Qin 

Fang · Hao Wu

Concrete
Structures
Under Projectile
Impact
Concrete Structures Under Projectile Impact
Qin Fang Hao Wu
•

Concrete Structures
Under Projectile Impact

123
Qin Fang Hao Wu
PLA University of Science and Technology PLA University of Science and Technology
Nanjing Nanjing
China China

ISBN 978-981-10-3619-4 ISBN 978-981-10-3620-0 (eBook)

DOI 10.1007/978-981-10-3620-0

Jointly published with Science Press, Beijing

ISBN: 978-7-03-051687-9 Science Press, Beijing
Library of Congress Control Number: 2017931342

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017
This work is subject to copyright. All rights are reserved by the Publishers, whether the whole or part
of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations,
recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission
or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar
methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this
publication does not imply, even in the absence of a specific statement, that such names are exempt from
the relevant protective laws and regulations and therefore free for general use.
The publishers, the authors and the editors are safe to assume that the advice and information in this
book are believed to be true and accurate at the date of publication. Neither the publishers nor the
authors or the editors give a warranty, express or implied, with respect to the material contained herein or
for any errors or omissions that may have been made. The publishers remains neutral with regard to
jurisdictional claims in published maps and institutional affiliations.

Printed on acid-free paper

This Springer imprint is published by Springer Nature

The registered company is Springer Nature Singapore Pte Ltd.
The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore
Preface

Concrete materials and structures have been widely used in both civil and military
engineering. The responses of concrete materials and structures hit by kinetic
energy projectiles as well as large aircrafts are the main focuses for both protective
structural engineers and weapon designers. For the local responses of concrete
targets hit by projectiles, the following four regimes of penetration mechanics can
be classified roughly by initial striking velocity of the projectile (denoted by V0).
They are the non-deformable penetration regime (V0
0.9 km/s for the traditional
earth penetration weapon, etc.), the mass abrasive projectile penetration regime
(0.9 km/s < V0
1.5 km/s for the advanced earth penetration weapon, etc.), the
semi-hydrodynamic (eroding) projectile penetration regime (1.5 km/s < V0
3
km/s for the long-rod penetrator), and the hydrodynamic penetration regime (V0 > 3
km/s for the shaped charge jet, space debris, etc.).
Aircraft crash on targets can be classified as the soft impact. The collision
velocity of the aircraft is around 120 m/s for the taking off and landing scenarios
and may reach about 250 m/s for the deliberate collision in the hijacking.
In this book, the authors present their theoretical, experimental, and numerical
investigations on concrete materials and structures under projectile and aircraft
impacts in recent years. The main contents consist of three parts.
Part I: Projectile Impact. Introduction of the local response of concrete targets
under projectile impact as well as a review of the existing empirical, analytical, and
numerical studies (Chap. 2); an extended cavity expansion model as well as a unified
rigid projectile impact model for both thick and thin concrete slabs (Chap. 3); mass
abrasive projectile penetration models (Chap. 4); eroding projectile and shaped
charge jet penetrations (Chap. 5); analysis of terminal ballistic trajectory (Chap. 6);
numerical simulation of projectile impact with a modified HJC model and a modified
K&C model (Chap. 7).
Part II: Aircraft Impact. Determination of aircraft impact force (Chap. 8);
numerical simulations of dynamic response of the prestressed nuclear power plant
containment against commercial aircraft A320 collisions (Chap. 9); two series of
aircraft engine model collision experiments on ultra-high-performance steel

v
vi Preface

fiber-reinforced concrete (UHP-SFRC) slabs and mesoscopic numerical simulations

(Chap. 10).
Part III: Protective Materials and Structures. Impact resistance of two new
types of ultra-high-performance cement-based composites (UHPCCs), e.g., basalt-
and corundum-aggregated UHP-SFRC (denoted by UHP-BASFRC and
UHP-CASFRC, respectively) hit by ogive- or flat-nosed projectiles (Chap. 11);
impact resistance of four types of concrete structures (UHP-BASFRC/fabric com-
posite slab; monolithic and segmented RC slabs with a rear steel liner;
SFRHSC/steel liner/sandy soil-layered composite target; and the rock-rubble
overlay) (Chap. 12).
We acknowledge Prof. Chen X.W. from China Academy of Engineering Physics
for helpful and valuable discussions, as well as our team members Kong X.Z.,
Peng Y., Zhang T., and Zhang J.H. from State Key Laboratory for Disaster
Prevention & Mitigation of Explosion & Impact, PLA University of Science and
Technology for their contributions to the preparation of this book. Finally, we
gratefully acknowledge the financial support of National Nature Science
Foundation of China (NSFC).

Nanjing, China Qin Fang

Hao Wu
Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2

Part I Projectile Impact

2 Projectile Impact Phenomena and Existing Studies . . . . . . . . . . . . . 9
2.1 Local Response of Concrete Targets Under
Projectile Impact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
2.2 Empirical Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Modified Petry Formula
(Petry 1910; Samuely 1939) . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 Ballistic Research Laboratory (BRL) Formula
(Beth 1941; Chelapati et al. 1972; Gwaltney 1968;
Adeli and Amin 1985). . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3 Army Corps of Engineers (ACE) Formula
(ACE 1946; Chelapati et al. 1972) . . . . . . . . . . . . . . . . 12
2.2.4 Modified National Defense Research Committee
(NDRC) Formula (NDRC 1946; Kennedy 1966) . . . . . 13
2.2.5 Ammann and Whitney Formula (Kennedy 1976;
Ben-Dor 2013) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.6 Whiffen Formula (Whiffen 1943) . . . . . . . . . . . . . . . . . 14
2.2.7 Kar Formula (Kar 1978) . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.8 CEA-EDF Perforation Formula (Berriaud 1978) . . . . . . 14
2.2.9 UK Atomic Energy Authority (UKAEA) Formula
(Barr 1990) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
2.2.10 Bechtel Formula (BPC 1974; Rotz 1976; Sliter 1980;
Bangash 1993). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.2.11 Stone and Webster Formula
(Sliter 1980; Jankov et al. 1976) . . . . . . . . . . . . . . . . . . 16

vii
viii Contents

2.2.12 Degen Perforation Formula (Degen 1980) . . . . . . . . . . . 17

2.2.13 Chang Formula (Chang 1981) . . . . . . . . . . . . . . . . . . . . 17
2.2.14 Haldar–Hamieh Formula
(Haldar and Hamieh 1984) . . . . . . . . . . . . . . . . . . . . . . 17
2.2.15 Adeli–Amin Formula (Adeli and Amin 1985) . . . . . . . . 18
2.2.16 Hughes Formula (Hughes 1984) . . . . . . . . . . . . . . . . . . 18
2.2.17 Healey and Weissman Formula (Bangash 1989) . . . . . . 18
2.2.18 IRS Formula (Bangash 1993) . . . . . . . . . . . . . . . . . . . . 19
2.2.19 CRIEPI Formula (Kojima 1991) . . . . . . . . . . . . . . . . . . 19
2.2.20 TM 5-855-1 Formula (TM 5-855-1 1986) . . . . . . . . . . . 19
2.2.21 UMIST Formula (Reid and Wen 2001; BNFL 2003;
Li et al. 2005) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
2.2.22 Wen Formula. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2.2.23 Berezan Formula (Sagomonyan 1974) . . . . . . . . . . . . . . 22
2.3 Semi-analytical Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3.1 Penetration. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
2.3.2 Perforation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
2.4 Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3 Rigid Projectile Penetration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.2 Classical Cavity Expansion Models . . . . . . . . . . . . . . . . . . . . . . 35
3.3 Extended Cavity Expansion Model . . . . . . . . . . . . . . . . . . . . . . . 36
3.3.1 Solution of the Extended Cavity Expansion Model . . . . 36
3.3.2 One-Dimensional Resistance Function of Concrete
Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.3.3 Analysis and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 48
3.4 Unified Deep Penetration Model . . . . . . . . . . . . . . . . . . . . . . . . . 51
3.4.1 Mean Resistance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
3.4.2 DOP . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
3.4.3 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
3.5 Perforation Model of Thick Concrete Slabs . . . . . . . . . . . . . . . . 59
3.5.1 Three-Stage Perforation Model . . . . . . . . . . . . . . . . . . . 60
3.5.2 Height of Rear Crater . . . . . . . . . . . . . . . . . . . . . . . . . . 60
3.5.3 Perforation Limit and Ballistic Limit . . . . . . . . . . . . . . . 61
3.5.4 Residual Velocity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.5.5 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
3.6 Perforation Model of Thin Concrete Slabs . . . . . . . . . . . . . . . . . 64
3.6.1 Perforation Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
3.6.2 Perforation Model of Thin Concrete Slabs . . . . . . . . . . 69
3.6.3 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
3.7 Unified Perforation Model of Thick and Thin
Concrete Slabs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
Contents ix

3.8 Deceleration-Time History . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79

3.8.1 Projectile Impact Test . . . . . . . . . . . . . . . . . . . . . . . . . . 82
3.8.2 Accuracy Analysis of Measured Deceleration-Time
Histories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
3.8.3 Derivation of Rigid-Body Deceleration-Time
Histories . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
3.8.4 Predictions of Deceleration-Time Histories . . . . . . . . . . 97
3.9 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
4 Mass Abrasive Projectile Penetration . . . . . . . . . . . . . . . . . . . . . . . . 107
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
4.2 Mass Abrasion Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108
4.2.1 Mass Loss Coefficient . . . . . . . . . . . . . . . . . . . . . . . . . . 110
4.2.2 Nose-Blunting Coefficient . . . . . . . . . . . . . . . . . . . . . . . 114
4.2.3 Prediction of Mass Loss in All Striking Velocity
Range . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
4.2.4 Validation of Mass Loss Coefficient . . . . . . . . . . . . . . . 121
4.2.5 Validation of Nose-Blunting Coefficient . . . . . . . . . . . . 123
4.3 Structural Stability Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
4.3.1 Drag Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 127
4.3.2 Limit Striking Velocity . . . . . . . . . . . . . . . . . . . . . . . . . 131
4.3.3 Normal Penetration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
4.3.4 Oblique Penetration . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
4.3.5 Penetration with Attacking Angle and Asymmetrical
Abrasion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
4.3.6 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 145
4.4 Stability Analyses of Terminal Ballistic Trajectory . . . . . . . . . . . 148
4.4.1 Drag Forces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 150
4.4.2 Free-Surface Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
4.4.3 Mass Abrasion Model of Projectiles . . . . . . . . . . . . . . . 154
4.4.4 Prediction of Terminal Ballistic Trajectory . . . . . . . . . . 155
4.4.5 Verifications of PENTRA2D . . . . . . . . . . . . . . . . . . . . . 158
4.4.6 Parametric Influential Analyses . . . . . . . . . . . . . . . . . . . 161
4.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 162
5 Eroding Projectile and Shaped Charge Jet Penetrations . . . . . . . . . 165
5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
5.2 Eroding Projectile Penetration . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
5.2.1 A Review of the Existing Studies . . . . . . . . . . . . . . . . . 167
5.2.2 Comparison Between the Existing Models . . . . . . . . . . 168
5.2.3 Projectile Penetration Test . . . . . . . . . . . . . . . . . . . . . . . 172
5.2.4 Penetration Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 182
5.3 Shaped Charge (SC) Jet Penetration . . . . . . . . . . . . . . . . . . . . . . 188
5.3.1 A Review of the Existing Studies . . . . . . . . . . . . . . . . . 188
x Contents

5.3.2 Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 190

5.3.3 Comparison with Test Data . . . . . . . . . . . . . . . . . . . . . . 196
5.3.4 Parametric Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . 204
5.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 209
6 Efficient Decoupled Analytical/Numerical Approach of Terminal
Ballistic Trajectory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
6.2 Forcing Function . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 213
6.2.1 Free-Surface Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
6.2.2 Layering Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 220
6.2.3 Separation and Reattachment Effect . . . . . . . . . . . . . . . . 227
6.3 Efficient Decoupled Analytical/Numerical Approach . . . . . . . . . 230
6.3.1 Analytical/Numerical Approach for Deformed
Projectiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 231
6.3.2 Analytical Approach for Rigid Projectiles . . . . . . . . . . . 231
6.4 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 233
6.4.1 Projectile Penetrations . . . . . . . . . . . . . . . . . . . . . . . . . . 233
6.4.2 Projectile Perforations . . . . . . . . . . . . . . . . . . . . . . . . . . 241
6.4.3 Projectile Penetration into Multi-layered Targets . . . . . . 244
6.5 Numerical Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 248
6.5.1 Perforation of Spaced Concrete Slabs . . . . . . . . . . . . . . 249
6.5.2 Penetration into Underground Structures . . . . . . . . . . . . 252
6.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 254
7 Numerical Simulation of Projectile Impact
on Concrete Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 255
7.2 Material Models in Hydrocodes . . . . . . . . . . . . . . . . . . . . . . . . . 256
7.2.1 General Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 256
7.2.2 Frequently Used Concrete Models . . . . . . . . . . . . . . . . . 257
7.2.3 Calibration of HJC Model Parameters . . . . . . . . . . . . . . 262
7.2.4 Validation of HJC Model Parameters . . . . . . . . . . . . . . 266
7.3 Computation of the Effective Strain Rate . . . . . . . . . . . . . . . . . . 271
7.3.1 Different Definitions of the Effective Strain Rate . . . . . . 276
7.3.2 Influence of Material Parameters . . . . . . . . . . . . . . . . . . 279
7.3.3 Verification and Correction . . . . . . . . . . . . . . . . . . . . . . 281
7.4 Modified HJC Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 286
7.4.1 Yield Surface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 288
7.4.2 Tensile Damage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 290
7.4.3 Lode Angle Dependency . . . . . . . . . . . . . . . . . . . . . . . . 292
7.4.4 Strain Rate Effect . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 295
7.4.5 Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 297
7.4.6 Parametric Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . 300
Contents xi

7.5 Modified K&C Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 305

7.5.1 Strength Surface Parameters . . . . . . . . . . . . . . . . . . . . . 305
7.5.2 DIFt . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 307
7.5.3 Relationship Between ƞ and k . . . . . . . . . . . . . . . . . . . . 308
7.5.4 Tensile Damage Accumulation . . . . . . . . . . . . . . . . . . . 309
7.5.5 Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 314
7.5.6 Parametric Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . 317
7.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 320

Part II Aircraft Impact

8 Aircraft Impact Force . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 325
8.2 Numerical Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 326
8.2.1 Commercial A320 Aircraft . . . . . . . . . . . . . . . . . . . . . . 327
8.2.2 Ling Ao NPP Containment . . . . . . . . . . . . . . . . . . . . . . 335
8.3 Calculating Methods for Aircraft Impact Force . . . . . . . . . . . . . . 339
8.3.1 Theoretical Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . 339
8.3.2 Numerical Simulation . . . . . . . . . . . . . . . . . . . . . . . . . . 343
8.4 Influence of Containment Radius on Impact Force . . . . . . . . . . . 346
8.4.1 Theoretical Analysis of Impact Impulse. . . . . . . . . . . . . 346
8.4.2 Numerical Simulation of Aircraft Impact Process . . . . . 347
8.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 350
9 Numerical Simulation of A320 Aircraft Impact on NPP
Containments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 353
9.2 Comparison Between the Integral and the Decoupled Impact
Simulations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
9.2.1 Impact Phenomenon Predicted by Missile-Target
Interaction Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . 354
9.2.2 Damage of the Prestressed NPP Containment . . . . . . . . 354
9.2.3 Containment Displacement . . . . . . . . . . . . . . . . . . . . . . 358
9.3 Parametric Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 360
9.3.1 Influence of Impact Position . . . . . . . . . . . . . . . . . . . . . 360
9.3.2 Influence of Impact Velocity . . . . . . . . . . . . . . . . . . . . . 365
9.3.3 Influence of Impact Angle . . . . . . . . . . . . . . . . . . . . . . . 369
9.3.4 Influence of Containment Wall Thickness . . . . . . . . . . . 372
9.3.5 Influence of Rebar Ratio . . . . . . . . . . . . . . . . . . . . . . . . 376
9.3.6 Influence of Prestressing Forces in Tendons . . . . . . . . . 380
9.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 384
xii Contents

10 Aircraft Engine Impact on UHP-SFRC Slabs . . . . . . . . . . . . . . . . . . 385

10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 385
10.2 Test Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 387
10.2.1 Aircraft Engine Model . . . . . . . . . . . . . . . . . . . . . . . . . . 387
10.2.2 UHP-SFRC Slabs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 388
10.2.3 Setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 390
10.3 Test Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
10.3.1 Impact Process. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 392
10.3.2 Deformation of Engine Missiles . . . . . . . . . . . . . . . . . . 393
10.3.3 Damage of UHP-SFRC Slabs . . . . . . . . . . . . . . . . . . . . 394
10.3.4 Test Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 396
10.4 Numerical Simulation and Comparison with the Test Data . . . . . 398
10.4.1 Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . . . . 398
10.4.2 Material Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 400
10.4.3 Numerical Simulation of the Experiment . . . . . . . . . . . . 404
10.4.4 Comparison with Riedel et al.’s Test . . . . . . . . . . . . . . . 410
10.5 Parametric Analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 416
10.5.1 Influence of Slab Thickness . . . . . . . . . . . . . . . . . . . . . . 416
10.5.2 Influence of Impact Velocity . . . . . . . . . . . . . . . . . . . . . 416
10.6 Modified Empirical Formula for Residual Velocity
of Engine Missiles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 419
10.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 423

Part III Protective Materials and Structures

11 UHPCC Targets Under Projectile Impact . . . . . . . . . . . . . . . . . . . . . 427
11.1 A Review on Projectile Impact Resistance of HSC . . . . . . . . . . 429
11.2 Static and Dynamic Properties of UHP-SFRC . . . . . . . . . . . . . . 433
11.3 Preparation of UHPCC . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
11.3.1 Compositions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
11.3.2 Mixing Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 435
11.4 Triaxial Compressive Test of UHP-BASFRC . . . . . . . . . . . . . . . 436
11.4.1 A Review on Triaxial Compressive Behavior
of Concrete . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 436
11.4.2 Specimens and Test Setup . . . . . . . . . . . . . . . . . . . .... 438
11.4.3 Test Results and Analyses . . . . . . . . . . . . . . . . . . . .... 439
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile
Penetration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 449
11.5.1 Projectile Penetration Test . . . . . . . . . . . . . . . . . . . . . . . 449
11.5.2 Rigid Projectile Penetration Analyses . . . . . . . . . . . . . . 456
11.5.3 Mass Abrasive Projectile Penetration Analyses . . . . . . . 463
Contents xiii

11.6 UHP-BASFRC Targets Under Flat-Nosed Projectile

Penetration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .... 465
11.6.1 Projectile Penetration Test . . . . . . . . . . . . . . . . . . . .... 466
11.6.2 Analysis and Discussion on DOP . . . . . . . . . . . . . .... 471
11.7 UHP-CASFRC Targets Under Ogive-Nosed Projectile
Penetration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 477
11.7.1 Static Tests of UHP-SFRC . . . . . . . . . . . . . . . . . . . . . . 478
11.7.2 Projectile Penetration Test . . . . . . . . . . . . . . . . . . . . . . . 481
11.7.3 Discussion on DOP . . . . . . . . . . . . . . . . . . . . . . . . . . . . 486
11.7.4 Theoretical Predictions of Mass Loss and DOP . . . . . . . 492
11.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 494
12 Concrete Structures Under Projectile Impact . . . . . . . . . . . . . . . . . . 497
12.1 UHP-BASFRC/Fabric Composite Slabs Against Small-Caliber
Arms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
12.1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 497
12.1.2 Bullet Perforation Test. . . . . . . . . . . . . . . . . . . . . . . . . . 498
12.1.3 Analysis of Terminal Ballistic Parameters . . . . . . . . . . . 506
12.2 Monolithic and Segmented RC Slabs
with a Rear Steel Liner . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510
12.2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 510
12.2.2 Test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 511
12.2.3 Analysis and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 515
12.3 Hard Projectile Impact on Layered SFRHSC Composite
Targets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
12.3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 527
12.3.2 Projectile Impact Test . . . . . . . . . . . . . . . . . . . . . . . . . . 528
12.3.3 Analysis and Discussion . . . . . . . . . . . . . . . . . . . . . . . . 534
12.4 Numerical Modeling of Rock-Rubble Overlays
Subjected to Projectile Penetration . . . . . . . . . . . . . . . . . . . . . . . 540
12.4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 540
12.4.2 3D Finite Element Model . . . . . . . . . . . . . . . . . . . . . . . 542
12.4.3 Numerical Analysis Approach . . . . . . . . . . . . . . . . . . . . 547
12.4.4 Validation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 549
12.4.5 Parametric Analysis and Discussion . . . . . . . . . . . . . . . 551
12.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 557
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 559
Nomenclature

b Attacking angle °
w Caliber-radius-head (CRH) of projectile
rr Cavity expansion stress Pa
/ Cone slope angle of rear crater °
ryp Dynamic yield strength of projectile Pa
ryt Dynamic yield strength of target Pa
da Half size of coarse aggregate m
eh Hoop strain
j Mechanical equivalent of heat
k0 Nose-blunting coefficient of projectile s2/km2
a Oblique angle °
m Poisson’s ratio
er Radial strain
s0 Shear strength of concrete target Pa
a0 Terminal yaw angle °
u Wake separation angle of projectile °
rh ,ru Hoop component of cavity expansion stress Pa
A Cross-sectional area of projectile m2
Acf Frontal crater area m2
C Mass loss coefficient of projectile s2/km2
C0 Bulk sound speed m/s
cr Rebar spacing in reinforced concrete target m
D Diameter of cylindrical concrete target m
d Projectile’s shank diameter m
da,max Maximum size of coarse aggregate in concrete m
Dcf Equivalent diameter of frontal crater m
Dcr Equivalent diameter of rear crater m
df Diameter of steel fiber mm
dr Rebar diameter m
Drk Diameter of rock rubble m

xv
xvi Nomenclature

Ec Elastic modulus of concrete target Pa

Ef Effective fracture energy of concrete J/m2
Es Young’s modulus of steel Pa
fc Unconfined compressive strength of concrete Pa
ft Unconfined tensile strength of concrete Pa
H Thickness of target m
Hcf Frontal crater height m
Hcr Rear crater height m
HM Moh’s hardness of coarse aggregate
hpen Depth of penetration (DOP) m
hper Perforation limit thickness m
hscab Scabbing limit m
I Dimensionless impact factor
K Bulk modulus of material Pa
L Projectile length m
l0 Projectile nose length m
La Half of wingspan of aircraft m
Leff Effective length of projectile m
lf Length of steel fiber mm
M Mass of projectile kg
Mr Residual mass of recovered projectile after hit kg
N Dimensionless projectile nose geometry factor
P Hydrostatic pressure Pa
P Perimeter of the projectile’s cross section m
P[x(t)] Longitudinal crushing force of aircraft N
Q Unit mass melting heat for penetrator material
Rc External radius of NPP containment m
S or R Dimensionless concrete strength parameter
u Particle radial displacement m
v Particle radial velocity m/s
V0 Initial striking velocity of projectile m/s
VBL Ballistic limit m/s
Vc Cavity expansion velocity m/s
Vca Volume fraction of coarse aggregate %
Vcf Frontal crater volume m3
Vcr Rear crater volume m3
Vf Volume fraction of steel fiber %
Vh Shaped charge jet tip velocity km/s
Vr Residual velocity of projectile after perforating target m/s
Vrf Ejecting velocity of rear fragments after projectile perforation m/s
Vrk Volume fraction of rock rubble %
Vt Shaped charge jet tail velocity km/s
Wca Weight fractions of coarse aggregate kg/m3
Wf Weight fraction of steel fiber kg/m3
y(x) Nose-shaped function of projectile
Nomenclature xvii

l1[x(t)] Longitudinal mass distribution of aircraft kg/m

l2[x(t)] Transverse mass distribution of aircraft kg/m
lm Sliding friction coefficient
q0 Density of target kg/m3
qp Projectile density kg/m3
Chapter 1
Introduction

1.1 Background

For the wide applications of concrete material in construction of both ground and
underground protective structures, such as military fortifications and nuclear power
plant containments, which are designed to withstand the intentional or accidental
impact loadings caused by high-speed projectiles, fragments, and so on, the
responses of concrete targets hit by kinetic energy projectiles are the main focus for
both civil engineers and weapon designers.
For the local responses of concrete targets hit by projectiles, the following four
regimes of penetration mechanics could be classified roughly by initial striking
velocity of the projectile (denoted by V0): (i) the non-deformable penetration regime
[V0  0.9 km/s, such as the traditional earth penetration weapon (EPW)], where
the projectile can be treated as rigid body; (ii) the semirigid mass abrasive pene-
tration regime [0.9 km/s < V0  1.5 km/s, such as the advanced earth penetration
weapon (AEPW)], where the effect of mass abrasion of the projectile nose should
be considered in penetration evaluation; (iii) the semi-hydrodynamic (eroding)
penetration regime (1.5 km/s < V0  3 km/s, such as the armor-piercing projec-
tile), where the erosions of both nose and shank of the projectile become obviously
and the classical or improved Alekseevskii–Tate models may provide reasonable
predictions; and (iv) the hydrodynamic penetration regime (V0 > 3 km/s, such as
the metal jet and space debris, etc.), which can be characterized as a fluid–fluid
interaction and governed by the steady-state fluid dynamics. The Bernoulli’s
equation may be used to predict the projectile penetration, in which the strengths of
both projectiles and targets can be neglected. It should be noted that the above
critical velocities are not absolute, which depend on the properties of concrete
targets and projectiles, such as the strength and density. In the existing related
published books regarding the above issues, Ben-Dor et al. (2006, 2013) compre-
hensively presented the empirical and analytical models for calculating high-speed

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 1
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_1
2 1 Introduction

penetration of rigid projectiles into various targets. Rosenberg and Dekel (2012,
2016) mainly discussed the eroding long-rod projectile penetrations into metallic
targets.
Besides, the deliberate collision of large commercial aircrafts with nuclear power
plant (NPP) and other key national infrastructures or buildings has attracted high
attention around the world since the beginning of twenty-first century, especially
after the “9.11” terrorists’ attacks. Before “9.11” event, the nuclear safety-related
concrete structures were not designed to resist any external impact loads greater
than a light military fighter. The large commercial aircraft impact belongs to a
beyond-design-basis event, which only requires risk evaluation using probabilistic
assessment methods as given in DOE Standard (DOE 2006). US Nuclear
Regulatory Commission (NRC 2009) requires that new NPPs to be built must take
impact load by large commercial aircrafts into consideration.

1.2 Scope

In this book, the authors present systematically their theoretical, experimental, and
the numerical research on the concrete materials and structures under projectile and
aircraft impacts in recent years. The main contents are introduced briefly as follows:
Part I: Projectile impact
Chapter 2. The local responses of concrete targets under projectile impact are
introduced firstly. The existing empirical and semiempirical penetration/perforation
models, as well as the numerical simulations on the projectile penetration/
perforation into concrete targets, are then reviewed in detail.
Chapter 3. The rigid projectile penetration and perforation into concrete targets
are discussed. Firstly, an extended dynamic spherical cavity expansion model is
proposed by using a hyperbolic yield criterion and Murnaghan equation of state to
describe the plastic behavior of concrete material under projectile penetration. The
extended model gives better predictions of DOP than the previous models based on
the linear yield criterion and equation of state. Secondly, based on the mean
resistance approach, a simple equation to predict the DOP of diversified targets
struck normally by hard projectiles with different nose shapes is proposed and
validated experimentally. Thirdly, by conducting the rigid projectile perforation test
on thin UHP-SFRC slabs, a unified rigid projectile perforation model for both thick
(H/d > 5) and thin (H/d  5) concrete slabs are established and verified by the
related tests. Finally, the accuracy analysis approach for the recorded ultra-high
g deceleration-time data is proposed. The capability of the theoretical models and
numerical simulations in predicting the projectile’s motion time histories is vali-
dated by comparing with the related test data.
Chapter 4. For the mass abrasive projectile penetration into concrete targets, the
mass loss and nose-blunting laws as well as the structural and terminal ballistic
1.2 Scope 3

trajectory stabilities are discussed. At first, the engineering model for the mass loss
and nose-blunting effects of projectiles is established and validated. Then, by
considering the asymmetrical nose abrasion as well as the initial oblique and
attacking angles, the limit striking velocity that the structural destruction of a
projectile takes place is formulated and discussed for different situations. Finally,
differential area force law (DAFL) method is employed to construct the finite
differential approach to predict the terminal ballistic trajectory of high-speed mass
abrasive projectiles. It simultaneously accounts for the free-surface effect of the
target attributed by oblique and attacking angles, as well as the asymmetric nose
abrasions. The validation of the proposed prediction program PENTRA2D is ver-
ified by comparing with the available test data.
Chapter 5. Eroding projectile and shaped charge (SC) jet penetrations into
concrete targets are discussed. Firstly, eighteen shots of flat-nosed cylindrical
projectiles high-speed (510–1780 m/s) penetrating into concrete targets are con-
ducted. Three penetration regimes, i.e., rigid projectile penetration, deforming
projectile penetration without eroding, and eroding projectile penetration, are
observed. By combining the A-T model and the unified one-dimensional target
resistance, a simple eroding projectile penetration model is established and vali-
dated. Furthermore, the transition velocities of the deforming projectile penetration
stage are discussed. Secondly, the numerical simulations of SC jet penetrating into
concrete targets are conducted. The whole processes of the SC jet including the
formation in the SC, elongation in the air, and penetration into concrete targets are
numerically reproduced. By comparing with the available test data, the constitutive
model and the corresponding parameters and finite element algorithm is validated.
Furthermore, the parametric influences on the ballistic performance of the SC jet are
discussed.
Chapter 6. The terminal ballistic trajectory of penetrators after penetrating into
concrete targets is another issue concerned both by weapon designers and by civil
engineers. Decoupled method is adopted to study the terminal ballistic trajectory, in
which the projectile is modeled with an explicit transient dynamic finite element
code, and the target is represented by the forcing function as the pressure boundary
conditions. The forcing function is constructed by comprehensively considering the
free-surface and layering effects of concrete targets as well as the separation and
reattachment effect at the projectile/target interface. The proposed methodology is
implemented in ABAQUS explicit solver via the user subroutine VDLOAD, and
the predicted projectile deformations and trajectories are compared with the existing
test data. Besides, terminal ballistic trajectory software for rigid projectiles is
compiled and verified, and two numerical examples of a three-layer spaced concrete
slab and an underground structure hit by a small diameter bomb (SDB) are
presented.
Chapter 7. Numerical simulations of projectile penetration into the concrete
targets are discussed. Firstly, the frequently used material models (HJC and K&C
models) are briefly introduced. Secondly, parameter calibration of the yield surface
and equation of state for normal strength concrete (NSC) and high-strength concrete
(HSC) of HJC model are conducted, which give better predictions of terminal
4 1 Introduction

ballistic parameters of projectiles. Thirdly, based on the relationship between the

effective strain rate and axial strain rate, the computation of the effective strain rate
in material models is corrected to ensure the effective strain rate equal to the axial
strain rate in uniaxial dynamic tests. Moreover, an iterative correction methodology
is proposed to eliminate the inertia effect in SHPB experiments. Finally, in order to
obtain the improved numerical simulations of the cratering and scabbing phe-
nomena in concrete slabs subjected to projectile impact, the yield surface, tensile
damage, and strain rate effect in the original HJC and K&C material models are
modified. The numerically predicted cratering size, scabbing size, and residual
velocities show reasonable agreements with the test observations.
Part II: Aircraft impact
Chapter 8. The fine finite element models of the commercial aircraft A320 and the
prestressed NPP containment are firstly established. By comparison with the
available test data with the numerical simulations, the corresponding constitutive
models and the related parameters are confirmed. Then, the whole collision process
of the aircraft A320 on the NPP containment is numerically simulated, and the
impact force is derived. The detailed information, e.g., impact forces of the different
aircraft parts (fuselage, wings, and engines), the high-frequency portion of the
force-time history, and the peak impact force caused by the collision of the external
broken wings, is reproduced. Finally, the weakening effect of the circular sectional
NPP containment on the impact force is analyzed, and the Riera function is further
modified by a reduction coefficient which is dependent on the dimensionless ratio of
the NPP containment diameter and aircraft wingspan.
Chapter 9. The detailed numerical simulations of the commercial aircraft A320
collision with the NPP containment are performed by using the missile-target
interaction method (integral method) and the loading-time history method (de-
coupled method), respectively. It derives that the prediction results of the
loading-time history method are significantly inclined to smaller damage, and the
missile-target interaction method is relatively more reliable and reasonable. Then,
the parametric influential analyses are performed. The effects of the impact position,
velocity and angle of the aircraft, the wall thickness, rebar ratio, and tendon force of
the containment on the impact force and impulse are discussed. The weakest
position of the barrel body of the NPP containment to resist the external aircraft
impact is confirmed.
Chapter 10. The effect of aircraft engine missile collision on UHP-SFRC slabs is
investigated experimentally and numerically. Firstly, the 1/10 scaled UHP-SFRC
slabs and engine missiles are adopted and two series of experiments with impact
velocities of about 120 and 250 m/s are conducted, respectively. Then, a meso-
scopic model of UHP-SFRC with considering the random distribution, volumetric
ratio, as well as the bonding and slipping effect of steel fibers is established. By
using the finite element program LS-DYNA, the impact processes, slab damages,
deformations, and residual velocities of the engine in the available tests are
numerically reproduced with the reasonable accuracy. Furthermore, based on the
1.2 Scope 5

parametric influential analyses of the slab thickness and impact velocity, a modified
empirical formula for predicting the residual velocity of engine missiles is presented
to guide the design of NPP containment.
Part III: Protective materials and structures
Chapter 11. The existing studies on projectile impact resistance of high-strength
concrete (HSC) as well as the static and dynamic properties of ultra-high-
performance steel fiber-reinforced concrete (UHP-SFRC) are reviewed firstly. Two
new types of ultra-high-performance cement-based composite (UHPCC), e.g.,
basalt and corundum aggregated UHP-SFRC (UHP-BASFRC and UHP-CASFRC),
are prepared under ordinary procedures without high-temperature and high-pressure
curing. The triaxial compressive strength, peak axial strain, damage modes, as well
as the failure criteria of UHP-BASFRC are studied experimentally with the con-
fining pressure as high as 100 MPa. Then, the ogive- and flat-nosed projectile
high-speed (299–1320 m/s) penetration tests on UHP-BASFRC or UHP-CASFRC
targets are carried out, the impact resistance of both UHPCC targets is evaluated
experimentally, and the analytical models are established and verified.
Chapter 12. The impact resistance of four types of concrete structures against
high-speed projectiles is studied. Firstly, the 7.62 mm API bullet impacting test on
bare and rear fabric (CFRP or UHMWPE) strengthened UHP-BASFRC slabs is
conducted. The perforation limit of the bare UHP-BASFRC slab is experimentally
determined, and the analytical model is proposed. In addition, the strengthening
effects of the two fabrics are compared. Secondly, twenty-five shots of reduce-
scaled projectile perforation test on five arrangements of monolithic and segmented
RC slabs with a rear steel liner is conducted. The impact resistances of the layered
bare RC and RC/steel composite targets are discussed. An explicit dimensionless
expression for predicting the terminal ballistic parameters of the projectile perfo-
rating the RC slab is proposed and validated. Thirdly, twenty shots of hard pro-
jectile high-speed impact test on the layered composite concrete target “SFRHSC
panel/steel liner/sandy soil” are carried out. A predicting approach for the ballistic
limits of RC/steel and RC/steel/sandy soil composite targets is proposed. Finally,
the generation algorithm of 3D rock-rubble particles with random sizes and shapes,
the taking and dropping and compacting algorithms to consider the random dis-
tribution of 3D rock-rubble particles in rock-rubble overlays, as well as the mapping
algorithm to generate the finite element grid are developed. The numerical results of
rock-rubble overlays subjected to the SDB projectile penetration agree well with the
existing test data.
 Part I
Projectile Impact
Chapter 2
Projectile Impact Phenomena
and Existing Studies

2.1 Local Response of Concrete Targets

Under Projectile Impact

For the local response of the concrete target hit by a non-deformable projectile,
three successive phenomenological responses are usually observed in test with the
gradually increasing of the initial striking velocity V0 of the projectile, as illustrated
in Fig. 2.1.
(i) Penetration, shown in Fig. 2.1a, a projectile penetrates into the concrete
target with a certain distance, the so-called spalling takes place in the vicinity
of the impacted location due to the extrusion of the projectile, while the distal
face (opposite to the impacted face) of the target has not any minor cracks.
(ii) Scabbing, shown in Fig. 2.1b, the projectile penetrates into the concrete
target with a certain distance, and the scabbing of fragments occurs on the
rear face of the target.
(iii) Perforation, shown in Fig. 2.1c, the projectile passes through the target with
certain residual velocity, and the fragments induced by shear plugging form
simultaneously ahead of the projectile.
For the thick concrete target, the whole perforation process consists of the
following three stages: impact cratering, tunneling, and rear shear plugging, such as
shown in Fig. 2.2a. For the thin concrete target, as shown in Fig. 2.2b, the cratering
process is followed directly by the rear shear plugging.
The mainly concerned terminal ballistic parameters during the projectile pene-
tration and perforation processes are the depth of penetration hpen (DOP, the normal
depth of the projectile penetrates into the target), scabbing limit thickness hscab (the
minimum thickness of the target required to prevent the scabbing at the rear face for
a given projectile striking velocity), perforation limit thickness hper (the minimum
thickness of the target required to prevent the perforation for a given projectile
striking velocity), the ballistic limit velocity VBL (the minimum initial impact

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 9
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_2
10 2 Projectile Impact Phenomena and Existing Studies

(a) (b) (c)

Spalling Scabbing

Fig. 2.1 Projectile impact on the concrete target a penetration b scabbing c perforation

(a)

Frontal crater

Tunnel

Rear crater

(b)

Frontal crater

Rear crater

Fig. 2.2 Frontal and rear faces and axial section plane view of a thick concrete slab b thin
concrete slab (Wu et al. 2015a; Beppu et al. 2008)

velocity of the projectile to perforate the target with the constant target thickness),
and the residual velocity of the projectile Vr (the remaining exit velocity of the
projectile after perforating a target).
In fact, the overall structural response of concrete targets has definite influence
on the perforation process of projectiles, which can be referred to Li et al. (2007).
However, this book mainly concerns with the local impact response of the concrete
target under projectile impact.
Generally, based on the field prototype or laboratory reduce-scaled projectile
shooting tests, the empirical (phenomenological), (semi-)analytical penetration/
perforation models as well as numerical simulations are the three main approaches
in the projectile impact investigations.
2.2 Empirical Models 11

2.2 Empirical Models

The empirical models are usually obtained by the curve fitting of projectile shooting
test data, thus the applications of the empirical models are dependent on the
parametric ranges covered in tests. For concrete targets, Kennedy (1976), Li et al.
(2005), and Ben-Dor et al. (2013) conducted detailed reviews and assessments on
the existing empirical penetration and perforation models. For instance, Ben-Dor
et al. (2013) assessed the validations of the empirical models by comparing the
prediction results with the projectile penetration data of 94 shots and projectile
perforation data of 102 shots.
In this section, several widely used empirical models which were proposed until
the end of the year 2014 are listed in the SI form. Limited by the length of this book
and to avoid the reduplication of the existing studies, the originations and modi-
fications of the models are simplified, which can be found in Li et al. (2005),
Ben-Dor et al. (2013), and Ranjan et al. (2014).

2.2.1 Modified Petry Formula (Petry 1910; Samuely 1939)


hpen M V02
¼ k 3 log10 1 þ ð2:1Þ
d d 19;974

Originally, in Petry I formula, k is taken as 6.36  10−4 for massive plain

concrete, 3.39  10−4 for normal reinforced concrete, and 2.26  10−4 for spe-
cially reinforced concrete. In the modified Petry II formula, k ¼ 0:0795kp and
Walter and Wolde-Tinsae (1984) proposed that
 
kp ¼ 6:34  103 exp 0:2973  107 fc ð2:2Þ

Amirikian (1950) further proposed that

hper hpen
¼2 ð2:3Þ
d d
hscab hpen
¼ 2:2 ð2:4Þ
d d

Amde et al. (1997) proposed that

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffi
Vr ¼ V0 1  0:5H=hpen ð2:5Þ
12 2 Projectile Impact Phenomena and Existing Studies

where H and fc are the thickness and unconfined compressive strength of concrete
targets, respectively. M and d represent the mass and diameter of the projectile,
respectively.

2.2.2 Ballistic Research Laboratory (BRL) Formula

(Beth 1941; Chelapati et al. 1972; Gwaltney 1968;
Adeli and Amin 1985)

The modified BRL formula was proposed by Gwaltney (1968) and Adeli and Amin
(1985) based on the original form proposed by Beth (1941) and Chelapati et al.
(1972) as follows


hpen 1:33  103 M 0:2 1:33
¼ p ffiffiffi
ffi d V0 ð2:6Þ
d fc d3

Chelapati et al. (1972) and Linderman et al. (1973) further proposed that

hper hpen
¼ 1:3 ð2:7Þ
d d
hscab hpen
¼2 ð2:8Þ
d d

2.2.3 Army Corps of Engineers (ACE) Formula

(ACE 1946; Chelapati et al. 1972)


hpen 3:5  104 M 0:215 1:5
¼ pffiffiffiffi d V0 þ 0:5 ð2:9Þ
d fc d3

hper hpen hpen hper

¼ 1:32 þ 1:24 for 1:35\ \13:5 or 3\ \18 ð2:10Þ
d d d d
hscab hpen hpen hscab
¼ 2:12 þ 1:36 for 0:65\  11:75 or 3\  18 ð2:11Þ
d d d d

The above Eqs. (2.10) and (2.11) are regressed from the impact tests of the
large-caliber (37–155 mm) penetrators, while for the 0.5 caliber bullets, they have
2.2 Empirical Models 13

hper hpen
¼ 1:23 þ 1:07 ð2:12Þ
d d
hscab hpen
¼ 2:28 þ 1:13 ð2:13Þ
d d

2.2.4 Modified National Defense Research Committee

(NDRC) Formula (NDRC 1946; Kennedy 1966)
(
hpen
¼ 2G0:5 for G  1
d
hpen ð2:14Þ
d ¼ G þ 1 for G [ 1

in which


5 N  M V0 1:8
G ¼ 3:8  10 pffiffiffiffi ð2:15Þ
d fc d

where N* is the projectile nose geometry factor, which equals to 0.72, 0.84, 1.0, and
1.14 for flat, hemispherical, blunt, and very sharp projectile noses, respectively.
8    2
< hper ¼ 3:19 hpen  0:718 hpen for
hpen
 1:35 or
hper
3
d d
  d d d
: hper ¼ 1:32 þ 1:24 hpen for 1:35\
hpen
\13:5 or 3\
hper
\18
d d d d

ð2:16Þ
8    2
< hscab ¼ 7:91 hpen  5:06 hpen for
hpen
 0:65 or hscab
3
d d
  d d d
: hscab ¼ 2:12 þ 1:36 hpen h
 11:75
d d for 0:65\ pen
d or d  18
3\ hscab
ð2:17Þ

2.2.5 Ammann and Whitney Formula

(Kennedy 1976; Ben-Dor 2013)

For the penetration of explosively generated high-speed fragments ðV0 [ 300 m/sÞ:


hpen 6  104  M 0:2 1:8
¼ pffiffiffiffi N d V0 ð2:18Þ
d fc d3
14 2 Projectile Impact Phenomena and Existing Studies

2.2.6 Whiffen Formula (Whiffen 1943)

Whiffen formula was suggested by British Road Research Laboratory as



0:1
97:51
hpen 2:61 M d V0 fc0:25
¼ pffiffiffiffi ð2:19Þ
d fc d3 da;max 533:4

where da;max is the maximum size of coarse aggregate in concrete. The application
ranges are V0 \1127:8 m=s, 5:52 MPa\fc \68:95 MPa; 0:136 kg\M\9979:2 kg
and 12:7 mm\d\965:2 mm.

2.2.7 Kar Formula (Kar 1978)

Aiming to the nuclear power plant protection and considering the projectile
materials and the coarse aggregate size of concrete targets, Kar proposed the fol-
lowing formula based on the NDRC formula
(
hpen
¼ 2G0:5 for G  1
d
hpen ð2:20Þ
d ¼ G þ 1 for G [ 1

in which

1:25 
1:8
E N M V0
G ¼ 3:8  105 pffiffiffiffi ð2:21Þ
Es d fc d
8    2
< hper da ¼ 3:19 hpen  0:718 hpen for
hpen
 1:35
d d d d
  ð2:22Þ
: hper da ¼ 1:32 þ 1:24 hpen for 1:35\
hpen
 13:5
d d d

8      2
< hscab d a Es 0:2 ¼ 7:19 hpen  5:06 hpen for
hpen
 0:65
d E d
  d d
ð2:23Þ
: hscab d a Es 0:2 ¼ 2:12 þ 1:36 hpen for 0:65\
hpen
\11:75
d E d d

where d a denotes a half of the coarse aggregate size, E and Es are Young’s moduli
of the projectile material and steel, respectively.

2.2.8 CEA-EDF Perforation Formula (Berriaud 1978)

Commisariat àl’Enerfie Atomoque-Electricité de France (CEA-EDF) proposed that

2.2 Empirical Models 15

hper M 0:5 V00:75

¼ 0:82 0:125 0:375 ð2:24Þ
d q0 fc d 1:5

2 2=3
1=6 dH
VBL ¼ 1:3q0 fc0:5 ð2:25Þ
M

Furthermore, by considering the non-circular missile cross section and rein-

forced concrete targets, Fullard et al. (1991) proposed

2 2=3
1=6 pH
VBL ¼ 1:3q0 fc0:5 ðr þ 0:3Þ0:5 ð2:26Þ
pM

where p is the perimeter of the projectile cross section and r represents the rein-
forcement ratio. The application ranges are V0 \200 m=s, 23 MPa\fc \46 MPa;
20 kg\M\300 kg; 0.35 < H/d < 4.17.

2.2.9 UK Atomic Energy Authority (UKAEA)

Formula (Barr 1990)
8h
>
< d ¼ 0:275  ð0:0756  GÞ
pen 0:5
for G  0:0726
hpen
d ¼ ð4G  0:242Þ
0:5
for 0:0726  G  1:0605 ð2:27Þ
>
: hpen
d ¼ G þ 0:9395 for G  1:0605

where


N  M V0 1:8
G ¼ 3:8  105 pffiffiffiffi ð2:28Þ
d fc d

The application ranges are 25 m/s\V0 \300 m=s, 22 MPa\fc \44 MPa,
5000 kg=m3 \M=d 3 \200;000 kg=m3 :

hscab
¼ 5:3G0:33 ð2:29Þ
d

The application ranges are 29 m=s\V0 \238 m=s; 26 MPa\fc \44 MPa,
3000 kg=m3 \M=d 3 \222;200 kg=m3 :
16 2 Projectile Impact Phenomena and Existing Studies

(
VBL ¼ Va h for Va  70 m/s
 V 2 i ð2:30Þ
VBL ¼ Va 1 þ 500
a
for Va [ 70 m/s

in which

2 2=3 h  c i
1=6 pH r
Va ¼ 1:3q0 kc1=2 ðr þ 0:3Þ1=2 1:2  0:6 ð2:31Þ
pM H

where cr is the rebar spacing distance, r is reinforcement ratio %

(each-way-each-face, EWEF), and kc ¼ fc for fc \37 MPa, and kc ¼ 37 MPa for
fc  37 MPa. The application ranges are 11 m/s\VBL \300 m/s, 22 MPa\fc
\52 MPa, 0:33\H=ðp=pÞ\5, 0\r\0:75%, 0:12\cr =H\0:49, 150 kg/m3 \
M=ðp2 H Þ\104 kg/m3 . When cr =H [ 0:49, Eq. (2.26) should be used instead.

2.2.10 Bechtel Formula (BPC 1974; Rotz 1976;

Sliter 1980; Bangash 1993)

Bechtel Power Corporation (BPC 1974; Rotz 1976; Sliter 1980; Bangash 1993)
proposed the scabbing limit for solid and hollow steel projectiles, respectively, as

0:4 0:5 
hscab M V0
¼ 38:98 0:5 1:2 ðsolid projectileÞ ð2:32Þ
d fc d

0:4 0:65 
hscab M V0
¼ 13:63 ðhollow projectileÞ ð2:33Þ
d fc0:5 d 1:2

2.2.11 Stone and Webster Formula (Sliter 1980;

Jankov et al. 1976)

1=3
hscab MV02
¼ ð2:34Þ
d ð0:013H=d þ 0:33Þd 3

The application ranges are 1:9 kg\M\12:8 kg, 27 m/s\V0 \157 m/s,
20:7 MPa\fc \31 MPa, 11:4 cm\H\15:2 cm and 1:5\hscab =d\3.
2.2 Empirical Models 17

2.2.12 Degen Perforation Formula (Degen 1980)

8    2
< hper ¼ 2:2 hpen  0:3 hpen for
hpen
\1:52 or
hper
\2:65
d d
 d d d
: hper ¼ 0:69 þ 1:29 hpen for 1:52 
hpen
 13:42 or 2:65 
hper
 18
d d d d

ð2:35Þ

where the penetration depth hpen is determined by the modified NDRC formula
(Eq. 2.14). The application ranges are 28:4 MPa\fc \43:1 MPa, 25 m/s\V0
\311:8 m/s, 0:15 m\H\0:61 m and 0:1 m\d\0:31 m.

2.2.13 Chang Formula (Chang 1981)

For a flat ended projectile, Chang (1981) proposed the dimensionless formulae as

0:25
0:5
hper 61 MV02
¼ ð2:36Þ
d V0 fc d 3

0:13
0:4
hscab 61 MV02
¼ 1:84 ð2:37Þ
d V0 fc d 3

where the application ranges are 16 m/s  V0  311:8 m/s, 0:11 kg  M  342:9 kg,
50:8 mm  d  304:8 mm and 22:8 MPa  fc  45:5 MPa.

2.2.14 Haldar–Hamieh Formula

(Haldar and Hamieh 1984)

Introducing the impact factor Ia ¼ MN  V02 fc d 3 , Haldar and Hamieh (1984) sug-
gested that
8h
>
< d ¼ 0:0308 þ 0:2251Ia
pen
for 0:3  Ia  4:0
hpen
d ¼ 0:6740 þ 0:0567Ia for 4:0\Ia  21 ð2:38Þ
>
: hpen
d ¼ 1:1875 þ 0:0299Ia for 21\Ia  455
18 2 Projectile Impact Phenomena and Existing Studies

2.2.15 Adeli–Amin Formula (Adeli and Amin 1985)

(
hpen
¼ 0:0416 þ 0:1698Ia  0:0045Ia2 for 0:3\Ia \4:0
d
hpen ð2:39Þ
d ¼ 0:0123 þ 0:196Ia  0:008Ia2 þ 0:0001Ia3 for 4:0  Ia \21

hper
¼ 1:8685 þ 0:4035Ia  0:0114Ia2 for 0:3\Ia \21 ð2:40Þ
d
hscab
¼ 0:9060 þ 0:3214Ia  0:0106Ia2 for 0:3\Ia \21 ð2:41Þ
d

where the application ranges are 27 m/s\V0 \312 m/s, 0:7  H=d  18,0:11 kg 
M  342:9 kg, d\0:3 m and hpen =d  2.

2.2.16 Hughes Formula (Hughes 1984)

Similar with
Haldar–Hamieh formula, by introducing the impact factor
Ih ¼ MV02 ft d 3 , Hughes (1984) suggested that

hpen Nh Ih
¼ 0:19 ð2:42Þ
d Sh
(
hper h hpen
¼ 3:6 pen for \0:7
d
hper
d
h
d
hpen ð2:43Þ
d ¼ 1:58 pen
d þ 1:4 for d  0:7
(
h hpen
hscab
¼ 5:0 pen for \0:7
d d
h
d
hpen ð2:44Þ
hscab
d ¼ 1:74 pen
d þ 2:3 for d  0:7

where Nh describes the projectile nose shape, and it equals to 1.0, 1.12, 1.26, and
1.39 for flat, blunt, spherical, and very sharp noses, respectively. The parameter
Sh ¼ 1:0 þ 12:3 lnð1:0 þ 0:03Ih Þ reflects the strain rate effect on concrete targets
during high-speed penetrations. The application range is Ih \3500.

2.2.17 Healey and Weissman Formula (Bangash 1989)

This formula is very similar to the modified NDRC and the Kar formulae and has
the form as
2.2 Empirical Models 19

(
hpen
¼ 2G0:5 for G1
d
hpen ð2:45Þ
d ¼ Gþ1 for G[1

where

 
1:8
E N M V0
G ¼ 4:36  105 pffiffiffiffi ð2:46Þ
Es d fc d

2.2.18 IRS Formula (Bangash 1993)

 
hpen ¼ 3703:376fc0:5 þ 82:152fc0:18 exp 0:104fc0:18 ð2:47Þ

The minimum shield thickness for total protection of a target against penetration,
perforation, and scabbing is
 
Hmin ¼ 3913:119fc0:5 þ 132:409fc0:18 exp 0:104fc0:18 ð2:48Þ

2.2.19 CRIEPI Formula (Kojima 1991)

The Central Research Institute of Electric Power Industry (CRIEPI) proposed

 
2=3

hpen 0:0265N  Md 0:2 V02 114  6:83  104 fc ðd þ 1:25H ÞHr r
¼ ð2:49Þ
d fc
2=3 ðd þ 1:25H ÞH

0:25
0:5
hper 61 MV02
¼ 0:9 ð2:50Þ
d V0 d 3 fc

0:13
0:4
hscab 61 MV02
¼ 1:75 ð2:51Þ
d V0 d 3 fc

where Hr = 0.2 m is the reference thickness of the concrete slab.

2.2.20 TM 5-855-1 Formula (TM 5-855-1 1986)

For the standard explosive formed fragment, the depth of penetration is expressed
as
20 2 Projectile Impact Phenomena and Existing Studies

8
< 0:027M0:25 V0
0:37 0:9

fc
for V0  V0
hpen ¼ ð2:52Þ
: 0:004M0:5 V0
0:4 1:8

fc
þ 0:395M 0:33 for V0 [ V0

and

hper ¼ 0:0311hpen M 0:033 þ 2:95M 0:33 ð2:53Þ

hscab ¼ 0:0334hpen M 0:033 þ 4:465M 0:33 ð2:54Þ

8 h
<  2 i0:555
V0 1  H hper for V0  V0
Vr ¼ ð2:55Þ
:    0:555
V0 1  H hper for V0 [ V0

where

V0 ¼ 5:13fc0:278 =M 0:044 ð2:56Þ

2.2.21 UMIST Formula (Reid and Wen 2001;

BNFL 2003; Li et al. 2005)

 
hpen 2 N MV02
¼ ð2:57Þ
d p 0:72rt d 3

rt ðMPaÞ ¼ 4:2fc ðMPaÞ þ 135 þ ½0:014fc ðMPaÞ þ 0:45V0 ðm/sÞ ð2:58Þ

where the application ranges are 50 mm\d  600 mm, 35 kg  M  2500 kg,
hpen =d  2:5 and 3 m/s\V0 \66:2 m/s.
Denoting the critical impact energies of a projectile causing through thickness
cone cracking, scabbing, and perforation as Ecrack, Escab, and Eper, the following
expressions are given
(i) H=d\5
( H  H 2
Ecrack
grt d 3 ¼ 0:00031 d þ 0:00113 d for 0\ Hd  2
   3 ð2:59Þ
Ecrack
grt d 3 ¼ 0:00325 Hd þ 0:00130 Hd for 2\ Hd \5


2
Escab N  H H
¼ 0:005441 þ 0:01386 ð2:60Þ
grt d 3 0:72 d d
2.2 Empirical Models 21

( Eper    2
grt d 3 ¼ 0:00506 Hd þ 0:01506 Hd for 0\ Hd  1
Eper    3 ð2:61Þ
grt d 3 ¼ 0:01 Hd þ 0:02 Hd for 1 d \5
H

where
8   qffiffiffi
< 0:5 þ 3 d rt for d
\ d
8 cr cr
g¼ qffiffiffi qdffiffiffir ð2:62Þ
: 0:5 þ 3 d rt for d
 d
8 dr cr dr

in which the total bending reinforcement ratio rt = 4r and r ¼ pdr2 =4Hcr is the
reinforcement ratio in EWEF, cr , and dr represent the rebar spacing and rebar
diameter, respectively.
(ii) H=d  5


Ecrack p H
¼  4:7 ð2:63Þ
rt d 3 4 d


Escab N  p H
¼  4:3 ð2:64Þ
rt d 3 0:72 4 d


Eper p H
¼  3:0 ð2:65Þ
rt d 3 4 d

2.2.22 Wen Formula

Based on the UMIST formula, Wen and Yang (2014) and Wen and Xian (2015)
proposed a unified approach for evaluating the terminal ballistic performance of
projectiles as


hpen 2 MV02
¼ ð2:66Þ
d p rd 3

rffiffiffiffiffi 
q0
r¼ a 0 þ b0 V0 Y ð2:67Þ
Y
8
< 1:4fc þ 45 for fc  75 MPa
Y¼ 150 for 75 MPa\fc \150 MPa ð2:68Þ
:
fc for fc  150 MPa
22 2 Projectile Impact Phenomena and Existing Studies

(i) For the thin target

H  H 4
Ecrack
/0 grd 3
¼ 0:0003 d þ 0:0003 d for 0\ Hd \5 ð2:69Þ
H  H 2
Escab
/0 grd 3
¼ 0:0066 d þ 0:012 d for 0\ Hd \5 ð2:70Þ
( Eper    2
/0 grd 3
¼ 0:0048 Hd þ 0:0018 Hd for 0\ Hd  1:3
Eper    3 ð2:71Þ
/0 grd 3
¼ 0:03 Hd þ 0:022 Hd for 1:3\ Hd \4

(ii) For the thick target

H 
Ecrack
rd 3 ¼ p4 d  4:7 for H
d 5 ð2:72Þ
H 
Escab
rd 3 ¼ p4 d  4:3 for H
d 5 ð2:73Þ
 p H 
4  d  3:0 for  4; V0 \300 m/s
Eper H
¼ p H
d ð2:74Þ
rd 3 4 d  2:0  4; V0  300 m/s
H
for d

where a0 , b0 and /0 are projectile nose shape related parameters, and g is related
with the projectile-nosed geometry and the arrangement of reinforcing steel mesh.
The detailed expressions of the above parameters can be referred to Wen and Yang
(2014) and Wen and Xian (2015).

2.2.23 Berezan Formula (Sagomonyan 1974)

Berezan formula was proposed as

hpen M
¼ k 5 3 V0 cos a ð2:75Þ
d 10 d

where k is the resistance coefficient which was given by Ben-Dor et al. (2013) for
diversified targets, and a is angle between the normal to the target surface and the
impact direction
2.3 Semi-analytical Models 23

2.3 Semi-analytical Models

2.3.1 Penetration

As illustrated in Sect. 2.1, the impact crater and the succeeding tunneling are
formed for a rigid projectile penetrating into the semi-infinite concrete target, the
schematic of which is shown in Fig. 2.3.
The resistance on a projectile during penetration process is commonly predicted
by the dynamic spherical or cylindrical cavity expansion theory (introduced in
Sect. 3.2). By using the onboard accelerometers, Forrestal et al. (2003a) measured
the deceleration-time histories of a projectile during penetrating into a semi-infinite
concrete target, shown in Fig. 2.4.
Figure 2.4 shows that the resistance acted on the projectile increased nearly
linearly in the cratering stage and maintained almost flat or a little decaying in the
tunneling stage until penetration stops. Forrestal et al. (1993, 1996) demonstrated
that the resistance on the projectile is proportional to the penetration depth during
the cratering stage, and the crater depth Hcf is also proportional to the diameter of
the projectile shank, e.g., Hcf = kd.
Forrestal et al. (1993, 1996) and Frew et al. (1998, 2006) suggested that k = 2
based on the test data and further proposed the prediction formulae for the depth of
penetration (DOP) as



ðL þ 0:5k0 d Þ qp Nq0 V12
hpen ¼ ln 1 þ þ 2d
2N q0 Sfc

 ð2:76aÞ
2M Nq0 V12
¼ 2 ln 1 þ þ 2d
pd q0 N Sfc

 " #
0:5
4w 1 ð 4w  1 Þ
k 0 ¼ 4w 
2
þ ð4w  1Þ 4w ð2w  1Þ sin1
0:5 2
ð2:76bÞ
3 3 2w

Fig. 2.3 Axial sectional

schematic of projectile
penetrating into the concrete
target
Hc,f
hpen
24 2 Projectile Impact Phenomena and Existing Studies

Fig. 2.4 Deceleration-time (a) 7.500

histories of projectile
penetrating into the concrete V0 =200 m/s
6.000
target (Forrestal et al. 2003a)

Deceleration (g)
4.500

3.000

1.500

0.000

-1.500
0 1 2 3 4 5 6
Time (ms)

(b) 10.000
V0 =250 m/s
8.000
Deceleration (g)

6.000

4.000

2.000

0.000

-2.000
0 1 2 3 4 5 6
Time (ms)

  
8w  1 V02  2d
L þ 0:5k 0 d
Sfc
q  0:544
N¼ ; V12 ¼   p  ; S ¼ 82:6  fc  106
24w2 1þN 2d q0
L þ 0:5k0 d qp

ð2:76cÞ

where L is shank length of a projectile, w ¼ s=d and s are the caliber-radius-head

(CRH) and circle radius of the ogival projectile nose, respectively.
In addition, by further considering the boundary influence of the target, Forrestal
et al. (2003a) modified Eq. (2.76a, b, c) as
h i
Nq0 V12
hpen ¼ 2M
pd 2 q0 N ln 1 þ R þ 2d hpen [ 2d ð2:77aÞ

8w  1 MV02  0:5pd 3 R
N¼ ; V12 ¼ ð2:77bÞ
24w2 M þ 0:5pd 3 Nq0
2.3 Semi-analytical Models 25

where V1 is the velocity of the projectile at the transition from the cratering stage to
the tunneling stage, and R is a concrete strength parameter similar to S in
Eq. (2.76c), which can be regressed from the penetration test data according to the
reverse expression of Eq. (2.77a).

Nq0 V02
R¼ 
 ð2:77cÞ
0:5pd 3 Nq0 pd 2 ðhpen 2d ÞNq0
1þ M exp 2M 1

Equations (2.76a, b, c) and (2.77a, b, c) proposed by Forrestal et al. have been

widely used. Nevertheless, Forrestal formulae are only applicable for ogive-nosed
projectiles and the friction between the projectile and target during penetrations is
neglected. Furthermore, the target strength parameter R in Eqs. (2.77a, b, c) must be
confirmed by the projectile penetration tests, thus the prediction function of
Eqs. (2.77a, b, c) is lost.
In addition, according to the two-stage (cratering + tunneling) penetration
model, Chen and Li (2002), Li and Chen (2003), and Li et al. (2004) obtained the
crater depth coefficient as k = (0.707 + l0/d) based on the slip-line field theory, in
which l0 is the projectile nose length. Introducing the dimensionless
projectile-nosed geometry function N and impact factor I0, and neglecting the crater
depth under the deep penetration, Eqs. (2.76a, b, c) was extended by Chen and Li
(2002) and Li and Chen (2003) to the dimensionless form which can be applied for
arbitrary-nosed projectiles.


hpen 2 I0
¼ N ln 1 þ ð2:78aÞ
d p N

MV02 M  0:5
I0 ¼ ; N¼ ; S ¼ 72:0  fc  106 ð2:78bÞ
N1 Sfc d 3 N2 q0 d 3

where N1 and N2 describe the influence of projectile nose geometry as well as the
frictions between projectiles and targets. For an arbitrary-nosed projectile, N1 and
N2 can be obtained by integrating the periphery function of projectile nose along the
nose length. In Chen and Li (2002), the explicit expressions of N1 and N2 for ogive,
conical, blunt, truncated ogive, hemispherical, and flat projectile nose were given.
Besides, Qian et al. (2000) introduced a resistance coefficient and target strength
parameter to analyze the DOP of a truncated ogive-nosed projectile penetrating into
concrete targets. However, the proposed formula is only valid when the length of
the truncated part is less than 1/3 of the original nose length. Jan and Henrik (2004)
extended the Forrestal formulae to the truncated ogive- and flat-nosed projectiles by
modifying the resistance function based on the cavity expansion theory.
26 2 Projectile Impact Phenomena and Existing Studies

2.3.2 Perforation

As for the projectile perforation, the damage of a concrete slab is schematically

shown in Fig. 2.5. Figure 2.6 shows the raw and low-pass filtered deceleration-time
histories of a projectile perforating a 300-mm-thick concrete slab. Different from
the decelerations of the projectile penetrating into the semi-infinite concrete target
shown in Fig. 2.4, nearly no constant deceleration plateau is observed.
Based on the two-stage “tunneling + shear plugging” perforation model,
Yankelevsky (1997) obtained the expression of perforation limit for the low speed
(V0 < 200 m/s) impact of the flat-nosed projectile. Chen and Li (2003), Li and
Tong (2003), and Chen et al. (2004, 2008b) proposed the prediction formulae of
perforation limit and ballistic limit for the reinforced concrete slab based on three
stage “cratering + tunneling + shear plugging” perforation model. It was found
that, if the projectile did not hit the rebar mesh, the relatively low reinforcement
ratio (less than 4%) had nearly no effect on the ballistic limit of the projectile. Based
on projectile perforation tests with 106 shots (Sliter 1980; Hanchak et al. 1992), the
predicted accuracy of the above-mentioned perforation limit formulae for the
damage modes of the concrete target (perforation, critical perforation and
non-perforation) reached as high as 83% (Li and Tong 2003).

Front crater
Tunnel
Rebar
H

Rear shear plug

Fig. 2.5 Schematics of three-stage perforation model of the RC slab

Fig. 2.6 Deceleration-time 20

history of projectile
perforating the target (Wu 15
et al. 2015a)
Deceleration (104 g)

10

-5
V0=641.5 m/s
-10 Filtered test data (1700Hz low-pass)

18 19 20 21 22 23 24
Time (ms)
2.3 Semi-analytical Models 27

Perforation limit hper was given by Chen and Li (2003), Li and Tong (2003), and
Chen et al. (2004, 2008b) as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi
hper hpen 1 þ 3S tan /  1 hpen
¼ þ for [k ð2:79Þ
d d 2 tan / d

where hpen and S are referred in Eqs. (2.78a, b), / is the cone slope angle of the rear
crater shown in Fig. 2.5. Dancygier (1998) proposed that tan / = 2.254 and 4.108
for normal strength concrete (NSC) and high strength concrete (HSC), respectively.
Ballistic limit VBL and the residual velocity Vr of the projectile can be obtained
reversely from Eq. (2.79) as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Sfc d 3 IBL p H Hplug k 1 þ 3S tan /  1
VBL ¼ ; IBL ¼   ; Hplug ¼
M 2 d d 2 2 tan /
ð2:80aÞ
2 0:5
Vr ¼ ðV02  VBL Þ ð2:80bÞ

Li et al. (2006) further obtained the critical impact kinetic energy of a projectile
for perforating concrete slabs, which showed better agreements with the flat-nosed
projectile perforation test data by comparing with the energy forms of UMIST and
NDRC formulae as well as Eq. (2.79) for the perforation limit. Ben-Dor et al.
(2010) obtained the perforation limit for the flat-nosed projectile by extending the
two-stage model, which gave a good agreement with the experimental damage
mode of the concrete target (Sliter 1980).
As for the projectile perforating on the segmented concrete targets, TM5-855-1
(1986) gives that a thick slab composed of several layers generally has a smaller
perforation resistance than a monolithic slab with the same thickness. Ben-Dor et al.
(2009) found theoretically that (i) the ballistic limit of multilayered concrete shield
did not depend on the order of the slabs in the shield; (ii) the monolithic shield was
superior to any layered shield with the same thickness; (iii) the largest decrease in
the ballistic limit velocity occurred when a shield was divided into a number of
slabs with the same thickness.

2.4 Numerical Simulation

Since the projectile penetration or perforation tests are difficult to perform and
always expensive, numerical simulations of projectile penetration or perforation are
playing an more and more important role. Numerical simulation approach can not
only reproduce the transient phenomena of the projectile and target interaction, but
also obtain numerical solutions (e.g., stress and strain) for the whole field. Besides,
28 2 Projectile Impact Phenomena and Existing Studies

it can provide the support and evidence for the establishment of analytical models
and empirical formulae. As far as the terminal ballistics is concerned, Zukas (1990)
conducted a comprehensive and informative survey of computer codes for impact
simulations.
Numerical simulations of high velocity impact events usually use the so-called
hydrocode, which can treat the complicate high pressure, high strain rate, and large
deformation problems. Generally, the hydrocode is an efficient tool for solving the
conservation equations for mass, momentum, and energy with the initial and
boundary conditions. Based on different representations of the conservation laws
for a continuum, it can be classified into Eulerian, Lagrangian, and Arbitrary
Lagrangian–Eulerian approaches. For Eulerian approach, the mesh is fixed, while
material flows through the mesh. The flow material through the fixed mesh deter-
mines the mass, pressure, velocity, and so on. The Eulerian approach is more
suitable for drastic distorted materials. However, since only one material is allowed
to exist in each element, the interface between the projectile and target is difficult to
capture clearly. As for Lagrangian approach, the mesh is attached to the material
and moves with it. Consequently, it is more appropriate for the material element
that is less distorted and the interface between the projectile and target can be
clearly observed. In addition, Arbitrary Lagrangian–Eulerian approach is a com-
bination of Lagrangian and Eulerian approaches. For example, when a rigid pro-
jectile penetrates into a relatively soft target, the Lagrangian approach is used for
the description of the projectile and the Eulerian approach for the soft target.
Besides, based on different discretization methods, numerical simulations can be
classified into finite difference method, finite element method and discrete element
method, smoothed particle hydrodynamics method, and so on.
Lagrangian approach often suffers from mesh entanglement in the target that can
prematurely terminate numerical computation. Therefore, the element erosion
technique is introduced to address this issue, but the erosion criteria are usually
empirical. In order to solve the above problems, Ortiz (1996) and Espinosa et al.
(1998) developed an adaptive element method, in which the algorithms for cor-
recting finite element mesh distortion through mesh rezoning, optimization, and
refinement were presented. When the mesh distortion reaches a critical condition
(usually represented by the accumulated equivalent plastic strain), rezoning and
refinement of these elements are performed. Algorithm to automatically convert
distorted finite elements into meshless particles is alternative selection, which was
proposed by Johnson et al. (2002) and Johnson and Stryk (2003). In this method,
when finite elements reach a user-specified plastic strain, they are converted to
particles and linked to the adjacent finite element grid. This method allows for the
use of accurate and efficient finite element approach in the lower distortion regions,
and for the use of meshless particle approach in the higher distortion regions. The
smoothed particle hydrodynamics (SPH) approach is also usually used to avoid the
mesh entanglement. However, it is generally known that SPH approach is
time-consuming and instable on the boundary. Recently proposed reproducing
kernel particle method (RKPM) (Liu et al. 1995) can well solve the above problems
and is successfully used in the projectile penetration simulations (Wu et al. 2012d;
2.4 Numerical Simulation 29

Fig. 2.7 Projectile penetration simulation results a adaptive element method (Espinosa et al.
1998) b F-M method (Johnson et al. 2002) c RKPM method (Wu et al. 2014c)

Wu et al. 2014c). Figure 2.7 presents the projectile penetration simulation results of
adaptive element, converting distorted finite elements into meshless particles (F-M)
and RKPM methods, respectively.
A proper material model is needed for correct simulation of projectile impact.
The dynamic behavior of metal material under intense dynamic loading is relatively
simple and Johnson–Cook model (Johnson and Cook 1983) is widely adopted.
However, the dynamic behavior of brittle material (e.g., concrete and rock) is rather
complex and is still the research hotspot.
Tu and Lu (2009) presented a comprehensive evaluation of several widely used
concrete material models in commercial hydrocodes. Among them, Holmquist–
Johnson–Cook (HJC) concrete model is widely used for impact simulation
(Holmquist et al. 1993). In this model, two stress invariants are employed to for-
mulate the current failure surface. Concrete is considered linear elastic before a
prescribed maximum failure surface, beyond which the accumulated damage that
depends on the equivalent plastic strain and plastic volume strain shifts the current
failure surface to a residual state by reducing the cohesion strength value. In order
to suppress the fracture by low magnitude tensile waves, a simple cutoff for the
fracture strain (EFMIN) is adopted. HJC model does consider most of the important
aspects of concrete behavior under impact loads, such as the pressure dependency,
strain rate effect, damage cracking, and compaction (Polanco-Loria et al. 2008).
However, from a critical point of view, many aspects are not taken into account,
such as (i) the independence of the third stress invariant cannot capture the tran-
sition of the deviatoric section from the triangular shape at low pressures to the
circular shape at high pressures; (ii) strain hardening is not considered; (iii) strain
softening is only dependent on the fracture strain cutoff which is usually difficult to
be determined; (iv) strain rate enhancement for compression and tension are
identical, which is contrary to the available experiments [CEB-FIB Model Code
1990 (1993)]; and (v) tensile damage is not considered, which may play an
important role in the simulation of the cratering and scabbing phenomena (Xu and
Lu 2006; Li and Hao 2014).
30 2 Projectile Impact Phenomena and Existing Studies

Fig. 2.8 Tensile damage

distribution during projectile
perforating a concrete slab
(Liu et al. 2009)

The previous modifications to HJC model are mainly focused on the compres-
sive behavior (Polanco-Loria et al. 2008) and strain rate effect (Polanco-Loria et al.
2008; Islam et al. 2012), and they are still not suitable for the predictions of the
cratering and scabbing phenomena. Liu et al. (2009) presented a modified version
of HJC model, where they employed TCK model (Taylor et al. 1986) to describe
the dynamic tensile behaviors of concrete, while the compressive behaviors of
concrete were still governed by HJC model. Based on this model, Liu et al.
(2009) gave reasonable predictions of the cratering and scabbing phenomena in
concrete slabs, as shown in Fig. 2.8. However, the yield surface of the above
modified model is discontinuous due to the separate treatment of compressive and
tensile behaviors. Besides, TCK model has nine differential equations to be care-
fully integrated during a time step, which may cause numerical computation
instability. In addition, some parameters, such as the volumetric strain rate expe-
rienced by concrete at fracture, are difficult to determine experimentally. Thus, a
more reasonable concrete material model should be proposed for the projectile
penetrating into concrete targets, and the shortcomings of HJC model discussed
above should be well addressed, especially for the correct reproducing the cratering
and scabbing phenomena observed in the experiments.
Gebbeken and Ruppert (2000) also proposed an extension form of HJC model
for concrete, in which the third stress invariant was accounted in the definition of
failure surface, the concrete damage was considered in the enlargement of failure
surface with the strain rate effect, and the tensile failure was represented by a single
parameter (Rankine criterion).
Riedel–Thoma–Hiermaier (RHT) model (AUTODYN 2003; Riedel et al. 1999)
is also an extended version of HJC model, in which the third stress invariant is
considered in the yield strength surface, but not in the residual strength surface. The
description of damage is similar to HJC model, but independent on the plastic
volume strain. Strain rate effect is taken into account by a dynamic amplification
factor, which could become problematic in the simulation of high strain rate
problems (Tu and Lu 2009). Tu and Lu (2010) presented several enhancements of
RHT model, such as a modified residual strength surface dependent on the third
stress invariant, as well as the improved tensile-to-compressive meridian ratio,
dynamic tensile strength increasing factor, and bilinear softening model based on
the fracture energy.
2.5 Summary 31

2.5 Summary

In this chapter, focusing on the local response of concrete targets under the impact
of projectile, a state-of-the-art review of the relevant studies is presented. Firstly,
local response of semi-infinite and finite thickness concrete targets under projectile
penetration or perforation is roughly described by the two or three successive
stages. Then, the frequently used empirical formulae, semi-analytical models, and
numerical simulations for projectile impact are reviewed in detail. Total 23 sets
prevalent empirical formulae proposed until the end of the year 2014 are listed in
the SI form, which can be referred as a handbook. Also, the semi-analytical models
(e.g., Forrestal series formulae, Li and Chen series formulae) for predicting the
terminal ballistic parameters of projectiles impacting on concrete targets are sum-
marized. Finally, different numerical simulation approaches, such as the
Lagrangian, Eulerian, Arbitrary Lagrangian–Eulerian approaches, as well as the
frequently used constitutive models of concrete material in hydrocodes are intro-
duced and evaluated.
Chapter 3
Rigid Projectile Penetration

3.1 Introduction

Terminal ballistic parameters of hard projectiles penetrating/perforating into con-

crete targets are the main focuses for both civil engineers and weapon designers.
The related experimental, analytical, and numerical studies can be referred to the
reviews by Li et al. (2005), Corbett et al. (1996), and Anderson and Bodner (1988).
The empirical formulae in Sect. 2.2 received widely applications, and some of them
are the fundamentals of the design codes (e.g., TM5-855-1, TM-1300) of protective
structures. However, the empirical formulae are only applicable within the para-
metric ranges that the tests covered, and some of them even are unit-dependent.
Numerical simulations can provide visible information on damage, stress, and
deformation field, while the related parameters in material model of targets are
difficult to be determined. Experiment-based analytical (semi-analytical) approa-
ches could be the most efficient and economical way to study the local effects of the
projectile impacting on concrete targets.
For rigid projectile penetration as introduced in Sect. 2.3.1, based on the cavity
expansion theory and assuming that the normal stresses acting on the projectile’s
nose equal to the expansion stresses of the spherical or cylindrical cavity, Forrestal
and Luk (1992a), Forrestal and Tzou (1997), Forrestal et al. (1995), Forrestal and
Warren (2008), Luk et al. (1991), and Warren et al. (2004a) proposed a series of
formulae to predict DOPs of ogive-nosed projectiles penetrating into soil, concrete,
ductile metal, and rock targets. For the complexity of solving the cavity expansion
theory, by introducing a dimensionless empirical parameter S which is dependent
only on the concrete compressive strength, Forrestal et al. (1994, 1996) and Frew
et al. (1998) simplified the expression of the cavity expansion stress and applied it
to analyze the projectile penetrations into concrete targets. Furthermore, by con-
sidering the target boundary influences on the penetration process, Forrestal et al.
(2003a) and Frew et al. (2006) proposed another parameter R to describe the
penetration resistance. However, the equivalent target strength parameter R must be

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 33
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_3
34 3 Rigid Projectile Penetration

confirmed by the actual projectile penetration test data and the prediction function
was lost. By introducing the dimensionless projectile-nosed geometry function and
impact factor, as well as considering the projectile/target interfacial frictions, Chen
and Li (2002) as well as Li and Chen (2003) extended the Forrestal’s formula to the
dimensionless form to analyze projectiles with arbitrary-nosed profile penetration
into diversified targets. Since the target strength parameter S used in Forrestal et al.
(1994, 1996), Frew et al. (1998), Chen and Li (2002), and Li and Chen (2003) was
obtained by fitting the test data of ogive-nosed projectiles penetrating into normal
strength concrete targets, Wu et al. (2012a) further regressed the expression of S for
high-strength concrete (HSC) targets subjected to projectiles with various nose
shapes, which was validated by the penetration tests on HSC targets with the
maximum compressive strength of *250 MPa. The resistances acting on the
projectile during the penetration in the above studies are dependent on the
instantaneous velocity of the projectile. For the derivation of DOP, this makes it
more complicated in integrating the motion equation, especially for the complex
projectile nose profile and consideration of the friction between the projectile and
target interface.
For rigid projectile perforations, as introduced in Sect. 2.3.2, based on the
three-stage (cratering + tunneling + shear plugging) perforation model, Chen et al.
(2008b) and Li and Tong (2003) proposed the formulae for predicting the ballistic
limit and perforation limit, as well as the residual velocity of the projectile after
perforations. By further considering the kinetic energy carried by the ejected
fragments on the rear face of the concrete slab, Wu et al. (2012a) proposed an
expression to predict the residual velocity of the projectile after perforations.
Recently, by considering the energy dissipated through the fracture of ejecting
fragments into pieces, Grisaro and Dancygier (2014a) proposed a modified energy
balance model of the projectile perforating on the concrete slab. However, the
perforation limit in this model is obtained by curve fitting of the test data and only
applied for residual velocity prediction. Besides, the height of rear crater and the
ejecting velocity of rear shear fragment are the two key factors in the projectile
perforation model for the concrete slab. However, the rear crater height equation
proposed in Chen et al. (2008b) and Li and Tong (2003) is very complex and its
accuracy is not validated for the lack of test data. More recently, Wu et al. (2015a)
conducted a series of projectile perforation tests on concrete slabs with thicknesses
ranged from 100 to 300 mm, and the striking and residual velocities of the pro-
jectiles as well as the dimensions of the front impact crater and rear shear crater
were recorded in detail.
In this chapter, cavity expansion theory (CET) is first introduced in Sect. 3.2,
and then in Sect. 3.3, a hyperbolic yield criterion and Murnaghan equation of state
are introduced to describe the plastic behavior of concrete targets under projectile
penetrations, an extended dynamic spherical cavity expansion model is proposed
and applied in the analyses of projectile penetrations. In Sects. 3.4–3.7, based on the
mean resistance approach, the projectile penetration and perforation models for both
thick and thin targets are established and validated. Then, a unified projectile
perforation model for both thick and thin concrete slabs is presented and applied
3.1 Introduction 35

into the resistance evaluation of the segmented targets. Finally, in Sect. 3.8, the
rigid projectile deceleration-time histories measured in tests given in Sect. 12.2 are
analyzed in detail.

3.2 Classical Cavity Expansion Models

Cavity expansion theory (CET) was firstly studied by Bishop et al. (1945), in which
the quasi-static equations for the expansion of cylindrical and spherical cavities
were developed and used to estimate the force applied on the wedge-shaped pen-
etrator when it was punched slowly into a metal target. Hill (1948) and Hopkins
(1960) developed the dynamic cavity expansion equations for an incompressible
target material, which was later applied by Goodier (1965) with considering the
target inertia effects, to predict the penetration depth of a rigid sphere into metal
targets. More recently, the works conducted by Forrestal’s team (Forrestal and Luk
1988a; Forrestal et al. 1988b, 1994, 1995; Luk et al. 1991; Frew et al. 2000)
developed cavity expansion penetration models for rigid projectiles penetrating into
ductile metal, soil, concrete, and rock targets, respectively. They developed
closed-form expressions for DOP of rigid projectiles and demonstrated good
agreements with experimental results for normal impact.
Based on CET, under the projectile penetrations, a spherically symmetric cavity
is assumed expand radially from the projectile surface, and the cavity radius
increases from zero at a constant velocity Vc. Generally, cavity expansion produces
plastic-cracked-elastic response regions for concrete material as shown in Fig. 3.1,
where r, t, c, c1, and cd represent the radial Eulerian coordinate, time,
cracked-plastic boundary velocity, elastic-cracked boundary velocity, and dilata-
tional velocity, respectively.
For concrete material concerned in this book, aiming to simplify the solution
procedure, the linear yield criteria were adopted in the dynamic cavity expansion
models in the existing works, such as the original and modified forms of Tresca
(Luk and Forrestal 1987), Drucker-Prager (Feng et al. 2015), and Mohr–Coulomb
(Forrestal and Tzou 1997; He et al. 2011; Forrestal et al. 1981, 1986) criteria. It is

Fig. 3.1 Response regions

for concrete material 1/

1/

1 Undisturbed
Elastic
Cracked
Plastic
Cavity

Vc t ct c1t cd t r
36 3 Rigid Projectile Penetration

well known that the shear strength–pressure relationship of concrete material is

nonlinear; thus, a more actual nonlinear yield criterion should be utilized.
Besides, as for the equation of state (EOS) associated with the existing dynamic
cavity expansion models, i.e., the relationship between the pressure and volumetric
strain was usually described by the bulk modulus of linear elastic deformation
(Forrestal and Tzou 1997; He et al. 2011; Forrestal et al. 1981, 1986) or a
linear-locked hydrostat (material becomes incompressible beyond a critical value of
volumetric strain) (Luk and Forrestal 1987; Forrestal and Luk 1992a). For
high-speed projectile penetration into concrete targets, the pressure at the
projectile/target interface can reach a magnitude of 1–3 GPa, where the relationship
of pressure-volumetric strain exhibits obvious nonlinear characteristic. Therefore,
neither the linear EOS nor linear-locked-hydrostat EOS could well represent the
actual situation. Feng et al. (2015) and Shi et al. (2014) used a three-stage and
P-alpha EOS to consider the nonlinear pressure–volumetric strain relationships of
concrete and soil targets under projectile penetrations, respectively. However, the
yield criteria used in the above studies cannot describe the nonlinear shear strength–
pressure relationship.

3.3 Extended Cavity Expansion Model

Considering the deficiencies of the yield criterion and EOS used in the existing
dynamic cavity expansion models introduced in Sect. 3.2, an extended dynamic
cavity expansion model is proposed in this section. Firstly, a hyperbolic yield
criterion and the nonlinear Murnaghan EOS are introduced to describe the plastic
behavior of concrete material under projectile penetrations. Then, a
one-dimensional resistance function of concrete targets with considering projectile
nose shape influences is presented. Furthermore, by combining with the
one-dimensional momentum theorem, the rigid projectile penetration model is
established. The validation of the extended cavity expansion model is verified by
comparing with the existing penetration test data as well as the predicted results of
the previous model (Forrestal and Tzou 1997). Besides, the above one-dimensional
resistance function of concrete targets will be combined with the Alekseevskii–Tate
equations to analyze eroding projectile penetration, which will be addressed in
Sect. 5.2.

3.3.1 Solution of the Extended Cavity Expansion Model

A hyperbolic yield criterion in Eulerian coordinates with spherical symmetry can be

described by
3.3 Extended Cavity Expansion Model 37

P
rr
rh ¼ a0 þ ð3:1aÞ
a1 þ a 2 P

P ¼ ðrr þ rh þ ru Þ=3; rh ¼ ru ð3:1bÞ

where rr is the radial component of the stress, rh and ru are the hoop components
of the stresses, respectively. P is the hydrostatic pressure, which are measured
positive in compression. a0 , a1 , and a2 are the constants, which need to be deter-
mined from the triaxial compression test data. Specially, Eqs. (3.1a, b) reduces to
Mohr–Coulomb yield criterion when a2 ¼ 0.
Murnaghan EOS is suitable to describe the adiabatic compression of solid
materials in the pressure range of 0–50 GPa (Rosenberg and Dekel 2009) and
expressed as

 c 
K q
P¼
1 ð3:2Þ
c q0

where K ¼ q0 C02 is the bulk modulus of target material with C0 being the bulk
sound speed, q0 and q are densities of the undeformed and deformed target
material, respectively; c is related to the slope s of the linear Hugoniot relationship
between the shock velocity and particle velocity by

c ¼ 4s
1 ð3:3Þ

In the following, the relationship between the cavity stress and cavity expansion
velocity is obtained numerically with the similarity transformation method, which is
then used to formulate the one-dimensional resistance function of concrete targets
in Sect. 3.3.2. The solutions of the target responses are derived in details.

3.3.1.1 Plastic Region

The equations of momentum and mass conservation in Eulerian coordinates with

spherical symmetry are


@rr 2 ð rr
r h Þ @v @v
þ ¼
q þv ð3:4aÞ
@r r @t @r



@v 2v @q @q
q þ ¼
þv ð3:4bÞ
@r r @t @r

where v is the particle velocity measured positive for outward motion.

Illustrated in Fig. 3.1, by introducing the following non-dimensional variables
and applying similarity transformation suggested by Forrestal and Tzou (1997),
38 3 Rigid Projectile Penetration

r v Vc ft rr P c c1
n¼ ; U¼ ; e¼ ; F¼ ; S¼ ; T¼ ; b¼ ; b1 ¼ ;
ct pffiffiffiffiffiffiffiffiffiffiffiffic c fc fc fc cp cp
cp ¼ Ec =q0
ð3:5Þ

the conservation equations in Eqs. (3.4a, b) are transformed into the

non-dimensional forms as

dT 2gðT Þ  c 1c dU
f ðT Þfc þ ¼ b2 Ec Tfc þ 1 ðn
UÞ ð3:6aÞ
dn n K dn

dU U fc dT
þ2 ¼ ðn
U Þ ð3:6bÞ
dn n K þ Tfc c dn

where ft ; fc , and Ec are the uniaxial tensile strength, compressive strength, and
elastic modulus of the concrete target, respectively. f(T) and g(T) are obtained with
the use of Eqs. (3.1a, b).

1 3a1 þ 6a2 Tfc þ 2a0 a2 þ 2

f ðT Þ ¼
3 a1 þ a2 Tfc

1 3Tfc a1 þ 3a2 T 2 fc2 þ 2a0 a1 þ 2a0 a2 Tfc þ 2Tfc a2

ð3:7aÞ
3 ða1 þ a2 Tfc Þ2

Tfc
gð T Þ ¼ a0 þ ð3:7bÞ
a1 þ a2 Tfc

By using Eqs. (3.1a, b), a dimensionless relationship between the radial stress
S and pressure T is derived as

1 3Tfc a1 þ 3a2 T 2 fc2 þ 2a0 a1 þ 2a0 a2 Tfc þ 2Tfc

S¼ ð3:8Þ
3fc a1 þ a2 Tfc

The standard forms of Eqs. (3.6a, b) suitable for numerical solution by the
Runge–Kutta method are

dU 2Uf ðT ÞðK þ Tfc cÞ þ 2gðT Þðn
U Þ

¼ ð3:9aÞ
dn nb2 E K þ Tfc c
1=c ðn
U Þ2
nf ðT ÞðK þ Tf cÞ
c K c

K þ Tf c
1=c
dT 2b2 Ec K c U ðK þ Tfc cÞðn
U Þ þ 2gðT ÞðK þ Tfc cÞ
¼ K þ Tfc c
1=c ð3:9bÞ
dn f nb E
2
ðn
U Þ2
f nf ðT ÞðK þ Tf cÞ
c c K c c

Equations (3.9a, b) can be solved with the initial values of U and T, i.e., U and
T in the plastic region at n ¼ 1. As discussed by Forrestal and Tzou (1997), the
3.3 Extended Cavity Expansion Model 39

cracked region will vanish when the cavity expansion velocity is larger than a
critical value (c > c1). Consequently, when Hugoniot jump conditions are used, the
initial values of U and T for Eqs. (3.9a, b) are obtained with the use of corre-
sponding values in the cracked region (elastic-cracked-plastic target response) or
elastic region (elastic-plastic target response) at n ¼ 1, respectively. The above two
different target responses are treated separately and presented as follows.
Corresponding to the two different response types discussed above, subscript 1 is
used to denote quantities in the cracked or elastic region at n ¼ 1, subscript 2 are
used to denote quantities in the plastic region at n ¼ 1, and subscripts 3 and 4
denote the quantities in the elastic and cracked regions at the elastic-cracked
interface (n ¼ b1 =b), respectively.

3.3.1.2 Elastic-Cracked-Plastic Response

The elastic region is governed by the elastic wave equation (neglecting the con-
vective term) and Hooke’s law (Forrestal and Luk 1988a)

@ 2 u 2 @u 2u 1 @ 2 u
þ
¼ ð3:10aÞ
@2r r @r r 2 c2d @ 2 t2
 
Ec @u u
rr ¼
ð1
m Þ þ 2m ð3:10bÞ
ð1 þ mÞð1
2mÞ @r r


Ec @u u
rh ¼
m þ ð3:10cÞ
ð1 þ mÞð1
2mÞ @r r

where m is Poisson’s ratio and u is the radial displacement. The dilatational velocity
cd is given by

Ec ð1
mÞ
c2d ¼ ð3:11Þ
ð1 þ mÞð1
2mÞq0

By introducing u ¼ u=ct into Eq. (3.10a) and using the similarity transforma-
tion, it derives

d2 u 2 du 2
1
a2 n2 þ
u¼0 ð3:12Þ
dn 2 n dn n2
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where a ¼ b ð1 þ mÞð1
2mÞ=ð1
mÞ. The general solution of Eq. (3.12) was
presented by Forrestal and Luk (1988a) as
40 3 Rigid Projectile Penetration

1
3a2 n2
u ¼ A0 an
B0 ð3:13Þ
3a2 n2

where A0 and B0 are the integration constants, which are determined by the radial
displacement at n ¼ 1=a and circumferential stress in the elastic region at n ¼ b1 =b
as follows

uðn ¼ 1=aÞ ¼ 0 ð3:14aÞ

rh ðn ¼ b1 =bÞ ¼
ft ð3:14bÞ

Using Eqs. (3.10c, 3.13, 3.14a, b) and the relationship between particle velocity
and radial displacement (U ¼ u
ndu=dn), the non-dimensional radial stress S3
and particle velocity U3 in the elastic region at the elastic-cracked interface
(n ¼ b1 =b) are obtained as
 
ð1
2mÞ þ 3ma2
ð1 þ mÞa3
S3 ¼ 2F ð3:15aÞ
1
2m
3a2 þ 2ð1 þ mÞa3
 
3ð1 þ mÞð1
2mÞft b1 1
a2
bU3 ¼ ð3:15bÞ
Ec 1
2m
3a2 þ 2ð1 þ mÞa3

When the Hugoniot jump conditions are applied on the elastic-cracked interface,
the non-dimensional radial stress S4 and particle velocity U4 in the cracked region at
the elastic-cracked interface (n ¼ b1 =b) can be obtained as follows

2F ðb1
bU3 Þ2
S4 ¼ S3 þ ð3:16aÞ
3
ðb1
bU3 Þ2

6ft ð1
2mÞðb1
bU3 Þ

bU4 ¼ bU3 þ h i ð3:16bÞ
3
ðb1
bU3 Þ2 Ec

In the cracked region, the concrete material is treated as linear compressible with
rh ¼ 0: The general solutions of the non-dimensional radial stress and particle
velocity in the cracked region were presented by Forrestal and Tzou (1997) as
" #
E0 3
Sð nÞ ¼ þ D0 1 þ ð3:17aÞ
bn ðbnÞ2
" #

3fc ð1
2mÞ E0 E0 4D0
U ð nÞ ¼ þ þ ð3:17bÞ
2Ec b 3 ðbnÞ2 bn
3.3 Extended Cavity Expansion Model 41

where E0 and D0 are the integration constants, which are determined by the
boundary conditions at the elastic-cracked interface (n ¼ b1 =b) given in
Eqs. (3.16a, b) as



6b21 2b1 fc S4 þ Ec b21 þ 3 ðbU4 Þ=3ð1
2mÞ
E0 ¼
2 ð3:18aÞ
fc b21
3

b21 ½fc S4 þ 2Ec b1 ðbU4 Þ=ð1
2mÞ

D0 ¼
2 ð3:18bÞ
fc b21
3

At the elastic-cracked interface (n ¼ b1 =b) in the cracked region, the hyperbolic

yield criterion satisfies, thus

rr =3
r r ¼ a0 þ ð3:19aÞ
a1 þ a2 rr =3

rr =fc ¼ E0 =b þ D0 1 þ 3=b2 ð3:19bÞ

By using Eqs. (3.19a, b), b is formulated as

 1=2
E0 þ E02 þ 12D0 ðY0 =fc
D0 Þ
b¼ ð3:20Þ
2ðY0 =fc
D0 Þ

where
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

3a1 þ a0 a2 þ 1 þ 9a21 þ 6a2 a0 a1
6a1 þ a20 a22 þ 2a0 a2 þ 1
Y0 ¼ ð3:21Þ
2a2

The non-dimensional particle velocity U1, radial stress S1, and pressure T1 in the
cracked region at the cracked-plastic interface (n ¼ 1) are summarized as follows:

fc b2 þ 3 ð1
2mÞE0 6fc D0 ð1
2mÞ

U1 ¼
ð3:22aÞ
2Ec b 3
Ec b2

Y0 Y0
S1 ¼ ; T1 ¼ ð3:22bÞ
fc 3fc

3.3.1.3 Elastic-Plastic Response

The cracked region will vanish when c > c1, and consequently, the response of the
concrete target is an elastic-plastic one. The general solution for u is the same as
42 3 Rigid Projectile Penetration

Eq. (3.13), except that the integration constants are determined by the following
boundary conditions.

uðn ¼ 1=aÞ ¼ 0 ð3:23aÞ

rr ðn ¼ 1Þ þ 2rh ðn ¼ 1Þ
rr ðn ¼ 1Þ
rh ðn ¼ 1Þ ¼ a0 þ ð3:23bÞ
3a1 þ a2 ðrr ðn ¼ 1Þ þ 2rh ðn ¼ 1ÞÞ

By using Eqs. (3.10b, c, 3.13, 3.23a, b), the non-dimensional particle velocity
U1, radial stress S1, and pressure T1 in the elastic region at the elastic-plastic
interface (n ¼ 1) are obtained as follows

3 A0 ð1
3a2 Þ
U1 ¼ þ 3A0 ð3:24aÞ
2 a2

Ec A0 ða3
1 þ ma3 þ 2m
3ma2 Þ ða
1ÞEc A0

S1 ¼ ; T1 ¼ ð3:24bÞ
fc ð1 þ mÞð2m
1Þa2 ð2m
1Þfc

3.3.1.4 Hugoniot Jump Conditions

The non-dimensional particle velocity, radial stress, and pressure in the cracked
region at the cracked-plastic interface and in the elastic region at the elastic-plastic
interface are obtained so far, respectively. However, in order to solve the
non-dimensional conservation Eqs. (3.9a, b), the non-dimensional particle velocity
and pressure in the plastic region at the cracked-plastic interface or in the plastic
region at the elastic-plastic interface are required, which are determined by the
Hugoniot jump conditions.
For both cracked and elastic regions at n ¼ 1, the concrete target is considered as
linear compressible, thus


Ec q0
T1 fc ¼ 1
ð3:25Þ
3ð1
2mÞ q1

When Murnaghan EOS is used for the plastic region, it has

1c
cfc
q2 ¼ q0 T2 þ1 ð3:26Þ
K

The non-dimensional Hugoniot jump conditions are (Forrestal and Tzou 1997)

U2 ¼ 1
ðq1 =q2 Þð1
U1 Þ ð3:27aÞ

3.3 Extended Cavity Expansion Model 43

S1 fc þ q1 U1 cðU1 c
cÞ ¼ S2 fc þ q2 U2 cðU2 c
cÞ ð3:27bÞ

The non-dimensional particle velocity U2, radial stress S2, and pressure T2 in the
plastic region at n ¼ 1 are obtained by solving Eqs. (3.8, 3.25–3.27a, b), which are
then served as initial values for the numerical solution of the non-dimensional
Eqs. (3.9a, b).
For numerical solution, the inverse procedure proposed by Forrestal and Tzou
(1997) is used to calculate the relationship between the non-dimensional radial
stress at the cavity surface and the cavity expansion velocity. The detailed solution
process is given as follows:
(i) A value of b1 is chosen and b is determined by Eq. (3.20).
(ii) The non-dimensional particle velocity U1 ; radial stress S1 , and pressure T1
are obtained. In details, if b1 [ b; the response of the concrete target is the
elastic-cracked-plastic type. Therefore, U1 ; S1 , and T1 are determined by
Eqs. (3.22a, b). Otherwise, the response of the concrete target is an
elastic-plastic one. Consequently, U1 ; S1 , and T1 are obtained by using
Eqs. (3.24a, b).
(iii) The non-dimensional particle velocity U2 ; radial stress S2 , and pressure T2 in
the plastic region at n ¼ 1 are obtained by solving Eqs. (3.8, 3.25–3.27a, b),
which are then served as the initial values for numerical solution of
Eqs. (3.9a, b).
The above numerical solution procedure starts from n ¼ 1 to the cavity surface
at n ¼ e. When the boundary condition of the particle velocity at the cavity surface
U ðn ¼ eÞ ¼ e is satisfied, the radial stress S and its corresponding cavity expansion
velocity Vc ¼ becp are obtained. The relationship of the radial stress at the cavity
surface rr to the cavity expansion velocity Vc is obtained by curve-fitting method as
!2
rr Vc Vc
¼ A þ B pffiffiffiffiffiffiffiffiffiffi þ C pffiffiffiffiffiffiffiffiffiffi ð3:28Þ
fc fc =q0 fc =q0

where A, B, and C are the non-dimensional material constants.

Concrete with compressive strength of 48 MPa is chosen as an example to gain
some insights into the extended dynamic cavity expansion model. The shear
strength–pressure experimental data and the fitted hyperbolic yield criteria are
presented in Fig. 3.2a. It can be clearly observed that the proposed hyperbolic yield
criterion reproduces the test data excellently with a0 = 27.58 MPa, a1 = 0.74, and
a2 = 2.14e−4 MPa−1, while the Mohr–Coulomb yield criterion fails to correlate
well with the test data. For the EOS of concrete material, the available experimental
data is limited. The isotropic compression data from Hanchak et al. (1992) for
48-MPa concrete as well as the full-scale explosive detonation and flyer-plate-
44 3 Rigid Projectile Penetration

Fig. 3.2 Experimental data (a) 0.8

for a yield criterion b EOS Hanchak et al. (1992)
0.7 Mohr-Coulomb yield criterion
Hyperbolic yield criterion Eq. (3.1)
0.6

0.5

(GPa)
0.4

-
0.3

r
0.2

0.1

0.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
P (GPa)

(b) 8
Hanchak et al. (1992)
7 Gebbeken et al. (2006)
Linear EOS
6
Murnaghan EOS Eq. (3.2)
5
P (GPa)

0
0.00 0.05 0.10 0.15 0.20 0.25
1- 0/

impact data from Gebbeken et al. (2006) for 51.2-MPa concrete are plotted on
Fig. 3.2b. Using the Murnaghan EOS with C0 = 1945 m/s and s = 2.76 gives a
satisfied agreement with the experimental data. In addition, it indicates that the
frequently used linear elastic EOS cannot agrees well with the experimental data.
Figure 3.3 presents separately the predictions of the interface velocities and
radial stresses at the cavity surface by the extended dynamic cavity expansion
model for concrete (fc = 48 MPa, q0 = 2440 kg/m3), where the velocity range is
pffiffiffiffiffiffiffiffiffiffi
0\Vc = fc =q0 \8 corresponding to 0\Vc \1100 m/s. It is observed that the
pffiffiffiffiffiffiffiffiffiffi
cracked region disappears when Vc = fc =q0 [ 2:8. It should be pointed out that, in
Fig. 3.3a, the non-intersection of the three lines in the circle is attributed to the
numerical solution deviation and the simplicity of Eqs. (3.16a, b) (Forrestal and
Tzou 1997).
3.3 Extended Cavity Expansion Model 45

Fig. 3.3 Predictions of (a) 20

48-MPa concrete targets
18
a interface velocities versus
cavity expansion velocity 16
b cavity surface radial stresses 14
versus cavity expansion
velocity 12

10

8
c1 /( fc/ 0)
1/2
6 elastic-cracked-
c/( fc/ 0)
1/2
4 plastic response
c/( fc/ 0) , elastic-plastic response
1/2
2

0
0 1 2 3 4 5 6 7 8
Vc /( fc/ 0 ) 1/2

(b) 80
Elastic-cracked-plastic response
70
Elastic-plastic response
60

50
/fc

40
r

30

20

10

0
0 1 2 3 4 5 6 7 8
Vc /( fc/ 0 )
1/2

3.3.2 One-Dimensional Resistance Function

of Concrete Targets

An arbitrary nose shape of a projectile is shown in Fig. 3.4, where d is the nose
diameter and V is the instantaneous penetration velocity. y = y(x) is the nose
contour function in the coordinate shown in Fig. 3.4, and l0 is the length of pro-
jectile nose. The normal compressive stress rr acted on the projectile nose is
obtained by Eq. (3.28) presented in Sect. 3.3.1.
Neglecting the tangential friction stress (Wu et al. 2012a), the resulting axial
resistance Fx is determined by integrating the normal compressive stresses on the
projectile nose surface as
46 3 Rigid Projectile Penetration

Fig. 3.4 Schematic of an y

arbitrary-nosed projectile
y(x) r

d/2
V

0 l0 x

(a) d (b) (c) d

d

ψ
d×
σr σr ψ
σr σr
d×

σr σr
σ
r
σr σr σr
σr
r
σ

d
(d) d (e)
ψ
d×

σr
r
σ
σr

σr

Fig. 3.5 Projectile nose geometry a ogive b hemispherical c conical d blunt, and e flat

pd 2  pffiffiffiffiffiffiffiffi 
Fx ¼ N 0 Afc þ N 1 B q0 fc V þ N 2 Cq0 V 2 ð3:29Þ
4

where N 0 ; N 1 , and N 2 are the projectile nose shape coefficients with the expressions
of

Zl0 Zl0
8 yy02 8 yy03
N 0 ¼ 1; N1 ¼ 2 pffiffiffiffiffiffiffiffiffiffiffiffiffi; N2 ¼ 2 ð3:30Þ
d 1 þ y02 d 1 þ y02
0 0

In particular, for the ogive-nosed projectile illustrated in Fig. 3.5a, it has

3.3 Extended Cavity Expansion Model 47


 
2f31 2w
1 ð2w
1Þf1 f1 p
N 0 ¼ 1; N 1 ¼ 4w2 þ þ arccos
;
3w3 2w 2w2 w 2
1 1
N2 ¼
; w  0:5
3w 24w2
ð3:31Þ

where f21 ¼ w2
ðw
1=2Þ2 , and w is the caliber-radius-head (CRH) of the nose.

For the hemispherical-nosed projectile illustrated in Fig. 3.5b, it has

N 0 ¼ 1; N 1 ¼ 2=3; N 2 ¼ 1=2; w ¼ 0:5 ð3:32Þ

For the conical-nosed projectile illustrated in Fig. 3.5c, it has

1 1
N 0 ¼ 1; N 1 ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; N2 ¼ ; f2 ¼ l0 =d ð3:33Þ
4f22 þ 1 4f22þ1

For the blunt-nosed projectile illustrated in Fig. 3.5d, it has

2
3=2
8w2 4w
1 1
N 0 ¼ 1; N1 ¼
; N2 ¼ 1
; w  0:5 ð3:34Þ
3 3w 8w2

For the flat-nosed projectile illustrated in Fig. 3.5e, it has

N0 ¼ N1 ¼ N2 ¼ 1 ð3:35Þ

It should be pointed out that when the contribution of the first order of the
velocity to the axial resistance is neglected, that is N 1 ¼ 0, Eqs. (3.31–3.35) reduce
to the expressions by Chen and Li (2002).
The averaged one-dimensional resistance Fmean could be derived by normalizing
the resistance Fx in Eq. (3.29) as

Fx pffiffiffiffiffiffiffiffi
Fmean ¼ ¼ N 0 Afc þ N 1 B q0 fc V þ N 2 Cq0 V 2 ð3:36Þ
pd 2 =4

Based on the following discussions both in Sects. 3.3.3 and 5.2, it indicates that
Fmean can be conveniently utilized in both motion equation for rigid projectile and
the pressure balance equation for eroding projectile; thus, Fmean can be regarded as a
unified resistance to projectile penetration.
48 3 Rigid Projectile Penetration

3.3.3 Analysis and Discussion

When the deformation and mass loss of projectile are negligible during penetration,
such as the projectiles shown in Fig. 3.6, the projectiles can be regarded as rigid
bodies in the analyses.

3.3.3.1 Penetration Analysis

The projectile penetration into concrete targets commonly induces a conical crater
followed by a tunnel with the diameter nearly equaling to the projectile shank
diameter. Since the contribution of the cratering stage to deep penetration depth is
small, the dynamic cavity expansion model is used in the whole penetration
process.
By applying one-dimensional momentum theorem equation, it has

Fig. 3.6 Unfired and posttest

photographs of projectiles
(Frew et al. 1998)
3.3 Extended Cavity Expansion Model 49

qp Leff dV=dt ¼
Fmean ð3:37Þ

where Leff ¼ 4M= qp pd 2 is the effective length of the projectile. Using the
equation of dV=dt ¼ VdV=dhpen ; the non-dimensional DOP can be obtained by
integrating Eq. (3.37) as follows
pffiffiffiffiffiffiffiffi ! pffiffiffiffiffiffiffiffi
hpen qp Afc N 0 þ N 1 B q0 fc V0 þ N 2 Cq0 V02 qp N 1 B q0 fc
¼ ln þ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
Leff 2N 2 Cq0 Afc N 0 2
N 2 Cq0 q0 fc 4ACN 0 N 2
N 1 B2
2 0 1 0 13
pffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi
6 B N 1 B q0 fc C B N 1 B q0 fc þ 2N 2 Cq0 V0 C7
6arctanBrffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi C B C7
4 @  A
arctan@rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  A5
2 2 2 2
q0 fc 4ACN 0 N 2
N 1 B q0 fc 4ACN 0 N 2
N 1 B

ð3:38Þ

where V0 is the initial velocity of the projectile.

3.3.3.2 Comparison with Test Data

In this section, based on the rigid projectile penetration tests conducted by Forrestal
et al. (1994, 1996, 2003a), the DOP predicted by Eq. (3.38) with the extended
dynamic cavity expansion model, and the DOP predicted the classical dynamic cavity
expansion model proposed by Forrestal and Tzou (1997) are compared, respectively.
Forrestal et al. (2003a) presented ogive-nosed projectiles penetration test data on
23-MPa concrete targets with different impact velocities. The corresponding shear
strength–pressure relations between the triaxial compression tests (Warren et al.
2004a) and the fitted hyperbolic and Mohr–Coulomb yield criteria are shown in
Fig. 3.7.

Fig. 3.7 Shear strength– 0.8

pressure relations for 23-MPa Warren et al. (2004a)
0.7 Mohr-Coulomb
concrete
0.6
Eq. (3.1)

0.5 - 15.2e6+P/(2.44+2.1e-10P)
(GPa)

r
0.4
-

0.3
r

0.2 - 26e6+0.35P
r
0.1

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
P (GPa)
50 3 Rigid Projectile Penetration

Fig. 3.8 Non-dimensional 5

DOP versus striking velocity Forrestal et al. (2003a)
for 23-MPa concrete targets Classic CET model in Forrestal and Tzou (1997)
4 Extended CET model in Eq. (3.38)

hpen/Leff
2

0
0 50 100 150 200 250 300 350 400 450
V0 (m/s)

The classical dynamic cavity expansion model proposed by Forrestal and Tzou
(1997) uses the linear elastic EOS for 23-MPa concrete material, while the pre-
sented extended dynamic cavity expansion model uses Murnaghan EOS fitted from
Fig. 3.2b.
Predictions by Eq. (3.38) for both extended and classical dynamic cavity
expansion models as well as the penetration test data (Forrestal et al. 2003a) are
shown in Fig. 3.8. It is observed that the extended dynamic cavity expansion model
gives better predictions, while the classical dynamic cavity expansion model
underestimates the experimental results. The reason lies in the inappropriate use of
the linear elastic EOS since the yield criteria are almost identical as shown in
Fig. 3.7. As shown in Fig. 3.2b, for 23-MPa concrete, the linear elastic EOS is
higher than Murnaghan EOS when the volumetric strain is less than 0.033, which
agrees with the experimental observations by Holmquist et al. (1993). Since the
pressure at the projectile/target interface is relatively low for the smaller experi-
mental velocity (less than 400 m/s), the resistance is overestimated using the linear
EOS, and consequently, the DOP is underestimated.
Forrestal et al. (1996) presented another DOP test data for 51-MPa concrete
targets and rigid ogive-nosed projectiles. Since the shear strength–pressure relation
for 51-MPa concrete is not available in relevant literatures, the test data from
48-MPa concrete shown in Fig. 3.2a is adopted, and Murnaghan EOS drawn from
Fig. 3.2b is utilized simultaneously. The predictions by both the extended and
classical dynamic cavity expansion models as well as the penetration test data are
presented in Fig. 3.9. It also indicates that the extended dynamic cavity expansion
model shows better agreement with the test data than the classical dynamic cavity
expansion model. The overestimated predictions from the classical dynamic cavity
expansion model lies both in the inaccurate linear Mohr–Coulomb yield criterion
and EOS illustrated in Fig. 3.2.
3.3 Extended Cavity Expansion Model 51

3.3.3.3 Discussion

Aiming to determine the parameters of the hyperbolic yield criterion and

Murnaghan EOS, i.e., a0, a1, a2, C0, and s, the shear strength–pressure data from
triaxial compression test and the pressure-volumetric strain test data from full-scale
detonation or flyer-plate-impact tests are required in the extended dynamic cavity
expansion model. However, these experiments are difficult to conduct and always
absent in the existing DOP tests. Under this circumstance, a0, a1, and a2 suggested
by Markovich et al. (2011) as well as C0 and s obtained by data fitting illustrated in
Fig. 3.2b are proposed as follows.

a0 ¼ 2:442fc0:4369 ; a1 ¼ 1:084fc
0:2463 ; a2 ¼ 0:03276fc
0:6416 ð3:39aÞ

C0 ¼ 1945m=s; s ¼ 2:76 ð3:39bÞ

where the unit of fc is MPa.

Figure 3.10 shows the predictions of non-dimensional DOP by Eq. (3.38) for
three sets of tests, in which the parameters of the hyperbolic yield criterion and
Murnaghan EOS are given in Eqs. (3.39a, b). It is observed that the predictions
agree with the experimental data reasonably; thus, the validities of Eqs. (3.38,
3.39a, b) are verified.

3.4 Unified Deep Penetration Model

In this section, aiming to establish a simple and effective projectile penetration

model into various targets (e.g., concrete, metal, and rock), by assuming that the
resistance acting on the projectile keeps unchanged during the penetration process,

Fig. 3.9 Non-dimensional 9

DOP versus striking velocity
8
Forrestal et al. (1996)
for 51-MPa concrete targets Classic CET model in
7 Forrestal and Tzou (1997)
6 Extended CET model in
Eq. (3.38)
Hpen/Leff

0
0 200 400 600 800 1000 1200 1400
V0 (m/s)
52 3 Rigid Projectile Penetration

Fig. 3.10 Experimental and (a) 3.5

predicted non-dimensional Forrestal et al. (2003a)
DOP with concrete target 3.0 Eqs. (3.38, 3.39)
strengths of a 39 MPa
b 62.8 MPa, and c 36.2 MPa 2.5

2.0

hpen/Leff
1.5

1.0

0.5

0.0
0 100 200 300 400 500 600
V0 /m/s

(b) 7
Forrestal et al. (1996)
6 Eqs. (3.38, 3.39)

4
hpen/Leff

0
0 200 400 600 800 1000 1200
V0 /m/s

(c) 7
Forrestal et al. (1994)
6 Eqs. (3.38, 3.39)

4
hpen/Leff

0
0 100 200 300 400 500 600 700 800 900 1000
V0 /m/s
3.4 Unified Deep Penetration Model 53

a new mean resistance approach based on dynamic cavity expansion approximation

is proposed. A simple and effective projectile penetration model is further given to
predict the DOP of different nose-shaped hard projectiles penetrating into diversi-
fied targets. Besides, the related mean resistance coefficient is confirmed as 0.4
based on the parametric analyses.

3.4.1 Mean Resistance

The instantaneous axial resistance Fx acting on the projectile during penetration

process mainly consists of two parts, the quasi-static resistance term and the
dynamic term (or the inertial term) induced by the projectile velocity

pd 2
Fx ¼ ðN1 rs þ BN2 q0 V 2 Þ ð3:40aÞ
4
Zl0
8l
N1 ¼ 1 þ 2m ydx ð3:40bÞ
d
0

Zl0 Zl0
8 yy03 8l yy02
N2 ¼ 2 dx þ 2m dx ð3:40cÞ
d 1þy 02 d 1 þ y02
0 0

where rs is the quasi-static target material strength; B is the coefficient of dynamical

resistance derived from dynamic cavity expansion analysis; N1, N2 are projectile
nose shape factors; lm is the sliding friction coefficient between the projectile and
target during impact; and y = y(x) is the nose shape function as shown in Fig. 3.4.
Without considering the initial impact cratering stage, the penetration depth hpen
can be derived by integrating the Newton’s second law with consideration of the
Eqs. (3.40a, b, c) (Forrestal et al. 1994)


2M BN2 q0 V02
hpen ¼ 2 ln 1 þ ð3:41Þ
pd BN2 q0 N1 rs

where M and V0 are the mass and the striking velocity of a projectile, respectively.
The deviation of considering the entrance crater effects or not is about kd/2 for
penetration depth based on the study of Chen and Li (2002), where kd is the frontal
impact crater depth. Forrestal et al. (1994) and Li and Chen (2003) suggested k = 2
and k = 0.707 + l0/d, respectively. The entrance cratering region is neglected since
it has little influence on DOP for deep penetrations. The resistance in Eqs. (3.40a, b,
c) is complicated since it is dependent on the instantaneous projectile velocity. As
described in the following, we will present a mean resistance function which keeps
54 3 Rigid Projectile Penetration

Fig. 3.11 Diagram of

projectile deceleration versus
50000 Instantaneous
instantaneous penetration deceleration
depth
1 Mean deceleration

Deceleration (g)
40000
A2 A1=A2
1' 2'
A1
30000
2

20000

10000

3
0 hpen
0.0 0.2 0.4 0.6 0.8 1.0
x (m)

unchanged during the tunneling stage. Figure 3.11 illustrates the typical decelera-
tion–instantaneous penetration depth curve (red dashed line 0-1-2-3) of the pro-
jectile penetrating into concrete targets, which could be obtained from Eqs. (3.40a,
b, c, 3.41). For rigid projectile penetration, the ascending part 0-1 and descending
part 1-2 correspond to the projectile cratering and tunneling stages, respectively,
and the projectile stops penetration at the point 2. For simplicity, the mean decel-
eration during the whole penetration process which is only dependent on the initial
striking velocity (black solid line 0-1′-2′-3) is proposed and illustrated in Fig. 3.11.
The precondition of which is that the works done by the actual and mean resistances
are equal, that is to say the areas A1 = A2 is satisfied in Fig. 3.11.
Introducing a mean resistance coefficient l into Eqs. (3.40a, b, c), the mean
resistance Fm can be written as

pd 2
Fm ¼ ð1 þ ldÞN1 rs ð3:42aÞ
4

I0 BN2 q0 V02
d¼ ¼ ð3:42bÞ
N N1 rs

The parameter d denotes the ratio of dynamic to quasi-static resistance. I0 and

N are two dimensionless parameters proposed by Chen and Li (2002)

MV02 M
I0 ¼ ; N¼ ð3:43Þ
N1 rs d 3 Bq0 d 3 N2

As for concrete and metal targets, the related parameters in Eqs. (3.42a, b) were
suggested as follows:
Concrete Forrestal et al. (1994), Frew et al. (1998):
3.4 Unified Deep Penetration Model 55

Table 3.1 Nose shape factor (without consideration of friction, Chen and Li 2002)
Nose shape N1 N2 Definition
Ogival 1 1=3w
1=24w2 (0 < N2 < 0.5) w is CRH

Conical 1 1= 1 þ 4w2 (0 < N2 < 1) w = l0/d

Blunt 1 1
1=8w (0.5 < N2 < 1)
2 w = r/d and r is the radius of sphere

rs ¼ Sfc ; S ¼ 82:6ðfc =106 Þ
0:544 ; B¼1 ð3:44aÞ

Metal Forrestal et al. (1995, 2008):


n  Zb
2Y 2E ð
ln xÞn 3Y
rs ¼ 1þ I ; I¼ dx; b¼1
; B ¼ 1:5
3 3Y 1
x 2E
0
ð3:44bÞ

where fc is the uniaxial compressive strength of concrete Pa, E is Young’s modulus,

Y is the yield stress, and n is the strain-hardening exponent to be determined by the
curve fitting to the compression stress–strain test data of metal targets. We can also
find the related parameters for soil, limestone targets, etc. in Forrestal and Luk
(1992a) and Warren et al. (2004b).
The nose shape factors N1 and N2 for common projectiles are listed in Table 3.1,
the other nose shape factor such as truncated-ogive nose and truncated-conical nose
can be found in Chen and Li (2002) and Wu et al. (2012a). For metal targets, the
friction at the projectile/target interface could not be neglected. Thus, we multiply
the nose shape factor N1 and N2 listed in Table 3.1 by a coefficient of 1.1 in the
following predictions for simplicity.
It can be found that a larger N corresponds to a sharper and slender projectile and
I0 is mainly dependent on the striking velocity according to Eq. (3.43). For a given
target, parameter d in Eq. (3.42b) is closely related to the striking velocity V0 and
nose shape factor N2. For instance, Fig. 3.12 shows the variation of d for various
projectiles penetration into a concrete target with fc = 40 MPa. The plane (V0, N2)
in Fig. 3.12 is limited in a realistic variation range of rigid projectiles (Li and Chen
2003), and it shows that the value of d is less than 5. Actually, d < 1 is always
satisfied for ogive-nosed or conical-nosed projectiles.

3.4.2 DOP

For deep penetration, considering the work done by the mean resistance equals to
the initial kinetic energy of a projectile, it is obtained that
56 3 Rigid Projectile Penetration

0
1
0.8 1000
0.6 800
0.4 600
N2 0.2 400
200 V0 (m/s)
0 0

Fig. 3.12 Dependence of d on striking velocity V0 and nose shape factor N2

1
Fm hpen ¼ MV02 ð3:45Þ
2

Substituting Eqs. (3.42a, b) into Eq. (3.45) gives the dimensionless DOP as

hpen 2 I0
¼ ð3:46Þ
d p ð1 þ ldÞ

When l = 0, the inertial term is neglected, Eq. (3.46) is simplified to

hpen 2
¼ I0 ð3:47Þ
d p

Equation (3.47) was proposed by Forrestal et al. (2003a) and Chen and Li
(2002) to predict the low-to-mid speed impact of non-flat projectiles. It should be
noted that Eq. (3.46) is only suitable for deep penetration, and adding kd/2 to the
DOP predicted by Eq. (3.46) is suggested for concrete targets when the final
penetration depth is relatively shallow.
The value of the mean resistance factor l is derived by substituting Eq. (3.41)
into the DOP in Eq. (3.46)

1 1
l¼
ð3:48Þ
lnð1 þ dÞ d

Figure 3.13 plots the relationship of the parameters l and d according to

Eq. (3.48). It indicates that l drops slightly with the increase of d, and the value of
l is almost around 0.4 when d < 5 as shown in Fig. 3.12. We attempt to set a
constant l1 instead of the coefficient l expressed by Eq. (3.48), by substituting l1
3.4 Unified Deep Penetration Model 57

Fig. 3.13 The relationship of 1.0

l with d Eq. (3.48)

0.8

0.6

0.4

0.2

0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0

and Eq. (3.48) into Eq. (3.46), the deviations of the two predicted DOPs can be
expressed as follows:

jl1
ljd
D¼  100% ð3:49Þ
1 þ l1 d

Figure 3.14 shows the above deviation versus the parameter d, which is pre-
dicted by Eq. (3.49) with l1 = 0.4. It indicates that the deviation is less than 7%
when d < 5. In particular, for the ogive-nosed or conical-nosed slender projectiles
(d < 1), the deviation is further less than 3%. Therefore, l = 0.4 is assumed
reasonably.
Now it is interesting to examine the application range of the simple formula
Eq. (3.47), which neglects the inertial term. Substituting Eq. (3.48) and l1 = 0 into
Eq. (3.49) will give the deviation of the DOP predictions by Eq. (3.47). If the

Fig. 3.14 The relationship of 8

D with d Eq. (3.49) with 1
=0.4
7

0
0 1 2 3 4 5
58 3 Rigid Projectile Penetration

deviation of 10% is accepted, the application range of Eq. (3.47) proposed by

Forrestal et al. (2003a) and Chen and Li (2002) is limited to d < 0.2.

3.4.3 Validation

Figure 3.15 illustrates the comparisons between the DOP predicted by Eq. (3.46)
with l = 0.4 and the available test data for the following projectiles and targets:
(a) ogive-nosed projectiles penetrating into concrete targets, (b) flat, conical-nosed
and truncated-ogive-nosed projectiles penetrating into concrete targets,
(c) ogive-nosed projectiles penetrating into metal targets, and (d) ogive-nosed
projectiles penetrating into limestone targets. It can be seen that good agreements
are obtained between the predicted results and the test data.

(a) (b)
120 Eq. (3.46) 25
Eq. (3.46)
Forrestal et al. (1994) Forrestal et al. (1994)
Conical
Forrestal et al. (1994) Forrestal et al. (1996)
100 Forrestal et al. (1996) Forrestal et al. (1996)
Shi et al. (2005) Shi et al. (2005)
Frew et al. (1998) Frew et al. (1998)
20 Flat
Liang et al. (2008) O'Neil et al. (1999) Teland et al. (2004) Teland et al. (2004)
80 Unosson and Nilsson (2006) Xu and Lin (2002) Teland et al. (2004) Tai (2009)
Xu and Lin (2002) Xu and Lin (2002) 15 Wang et al. (2006)
hpen/ d

hpen/ d

Zhang et al. (2005) Truncated-ogive

60 Zhang et al. (2005) Qian et al. (2000)
Zhou et al. (2006a)
10
40

20 5

0 0
0 30 60 90 120 150 180 0 5 10 15 20 25 30 35 40 45
I0 I0

(c) (d)
75 90
80
60 70
60
45
50
hpen/ d
hpen/ d

Eq. (3.46) (with friction) 40

30 Forrestal et al. (1988b)
Forrestal et al. (1988b) 30 Eq. (3.46)
Forrestal et al. (1988b)
Frew et al. (2000) d=12.7mm
15 Forrestal et al. (1991) 20 Frew et al. (2000) d=7.11mm
Forrestal et al. (1992b)
Frew et al. (2000) d=25.4mm
Forrestal and Piekutowski (2000) 10 Forrestal and Hanchak (2002) d=7.11mm
Piekutowski et al. (1999)
0 0
0 30 60 90 120 150 0 20 40 60 80 100 120 140 160 180
I0 I0

Fig. 3.15 Test data and predicted curves of hpen/d–I0 a concrete b concrete c metal, and
d limestone
3.4 Unified Deep Penetration Model 59

It should be pointed out that the values of the parameter N are different for the
projectiles in the related tests, and we select the value of N which is corresponding
to the tests with the maximal impact factor I0 for prediction. The reason is that the
values of N have slight influence on the DOP when I0 is relatively small, which can
be found in Eq. (3.49) and Fig. 5 from Li and Chen (2003).

3.5 Perforation Model of Thick Concrete Slabs

For the existing experimental studies of projectile perforating the concrete slab,
Hanchak et al. (1992) conducted perforation experiments with 25.4-mm-diameter
projectiles (301–1058 m/s) into 178-mm-thick concrete slabs (48 and 140 MPa)
and measured residual velocities with X-ray photographs. Cargile et al.
(1993) captured and discussed the residual velocities of projectiles with 50.8 mm in
diameter perforating concrete slabs with four thicknesses (284.4, 254, 215.9, and
127 mm) at a striking velocity of approximately 312.9 m/s. Besides, four shots of
projectiles perforation test with striking velocities of 379–470 m/s on
254-mm-thick slabs were performed. Unosson and Nilsson (2006) published four
experimental residual velocities from high-speed camera, in which the projectile
diameter and concrete thickness were 75 and 400 mm, respectively. Li et al.
(2013) acquired the residual velocities of the projectiles by designing a foil screen
behind the concrete slab, in which the projectiles with 64 mm in diameter were
used to strike five concrete slabs with the thicknesses of 300–700 mm at a nominal
impact velocity of 400 m/s. Wu et al. (2015a) performed twenty-five shots of
projectiles (25.3 mm in diameter) perforation test on five configurations of
monolithic and segmented concrete slabs (100–300 mm thick), and the residual
velocities of the projectiles after perforating every slab were recorded.
For the analytical works, based on the two-stage (cratering + shear plugging) or
three-stage (cratering + tunneling + shear plugging) projectile perforation models,
Chen et al. (2004, 2008b) proposed a formulae to predict the projectile residual
velocity, which was validated by the test data from Hanchak et al. (1992).
Considering the kinetic energy carried by the rear-ejected fragments of concrete
slabs, Wu et al. (2012a) presented a modified expression for the residual velocity of
projectiles. Furthermore, Grisaro and Dancygier (2014a) proposed a modified
energy method to assess the residual velocity of projectiles perforating the concrete
barrier, in which the energies dissipated through fracture of the ejecting crater into
fragments as well as the additional cracking of the slab were empirically assumed.
In this section, for the projectile perforation of thick concrete targets illustrated
in Fig. 2.2a, an experiment-based simplified semi-analytical perforation model for
thick concrete slabs is further derived. The existing method for predicting the rear
crater height is improved and the kinetic energy carried by the scabbing fragments
is considered quantitatively.
60 3 Rigid Projectile Penetration

3.5.1 Three-Stage Perforation Model

The three-stage (cratering + tunneling + shear plugging) projectile perforation

model, as shown in Fig. 3.16, is always utilized to analyze thick concrete slabs
under projectile impact. Consider a projectile normally impacting on a concrete slab
with the thickness of H and initial velocity V0, the perforation process mainly
consists of the initial cratering, tunneling, and rear plugging. The plugging can be
assumed as a nearly frustum of cone with the slope angle of /. The ejecting velocity
of the fragments Vrf is assumed as Vrf ¼ gVr ; where Vr denotes the residual velocity
of the projectile.

3.5.2 Height of Rear Crater

The height of rear crater Hcr is usually predicted by theoretical formula (Wu et al.
2012a; Chen et al. 2008b; Li and Tong 2003) on the basis of the assumption that
plug is separated from the surrounding material as soon as the shear failure criterion
is satisfied along the plugging surface. This approach is complex and the predicted
heights are not verified by the related experimental data.
Figure 3.17 shows the dimensionless rear craters heights Hcr/d from the existing
projectile perforation tests. It indicates that all the test data of Hcr/d is about 2.5 and
it shows that the rear crater height is not influenced heavily by the striking velocity,
slab’s height, and strength of concrete targets. Thus, the height of rear crater during
the projectile perforation of thick concrete slabs is proposed as

Hcr ¼ 2:5d ð3:50Þ

Fig. 3.16 Schematic for Front Cratering

projectile normally
perforation of the thick
concrete target

Tunnelling
Projectile
H
Hcr

Target

Fragments Vrf = Vr
3.5 Perforation Model of Thick Concrete Slabs 61

Fig. 3.17 Test data of rear 5.0

crater height Wu at al. (2015a) (0.15m, 41MPa)
4.5 Wu at al. (2015a) (0.2m, 41MPa)
Wu at al. (2015a) (0.3m, 41MPa)
4.0 Li et al. (2013) (0.3m, 34.26MPa)
Li et al. (2013) (0.4m, 34.26MPa)
3.5 Hanchak et al. (1992) (0.178m, 48MPa)
Hanchak et al. (1992) (0.178m, 140MPa)

Hcr /d
3.0
Hcr=2.5 d
2.5

2.0

1.5

1.0
300 400 500 600 700 800 900 1000 1100
V0 (m/s)

3.5.3 Perforation Limit and Ballistic Limit

Perforation limit hper equals to the sum of penetration depth hpen and rear crater
height Hcf when the critical perforation occurs (Wu et al. 2012a; Chen et al. 2008b;
Li and Tong 2003). Li et al. (2013) and Máca et al. (2014) experimentally obtained
the critical perforation as shown in Fig. 3.18 and indicated that the projectile nose
protruded out of the back face of the target with a distance of about kd/2. By
considering the counterbalance of the entrance crater depth and the above
projectile-protruded distance, summing the DOP without considering the front
crater effect from Eq. (3.46) and the height of the rear crater given by Eq. (3.50),
the dimensionless perforation limit thickness can be derived as

hper 2 I0
¼ þ 2:5 ð3:51Þ
d p ð1 þ ldÞ

The ballistic limit VBL can be deduced reversely by solving Eq. (3.51) as

Fig. 3.18 Photographs of

(a) (b)
rear crater from critical
perforation tests conducted by
a Li et al. (2013) and b Máca
et al. (2014)

Projectile
Projectile
62 3 Rigid Projectile Penetration


pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi v p H
VBL ¼ N1 rs d 3 IBL =M ; IBL ¼ ; v¼
2:5 ð3:52Þ
1
lv=N 2 d

In particular, setting l = 0 in Eq. (3.51) when d < 0.2, Eqs. (3.51) and (3.52)
can be simplified to


hper 2I0 p H
¼ þ 2:5; IBL ¼
2:5 ð3:53Þ
d p 2 d

3.5.4 Residual Velocity

The residual velocity of the projectile Vr is deduced from the conservation of kinetic
energy

MV02 ¼ MVBL
2
þ MVr2 þ q0 XVrf2 ð3:54Þ

where Vrf ¼ gVr denotes the ejecting velocity of the rear plug concrete fragment
illustrated in Fig. 3.16, and X is the volume of the ejected frustum-of-cone frag-
ment, which has the expression of
p
X¼ Hcr ð4 tan2 /Hcr2 þ 6d tan /Hcr þ 3d 2 Þ ð3:55Þ
12

Substituting Eq. (3.50) into Eq. (3.55) and adopting / = 63.5° for normal
strength concrete as reported by Wu et al. (2015a) gives X ¼ 87:5d 3 .
Rearranging Eq. (3.54) gives the residual impact factor Ir and residual velocity
Vr of the projectile perforating thick concrete slabs as

1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Ir ¼ nðI0
IBL Þ; n¼ ; Vr ¼ N1 rs d 3 Ir =M ð3:56Þ
1 þ g2 q0 X=M

3.5.5 Validation

Figure 3.19 illustrates the comparisons between the dimensionless ballistic limit
curves predicted by Eqs. (3.52) and (3.53), as well as the projectile perforation tests
on concrete slabs. The solid and hollow symbols in the figure represent the target
are perforated and unperforated in the tests, respectively. It is demonstrated that the
predicted ballistic limits agree well with the perforation test data for thick concrete
slabs. It should be pointed out that the experimental data of the rear crater height is
employed in Fig. 3.19a since Li’s test data (Li et al. 2013) of the rear crater height
3.5 Perforation Model of Thick Concrete Slabs 63

Fig. 3.19 V/(Sfcd3/M)0.5 (a) 4.0

versus H/d for concrete slabs
3.5

3.0

0.5
V/(Sfc d /M )
2.5

3
2.0

Li et al. (2013)
1.5
Perforation
Critical perforation
1.0 Eq. (3.53)
Eq. (3.52)
0.5
4 6 8 10 12 14
H/d

(b) 7

5
with 1mm rear
0.5

steel liner
V/(Sfc d /M )

4
3

3
Eq. (3.53)
Eq. (3.52)
2
Cargile et al. (1993)
Hanchack et al. (1992)
1
Hanchack et al. (1992)
Wu et al. (2015a)
0
2 4 6 8 10 12 14
H/d

shows disperse. The predictions by Eqs. (3.52) and (3.53) are almost the same, the
reason of which lies in that d < 0.2 is satisfied in Fig. 3.19.
The residual velocities of projectiles after perforating thick concrete slabs pre-
dicted by Eq. (3.56) with the different η values, as well as the test data, are shown in
Fig. 3.20. It indicates that the residual velocities calculated by Eq. (3.56) are
consistent well with the test data for η = 0.2.
64 3 Rigid Projectile Penetration

(a) (b)
900
1200 Hanchak et al. (1992)
Hanchak et al. (1992) (fc=48MPa, H=0.178m) 800 (fc=140MPa, H=0.178m)
1000 Eq. (3.56)
700 Eq. (3.56)

800 600

Vr (m/s)
Vr (m/s)

500
600 400
400 300 Unperforated
Unperforated
200
200
100
0 0
200 400 600 800 1000 1200 300 400 500 600 700 800 900 1000 1100
V0 (m/s) V0 (m/s)
(c) (d)
800 800
Wu et al. (2015a) (fc=41MPa, H=0.15m) Wu et al. (2015a) (fc=41MPa, H=0.2m)
700 700 Eq. (3.56)
Eq. (3.56)
600 600
500 500
Vr (m/s)
Vr (m/s)

400 400
300 300
200 200
100 100
0 0
300 400 500 600 700 800 900 300 400 500 600 700 800 900
V0 (m/s) V0 (m/s)

(e)
700
Cargile et al. (1993) (fc=39.9MPa, H=0.254m)
600
Eq. (3.56)
500
Vr (m/s)

400
300
200
100
0
300 400 500 600 700
V0 (m/s)

Fig. 3.20 Predictions of residual velocity versus initial velocity and test data

3.6 Perforation Model of Thin Concrete Slabs

Most of the existing experimental studies for projectile perforations (Hanchak et al.
1992; Cargile et al. 1993; Unosson and Nilsson 2006; Li et al. 2013; Wu et al.
2015a) are carried out on relatively thick concrete slabs (H/d > 5, H is the slab
thickness and d is the projectile diameter), only few shots are conducted on rela-
tively thin concrete slabs. Moreover, although Chen et al. (2008b) and Wu et al.
(2012a) developed projectile perforation models for thin concrete slabs, the validity
of those models on thin concrete slabs is not clear since the related test data of the
3.6 Perforation Model of Thin Concrete Slabs 65

projectile residual velocity after perforating the thin slabs is scarce. Additionally,
our previously proposed model in Sect. 3.5 (Peng et al. 2015a) is only applicable
for relatively thick concrete slabs. Therefore, the projectile perforation tests on thin
concrete slabs as well as the validated analytical models are both required.
In this section, for the projectile perforation of the relatively thin slab illustrated
in Fig. 2.2b, two groups of projectile perforation tests on thin
ultra-high-performance steel fiber-reinforced concrete (UHP-SFRC) slabs are firstly
conducted, in which the 25.3-mm-diameter projectiles are propelled to perforate the
UHP-SFRC slabs with thicknesses of 40–100 mm. UHP-SFRC is a relatively new
cement-based composite with prominent anti-strike properties; thus, the test is a
supplement to validate its protective performance in the aspect of projectile per-
foration. Secondly, an experimental-based projectile perforation model for thin
concrete slabs (H/d  5) is further established, which can be regarded as a sup-
plement to the model for thick slabs (H/d > 5) presented in Sect. 3.5. Finally, based
on the available experimental data, predicted results by the proposed and existing
formulae are compared and discussed.

3.6.1 Perforation Test

3.6.1.1 Projectile

Shown in Fig. 3.21a, a ogive-nosed projectiles with a diameter of 25.3 mm and a

mass of 331 g is used. The projectile is made of DT300 (SiMnCrNiMoV) steel rods
with the yield strength of 1500 MPa. The dimensions of the projectile is illustrated

(a)

122mm
(b)
18mm
m
5m
25.3mm

18.4mm
16mm

mm
31.5

25mm
3CR

109.9mm
152mm

Fig. 3.21 Projectile a photograph and b dimensions

66 3 Rigid Projectile Penetration

Fig. 3.22 Powder gun

in Fig. 3.21b, where the cartridge thickness is 3.45 mm and the caliber-radius-head
(CRH) of the ogival nose is 3. The projectile is filled by the polymer inert material
with the density of 1.5 ± 0.05 g/cm3 to substitute the filled charge and adjust the
centroid of the projectile. A 25.3-mm-caliber smooth-bore powder gun in CAEP, as
shown in Fig. 3.22, is utilized to launch the projectile to reach the striking velocities
between 250 and 478 m/s by adjusting the charge weight.

3.6.1.2 UHP-SFRC Slabs

The mix proportions of UHP-SFRC in the test is listed in Table 3.2, in which the
components are normalized to the cement weight and the wet density of
UHP-SFRC is 2530 kg/m3. The detailed properties of the mixing ingredient as well
as the preparation procedure can be referred to Peng et al. (2016c). The volumetric
ratio of the mixing fiber is designed as 2% and the straight brass-coated steel fiber is
chosen since it provides a good trade-off between workability and mechanical
properties of concrete. The diameter, length, and tensile strength of the steel fibers
are 0.175, 13 mm, and 3000 MPa, respectively. The size of the basalt aggregate is
controlled strictly to less than 10 mm. HRWR represents the polycarboxylic-type
high-range water reducer.
As listed in Table 3.3, seven thin square UHP-SFRC slabs with dimensions of
400 mm  400 mm and nominal thickness of 40–100 mm (about 1.6–4 times of
the projectile diameter) are cast. The slab label denotes the nominal thickness, and
H is the actual slab thickness. The circumferential boundary effect of the slab during
perforation process could be neglected according to the previous analyses in Peng
et al. (2016b), which is also validated by the experimental observation as shown in

Table 3.2 Mix proportions of UHP-SFRC

Cement Water Fly Silica Ultra-fine Fine Basalt Steel HRWR
ash fume slag sand fibers
1.00 0.26 0.17 0.17 0.33 1.67 1.67 0.33 0.03
3.6 Perforation Model of Thin Concrete Slabs 67

Fig. 3.25. All slabs are cured atmospherically for 160 days before the impact test.
The average 160 days unconfined compressive strength of UHP-SFRC is
128.4 MPa, which is acquired experimentally on cylinder specimens with dimen-
sions of U75  150 mm3.

3.6.1.3 Test Setup

Figure 3.23 illustrates the sketch of the projectile perforation test. The square
UHP-SRFC slab is mounted on a stationary stiff steel frame with its frontal face
perpendicular to the barrel of the powder gun. The configuration of the steel frame
and the constraints of the slab are introduced at length in Wu et al. (2015a). Striking
velocity V0 is measured by a pair of metal wire nets that placed in the straight
trajectory of the projectile, which forms part of the electrical circuits connected to
an oscilloscope. Sequential breakage of the wire net by the projectile will generate
voltage changes; thus, impact velocity could be calculated by dividing the distance
between the two metal wire nets by a time interval between the voltage changes.
The residual velocity of the projectile after perforating the slab Vr is recorded by the
high-speed camera system and listed in Table 3.3.
The impact process of the projectile is recorded by the high-speed camera
system, and Fig. 3.24 shows the typical photographs during projectile perforation,
which indicates that the projectile perforates the target perpendicularly.

3.6.1.4 Test Results

As listed in Table 3.3, five shots with a nominal velocity of 346 m/s into
UHP-SFRC slabs with thicknesses of 40, 50, 60, 70, and 100 mm are conducted.
Additionally, we perform two shots of projectile perforation test into 50-mm-thick
slabs with various striking velocities from 250 to 478 m/s. The dimensions of front
and rear craters including the crater height Hcf (or Hcr), crater volume Vcf (or Vcr)

Table 3.3 Test data

Shot Slab label H (mm) M (g) V0 Vr Front crater Rear crater
no. (m/s) (m/s) Dcf Hcf Vcf Dcr Hcr Vcr /
(cm) (cm) (cm3) (cm) (cm) (cm3) (°)
1-1 Slab 50-1 50.75 330.4 250 171 7.13 2.2 30 10.3 2.8 102 54.2
1-2 Slab 50-2 50.50 331.0 478 425 8.23 2.6 63 10.0 2.4 83 57.3
1-3 Slab 50-3 50.25 331.0 347 283 8.2 2.3 38 10.8 2.7 132 56.8
1-4 Slab 40 41.00 330.8 352 304.9 6.23 1.9 20 10.6 2.1 105 62.5
1-5 Slab 60 60.38 330.6 341 251 7.75 2.6 60 11.3 3.4 152 52.2
1-6 Slab 70 70.25 331.4 348 233 10.7 3.4 107 12.9 3.6 180 55.2
1-7 Slab 100 97.00 330.5 342 156 9.6 3.5 110 21.9 4.9 650 63.2
68 3 Rigid Projectile Penetration

UHP-SFRC
Projectile slab

1000 mm
Wires nets for V0

Bullet-proof
Wall glass

0000
High-speed
camera
Oscilloscope

Fig. 3.23 Sketch of projectile perforation test setup

Slab Slab
Projectile Projectile

Fig. 3.24 Photographs of projectile perforation from high-speed camera

and equivalent diameter Dcf (or Dcr) are also given, where the subscript f and
r denote the front and rear faces of the slab, respectively. Referred to Wu et al.
(2015a), the equivalent diameter Dcf (or Dcr) is obtained by averaging the maximal
dimensions of the crater surface from the four directions: horizontal, vertical, 45°
and 135°, respectively. The slope angle / as listed in the last column of Table 3.3,
which denotes the angle between the moving direction of projectile and the conical
edge of the rear crater (Chen et al. 2008b), is obtained by
/ ¼ tan
1 ½ðDcr
dÞ=2Hcr , where d is the diameter of the projectile. It can be
obtained from Table 3.3 that the average value of / is 57.3°, which is nearly
3.6 Perforation Model of Thin Concrete Slabs 69

(a) (b) (c)

(d) (e) (f)

Fig. 3.25 Damage of the slabs a Slab 40-F, b Slab 70-F, c Slab 100-F, d Slab 40-R, e Slab 70-R,
and f Slab 100-R

identical with / = 55° obtained from the test of the UHP-SFRC slabs against the
small-caliber bullet (Peng et al. 2016a). Besides, the actual mass of each projectile
before impact M is also listed in the table.
Figure 3.25 gives the typical damage of the slabs, in which the labels F and R
represent the front and rear faces of the slab. From the Table 3.3 and Fig. 3.25, the
thin concrete slabs only form the frustum-of-cone-shaped frontal and rear craters
under projectile perforation, which is different from the relatively thick slab.
Meanwhile, Fig. 3.25 also shows that localized response occur in the central region
of the slabs and the dimension of the slabs is sufficient large to neglect the boundary
effect in the test.

3.6.2 Perforation Model of Thin Concrete Slabs

As introduced in Chap. 2, several models were proposed to predict the residual

velocity of projectiles after perforating concrete slabs. However, the validities of
70 3 Rigid Projectile Penetration

those models are only verified by the test data of relatively thick concrete slabs. As
follows, two representative semi-analytical models are firstly presented, and an
experimental-based projectile perforation model for thin concrete slabs (H/d  5)
is further established.

3.6.2.1 Existing Perforation Models

Two representative perforation models are introduced as follows.

(i) Chen’s model
Chen et al. (2008b) proposed the formula to predict the projectile residual
velocity Vr as

Vr ¼ ðV0
VBL Þ; for H=d  vc ð3:57aÞ

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Vr ¼ V02
VBL 2 ; for H=d [ vc ð3:57bÞ

where VBL is the ballistic limit, H is the thickness of concrete slabs, d is the
projectile diameter, and vc is a dimensionless critical thickness of concrete slabs
with the expression as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi
1 þ 3S tan /
1
vc ¼ þk ð3:58aÞ
2 tan /

S ¼ 72ðfc =106 Þ
0:5 or S ¼ 82:6ðfc =106 Þ
0:544 ð3:58bÞ

where k = 0.707 + l0/d is a dimensionless parameter, l0 is the length of the pro-

jectile nose, and fc is the compressive strength of concrete slabs.
Ballistic limit VBL in Eqs. (3.57a, b) is calculated as
qffiffiffiffiffiffiffiffiffi

pd 3 Sfc H ð3:59aÞ
VBL ¼ 4kM d
Hcr ; for H=d  vc
qffiffiffiffiffiffiffiffiffiq ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

pd 3 Sfc ð3:59bÞ
VBL ¼ 2M
H
d
vc þ 2 ;
k
for H=d [ vc

where M is the mass of the projectile and Hcr is the height of the rear crater
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffiffi
2 pffiffiffi pffiffiffi

1 þ 3S=4k þ 3SH tan /=kd
1 þ 3S=4k

Hcr ¼ ; for H=d  vc
2 tan /
ð3:60aÞ
3.6 Perforation Model of Thin Concrete Slabs 71

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffi
1 þ 3S tan /
1 ð3:60bÞ
Hcr ¼ 2 tan / ; for H=d [ vc

It should be noted that, for the slight reinforcement and the steel mesh not hit by
the projectile, the effect of reinforcement on residual velocity can be neglected;
thus, the reinforcement ratio in the original Chen’s formula is omitted at present.
(ii) Grisaro and Dancygier’s model
Grisaro and Dancygier (2014a) developed an energy balance approach for the
assessment of residual velocity, by assuming that the part of striking energy is
dissipated in the fracture of the ejecting crater into concrete fragments as well as
additional cracking of the concrete slab, the formula for the residual velocity was
given as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi




Vr VBL 2 VBL 2a1 VBL 2b1
¼ 1

þ ð3:61Þ
V0 V0 V0 V0

The coefficients a1 and b1 in Eq. (3.61) are empirically obtained as a1 = 0.9 and
b1 = 1.6 by fitting the experimental results reported by Hanchak et al. (1992).
Ballistic limit VBL is determined according to the NDRC model (NDRC 1946).

1:81
dGðHÞ
VBL ¼ 59:53  1000d ð3:62aÞ
KN  M

14:95
K ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffi ð3:62bÞ
fc =106

where N  is the projectile nose geometry factor, which equals to 0.72, 0.84, 1.0, and
1.14 for flat, hemispherical, blunt, and sharp noses, respectively. G(H) is expressed
as
 2
GðHÞ ¼ 0:25
hpen
d ; for hpen =d  2 ð3:63aÞ

GðHÞ ¼
hpen
d
1; for hpen =d [ 2 ð3:63bÞ

where
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
hpen
d ¼ 2:2214
4:9348
1:3928 Hd ; for H=d  3 ð3:64aÞ

hpen
d ¼ 0:806 5 Hd
1:0645; for 3\H=d\18:06 ð3:64bÞ

The prediction of the rear crater depth by Chen’s model is not validated by the
related test data, although the residual velocity predicted by Chen’s model showed a
72 3 Rigid Projectile Penetration

good agreement with the test data from Hanchak et al. (1992). The prediction of
perforation limit by Grisaro and Dancygier’s model is empirically obtained by
curve fitting of the test data.

3.6.2.2 Proposed Perforation Model

In this section, by considering the kinetic energy carried by the rear-ejected frag-
ments, a semi-analytical formula to predict the residual velocity of the projectile
perforating thin slab (H/d  5) is presented, which conforms to the proposed
perforation model for the relatively thick slab (H/d > 5) introduced in Sect. 3.5.
(1) Penetration depth
In Sect. 3.4, based on the mean resistance function, we propose a simple and
effective formula for deep projectile penetration, in which the initial impact cra-
tering stage is neglected. While for the projectile perforation into thin slabs, the
initial cratering phase must be considered. Projectile deceleration-time histories
measured by Forrestal et al. (2003a) showed a rise time at the initial impact stage
and it was assumed that this stage ends when the projectile nose has completely
entered the target (Teland and Sjøl 2004). Thus, according to Peng et al. (2015a)
and Teland and Sjøl (2004), the drag force acted on the projectile nose can be
expressed as
x
Fx ¼ Fm ; for x  l0 ð3:65aÞ
l

pd 2
Fx ¼ Fm ¼ ð1 þ ldÞSfc ; for x [ l0 ð3:65bÞ
4

where Fm is the mean resistance, l is the mean resistance coefficient, and d denotes
the ratio of dynamic to quasi-static resistance. The parameters x and l0 are the
projectile’s instantaneous displacement and nose length, respectively. When
d < 0.2, which are associated with sharp-nosed projectiles or low-velocity blunt
projectile penetrations, l = 0 holds true as discussed in Sect. 3.4. Thus, the work
done by the resistance equals to the initial kinetic energy of projectile

Zhpen
1
Fx dx ¼ MV02 ð3:66Þ
2
0

where hpen denotes the final depth of penetration (DOP) and its expression can be
deduced by substituting Eqs. (3.65a, b) into Eq. (3.66) as
3.6 Perforation Model of Thin Concrete Slabs 73

rffiffiffiffiffiffiffiffiffiffi
hpen 4I0 l0 p l0
¼ ; for hpen  l0 or I0 \ ð3:67aÞ
d p d 4d
hpen 2I0 l0 p l0
¼ þ ; for hpen [ l0 or I0  ð3:67bÞ
d p 2d 4d

where

MV02
I0 ¼ ; S ¼ 82:6ðfc =106 Þ
0:544 ð3:67cÞ
Sfc d 3

Fig. 3.26 Rear crater height (a) 55

from a present test and b test
50 Test data
in Wu et al. (2015a)
45 Chen et al. (2008), Eq. (3.60)
Hc=0.5H
40
35
Hcr (mm)

30
25
20
15
10
5
0
0 10 20 30 40 50 60 70 80 90 100 110
H (mm)

(b) 90
80 Test data (H/d 4) Wu et al. (2015a)
Chen et al. (2008), Eq. (3.60)
70 Hcr=0.5H
Hcr (mm)

60

50

40

30

20

10
400 450 500 550 600 650 700 750
V0 (m/s)
74 3 Rigid Projectile Penetration

(2) Height of rear crater

The dimensions of the rear craters after the projectile perforations are given in
the present and previous tests conducted by Wu et al. (2015a). The height of rear
crater for the 40–100-mm-thick UHP-SFRC slabs and the 100-mm-thick normal
strength concrete slab (41 MPa) is given in Fig. 3.26, and the corresponding pre-
dictions by Chen et al. (2008b) are also given for comparison. The projectiles in the
two groups of tests are both 25.3 mm in diameter. It is shown that Eq. (3.60a)
underestimates the experimental data, while the prediction of Hcr = 0.5H shows
good agreement with the test data. It is noted that Hcr = 0.5H is irrelevant to the
varying striking velocities nearly from 400 to 750 m/s.
In Sect. 3.5, we propose that Hcr = 2.5d for thick concrete slabs based on test
data. Similarly, at present, the height of rear crater for thin concrete slabs is
assumed as

Hcr ¼ 0:5H for H  5d ð3:68Þ

Equation (3.68) is limited in the range of H  5d, which is consistent with the
rear crater height formula for thick slabs when H = 5d. Therefore, we give the
definitions of thin concrete slabs as H  5d and the thick slabs as H > 5d for
convenience.
(3) Perforation limit and ballistic limit
Perforation limit denotes the minimum thickness of the slab required to prevent
the perforation. In Sect. 3.5, we give the perforation limit as hper = hpen + Hcr − l0/
2 for thick slabs, based on the fact that the projectile nose protruded out of the rear
crater with distance of l0/2 approximately when the critical perforation occurs (Li
et al. 2013; Máca et al. 2014; Peng et al. 2016a). When the slab’s thickness
decreases to zero, the protruded distance should be zero but not l0/2, so we give the
assumption that the projectile-protruded distance is linear to the thickness of thin
slabs when H  5d. Therefore, the perforation limit for thin concrete slabs can be
given by considering Eq. (3.68)

hper hpen H l0 H H
¼ þ
 ; for 5 ð3:69Þ
d d 2d 2d 5d d

It should be pointed out that we express the protruded distance as a half of the front
crater height in Sect. 3.5.2, which is different with the expression here but actually
with the same values (kd l0 satisfies for the general ogive-nosed projectile).
The ballistic limit VBL can be deduced reversely by solving Eq. (3.69) and
replaced hpen with Eqs. (3.67a, b, c) as
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
VBL ¼ Sfc d 3 IBL =M ð3:70aÞ

where
3.6 Perforation Model of Thin Concrete Slabs 75


H l0 H 2 pd 10l0
IBL ¼ þ  ; for H  ð3:70bÞ
2d 2d 5d 4l0 5d þ l0


p H l0 l0 H 10l0 H
IBL ¼
þ  ; for \ 5 ð3:70cÞ
2 2d 2d 2d 5d 5d þ l0 d

Figure 3.27 illustrates the schematics of three perforation scenarios corre-

sponding to Eqs. (3.70b, c) and (3.77d), respectively.
(4) Residual velocity
Similar with Eq. (3.54), the residual velocity of the projectile Vr can be deduced
from the conservation of kinetic energy as

MV02 ¼ MVBL
2
þ MVr2 þ q0 XVrf2 ð3:71Þ

where Vrf ¼ gVr denotes the ejecting velocity of the rear plug concrete fragments
and η = 0.2 was obtained in Sect. 3.5.5. X is the volume of the ejected
frustum-of-cone fragments, which has the expression of
p

X¼ Hcr 4 tan2 /Hcr2 þ 6d tan /Hcr þ 3d 2 ð3:72Þ

12

in which Hcr is given by Eq. (3.68) when H  5d. Rearranging Eq. (3.71) gives
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
Vr ¼ Sfc d 3 Ir =M ; Ir ¼ ðI0
IBL Þ ð3:73Þ
1 þ g2 q0 X=M

By substituting the unified Eqs. (3.70a, 3.70b, 3.70c) into Eq. (3.73), the
residual velocity of the projectile can be determined.

3.6.3 Validation

In this section, the existing and proposed models are evaluated by comparing with
the available projectile perforation test data on thin concrete slabs. Figure 3.28
illustrates the comparisons between the experimental projectile residual velocity
and the predictions by the proposed model, Chen’s model and Grisaro and
Dancygier’s model, respectively.
It indicates that the predictions by the proposed model show high degree of
accuracy with the available test data for thin concrete slabs, and the validity of the
proposed perforation model for thin concrete slabs is verified.
While the predictions by Chen’s model and Grisaro and Dancygier’s model
either underestimate or overestimate the residual velocities for thin concrete slabs,
e.g., H/d  3.6 in Fig. 3.28a–c. When the slab becomes thicker, the deviations
76 3 Rigid Projectile Penetration

Fig. 3.27 Schematics of (a) d

perforation scenarios
corresponding to Case 1: Projectile
a Eq. (3.70b) b Eq. (3.70c) hpen l0; H 5d
and c Eq. (3.77d)

Concrete slab

H
H/2
Plug region

d
(b)
Projectile
Case 2:
hpen >l0; H 5d

Concrete slab

H
Plug region H/2

d
(c)
Case 3: Projectile
hpen >l0; H>5d

Concrete slab
H
2.5d

Plug region
3.6 Perforation Model of Thin Concrete Slabs 77

(a) (b)
600 350
Test data
( V0=346m/s)
Proposed model, Eq. (3.73)
500 Chen et al. (2008) model 300
Grisaro and Dancygier (2014) model H=86.3mm
400 250
Vr (m/s)

Vr (m/s)
200
300
150
200
100 Test data
(H=50mm)
100 50
Proposed model, Eq. (3.73)
Chen et al. (2008) model
Grisaro and Dancygier (2014) model
0 0
150 200 250 300 350 400 450 500 550 600 0 20 40 60 80 100 120 140
V0 (m/s) H (mm)
(c) (d)
350 900
( V0=312.9m/s)
Wu et al. (2015a) (H/d=4)
300 800
Proposed model, Eq. (3.73)
700 Chen et al. (2008) model
250 H/d=3.6591
Grisaro and Dancygier (2014) model
Vr (m/s)

600
Vr (m/s)

200 500
150 400
300
100 Cargile et al. (1993)
Proposed formula, Eq. (3.73)
200
50 Chen et al. (2008) model 100
Grisaro and Dancygier (2014) model
0 0
0 1 2 3 4 5 6 100 200 300 400 500 600 700 800 900
H/d V0 (m/s)

Fig. 3.28 Comparisons of experimental and predicted residual velocities

between the test data and the predictions by the above two existing models both
decrease, shown in Fig. 3.28b, c. As for the thickness of the slab approaching to the
thick one, all the three models show good agreements with the experimental results.

3.7 Unified Perforation Model of Thick and Thin Concrete

Slabs

Based on the discussions in Sects. 3.5 and 3.6, a unified projectile perforation
model for both thick and thin concrete slabs can be derived. The perforation limit
hper is expressed as

hper hpen Hcr l0 H H

¼ þ
 ; for 5 ð3:74aÞ
d d d 2d 5d d
hper hpen Hcr l0 H
¼ þ
; for [5 ð3:74bÞ
d d d 2d d
78 3 Rigid Projectile Penetration

where the penetration depth has the form as

rffiffiffiffiffiffiffiffiffiffi
hpen 4I0 l0 p l0
¼ ; for hpen  l0 or I0 \ ð3:75aÞ
d p d 4d
hpen 2I0 l0 p l0
¼ þ ; for hpen [ l0 or I0  ð3:75bÞ
d p 2d 4d

MV02
I0 ¼ ; S ¼ 82:6ðfc =106 Þ
0:544 ð3:75cÞ
Sfc d 3

and the rear crater depth Hcr is

Hcr ¼ 2:5d for H [ 5d ð3:76aÞ

Hcr ¼ 0:5H for H  5d ð3:76bÞ

The ballistic limit VBL can be deduced reversely by solving Eqs. (3.74a, b) as
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
VBL ¼ Sfc d 3 IBL =M ð3:77aÞ

where


H l0 H 2 pd 10l0
IBL ¼ þ  ; for H  ð3:77bÞ
2d 2d 5d 4l0 5d þ l0


p H l0 l0 H 10l0 H
IBL ¼
þ  ; for \ 5 ð3:77cÞ
2 2d 2d 2d 5d 5d þ l0 d


p H H
IBL ¼
2:5 ; for [5 ð3:77dÞ
2 d d

The residual velocity Vr can be further obtained as

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1
Vr ¼ Sfc d 3 Ir =M ; Ir ¼ ðI0
IBL Þ ð3:78Þ
1 þ g2 q 0 X=M

where the volume of ejected frustum-of-cone fragments X can be derived by

Eqs. (3.55) and (3.72) for thick and thin slabs, respectively.
There are two points should be mentioned here: (i) the special scenario that 10l0/
(5d + l0) > 5, or l0 > 5d is not included in Eqs. (3.77a, b, c, d) since such projectile
is very sharp and scarcely used; (ii) for the blunt projectile with high velocity, the
perforation limit hper in Eqs. (3.74a, b) or the ballistic limit VBL in Eqs. (3.77a, b, c,
d) should be revaluated by recalculating the DOP according to Eqs. (3.65a, b, 3.66)
with l = 0.4 as discussed in Sect. 3.4.
3.7 Unified Perforation Model of Thick and Thin Concrete Slabs 79

(a) (b)

Impacting direction

Fig. 3.29 Configurations of two segmented targets a 100 mm + 200 mm and b 100 mm +
100 mm + 100 mm

The above-proposed unified perforation model is used to analyze a series of

perforation tests on spaced segmented slabs conducted by Wu et al. (2015a), in
which a single slab was inadequate to stop the projectile and the subsequent slabs
were all struck and perforated. Two typical segmented targets’ configurations are
shown in Fig. 3.29.
Figure 3.30 plots the residual velocities of the projectile after perforating each
slab of the segmented concrete targets, as well as the corresponding predictions by
Eq. (3.78). In Fig. 3.30, the horizontal axis V0 is the velocity of the projectile when
it hits the first slab, and the predicted residual velocities of the projectiles perforated
the previous slab are regarded as the striking velocity for the subsequent slab. It can
be seen that the predictions by the unified perforation model consist well with the
test data for the segmented slab. It should be noted that a steel plate (1 mm thick) is
welded onto the rear face of the last slab in each configuration, and it is neglected in
the discussion as its effect on the residual velocity is not obvious drawn by Wu et al.
(2015a).

3.8 Deceleration-Time History

Rigid-body deceleration-time history of a projectile could provide a better under-

standing for the penetration/perforation process than the scattered terminal ballistic
parameters such as the final DOP and residual velocity. The deceleration-time
history as well as its single and double integrations (velocity- and
displacement-time history) directly describe the instantaneous endured resistance
and motion of the projectile, which could represent the most basic characteristics of
the impact process. Meanwhile, they are also the fundamentals of the smart fuze
technique which would allow the triggering of the detonation at the expected
location. Besides, the deceleration data is crucial in the strength design of the
projectile as well as its mechanical components and instrumentation package.
80 3 Rigid Projectile Penetration

(a) (b)
1000 1000
st
st
Test data (1 layer) Wu et al. (2015a) Test data (1 layer) Wu et al. (2015a)
nd
nd
Test data (2 layer) Wu et al. (2015a) Test data (2 layer) Wu et al. (2015a)
800 800 st
st
Prediction for 1 layer Prediction for 1 layer
nd
nd
Prediction for 2 layer Prediction for 2 layer
600

Vr (m/s)
Vr (m/s)

600
H=100mm H=200mm
400 400

H=100mm+200mm H=200mm+100mm
200 200

0 0
100 200 300 400 500 600 700 800 900 1000 300 400 500 600 700 800 900 1000
V0 (m/s) V0 (m/s)

(c) (d)
1000 st 1200
Test data (1 layer) Wu et al. (2015a) st
Test data (1 layer) Wu et al. (2015a)
nd nd
Test data (2 layer) Wu et al. (2015a) 1000 Test data (2 layer) Wu et al. (2015a)
800 st
Prediction for 1 layer Test data (3rd layer) Wu et al. (2015a)
nd st
Prediction for 2 layer 800 Prediction for 1 layer
Vr (m/s)

Vr (m/s)

600 nd
Prediction for 2 layer
H=150mm 600 Prediction for 3rd layer
400
400 H=100mm H=100mm+100mm
H=150mm+150mm
200 200
H=100mm+100mm+100mm
0 0
300 400 500 600 700 800 900 1000 100 200 300 400 500 600 700 800 900 1000
V0 (m/s) V0 (m/s)

Fig. 3.30 Predictions of residual velocity versus initial velocity for segmented concrete slabs
a 100 mm + 200 mm, b 200 mm + 100 mm, c 150 mm + 150 mm, and d 100 mm +
100 mm + 100 mm

Therefore, many researchers have made efforts in the measurements and analyses of
this ultra-high g (103–105g) deceleration-time data.
As for the projectile deceleration during penetration, in the early 1990s, Forrestal
and Luk (1992a) reported six sets of deceleration-time data for projectile penetra-
tion into soil targets with the peak rigid-body decelerations about 1200g. An
onboard, digital-recording system which contained two accelerometers in forward
and aft locations of the projectile was employed to measure the deceleration data.
The rigid-body deceleration of the projectile was obtained by filtering raw signals
from the two accelerometers until no difference could be detected. For concrete
targets, Forrestal et al. (2003a) and Frew et al. (2006) conducted series of pene-
tration experiments with deceleration-time measurement and gave analysis to the
deceleration data with a penetration model derived by the cavity expansion theory
(Forrestal et al. 1994). Later, Warren et al. (2014) pointed out that the
deceleration-time data in Forrestal et al. (2003a) contained both rigid-body decel-
eration as well as the structural response, and the higher frequency responses were
not measured since the data recorder was limited to 15 kHz.
3.8 Deceleration-Time History 81

As for the projectile deceleration during perforation, Cargile et al. (1993) used an
onboard deceleration recorder to monitor the movement of the projectile during
perforation into the thin unreinforced concrete slab. Gao et al. (2004) gave the
deceleration-time histories of projectiles perforating the double- and three-layer
spaced reinforced concrete (RC) slabs. Besides, Booker et al. (2009) obtained
deceleration-time test data of the projectile impacting on the eight-layer spaced RC
slabs with different filling materials (air or grout) by mounting two accelerometers
at the nose and tail of the projectiles. In general, the above data only provided
qualitative understandings of the deceleration characteristic during projectile
perforations.
Since the measured deceleration-time data contains not only the rigid-body
deceleration but also the structural response, the accuracies of most measured
deceleration-time data are not verified. Furthermore, for the concerned rigid-body
deceleration, there are not appropriate signal processing approaches to remove the
structural response from the crude deceleration-time signal, although some
researchers have discussed on this issue (Forrestal and Luk 1992a; Wang et al.
2007a; Huang et al. 2009; Fan et al. 2012).
On the other hand, as for the analysis of the projectile impact on the concrete
target, numerical simulations and the resistance-based theoretical models have been
proposed over years to predict the projectile terminal ballistic parameters.
Numerical simulations could provide deceleration-, velocity- and
displacement-time histories for both the projectile penetration and perforation,
while the resistance-based theoretical models could only give the time-resolved
histories for projectile penetration. Although the numerical simulations and the
theoretical predictions are verified by the scattered terminal ballistic data such as
DOP, the accuracy of their prediction for instantaneous motion characteristic of
projectiles is still unclear since the reliable experimental penetration/perforation
deceleration-time data is limited.
In this section, the measured deceleration-time data from the test introduced in
Sect. 12.2 is discussed in detail. Firstly, the rigid projectile penetration and per-
foration test with the measurements of residual velocity and deceleration-time
history is introduced briefly. Then, the accuracy and the signal processing approach
of the deceleration-time data are discussed. After that, the deceleration-time models
for projectile penetration and perforation are given. The experimental
deceleration-time data and the integrated velocity- and displacement-time histories
are compared with the predicted results from theoretical models and numerical
simulations.
82 3 Rigid Projectile Penetration

Fig. 3.31 Projectile and accelerometer a photograph and b dimensions

Location for
Cartridge case data recorder Projectile Smooth-bore gun

149mm 152mm 1857mm

2158mm

Fig. 3.32 Schematic for powder gun with projectile before launch

3.8.1 Projectile Impact Test

3.8.1.1 Projectiles and Concrete Slabs

The ogive-nosed projectiles, as shown in Fig. 3.31a, are made of 45CrNiMoV steel
rods and heat treated to a hardness of HRC45 and the yield strength of 1420 MPa in
the test. Projectile dimension is illustrated in Fig. 3.31b, where the shank diameter
is 25.3 mm and the caliber-radius-head (CRH) of the ogival nose is 3. Before the
shot, an onboard deceleration data recorder (shown in Fig. 3.31a) is firstly fitted
into a circular hole in the projectile as shown in Fig. 3.31b and then tightened by a
screw cap at the tail of the projectile; the total nominal mass of the instrumented
projectile is 428 g. The deceleration-time data recorder is mainly composed of the
piezoresistive accelerometer, recording electronics, and battery.
The deceleration-time data recorder is designed to digitize and record the
acceleration during the launch in the barrel as well as the deceleration during the
penetration or perforation into targets. The sampling amplitude and frequency of
accelerometer are 2.5  105g and 1.05  105 Hz, respectively. After impact test,
the deceleration data measured by the piezoresistive accelerometer mounted in the
data recorder is retrieved from the recovered recorder.
As illustrated in Fig. 3.32, a 25.3-mm-caliber smooth-bore powder gun is uti-
lized to launch the projectiles to achieve the designed striking velocities from 540 to
641.5 m/s. The total barrel length of the powder gun and the distance between the
projectile nose and the powder gun muzzle are 2.158 and 1.857 m, respectively.
RC slabs with the dimensions of 675 mm  675 mm and thicknesses of 300,
200, and 100 mm are cast, respectively. Several layers of square-pattern steel mesh
3.8 Deceleration-Time History 83

are utilized to reinforce the RC slabs, the geometry, and the layout of which can be
referred from Wu et al. (2015a). The unconfined cylindrical compressive strength
and wet density of concrete are 41 MPa and 2350 kg/m3, respectively.

3.8.1.2 Test Setup and Results

Projectile impact test on the monolithic or two spaced RC slabs with the total
thickness of 300 mm is conducted, and the experimental setup is illustrated in
Fig. 3.33. The RC slabs are located 3.66–3.72 m (S1) from the powder gun muzzle
and the distance between the two slabs for the spaced segmented targets is about
27 cm (S2). A steel plate with 1 mm thick is welded on the rear face of the last RC
slab in each shot. The slabs are fixed with their frontal face perpendicular to the gun
and aligned to ensure the projectiles do not strike the steel reinforcement. In
Fig. 3.33, the symbol V0 represents the projectile striking velocity from the electric
probes, while V1r and V2r denote the residual velocities of the projectile after
perforating the first and second RC slabs (measured from high-speed camera),
respectively.
Figure 3.34 gives the recorded raw deceleration-time histories (only zero shif-
ted) in Shots 1-1, 1-2, 3-2, and 1-3, respectively. In which, the projectile penetrates
into the 300-mm-thick RC slab with DOP of 254 mm in Shot 1-2, and projectiles
perforate the 300-mm-thick RC slabs in Shots 1-1 and 1-3. While in Shot 3-2, the
projectile perforates the two spaced RC slabs with the thickness of 200 mm (1st
slab) and 100 mm (2nd slab). In Fig. 3.34, the positive value represents the
deceleration while the negative one means the acceleration. Obviously, the whole
ballistic range of deceleration-time history is obtained, including the projectile
accelerating in the barrel, flight in the air, decelerating in the concrete target, and
after-perforation flight or penetration stop.

2.158m S1 H1 S2 H2

V1r V2r
V0

Projectile
Sandbags
Powered gun

RC slab
Steel linear
Steel frame
Bullet-proof glass
Thickened brick wall High-speed
camera

Fig. 3.33 Test setup

84 3 Rigid Projectile Penetration

(a) (b)
15.0 1-2 20
1-1
12.5
15

Deceleration (10 g)
Deceleration (10 4g)

10.0

4
7.5 10
5.0 5
2.5
0
0.0
-2.5 -5
Stage 1 Stage 2 Stage 4
Stage 1 Stage 2
-5.0 Stage 3 -10 Stage 3

-7.5
0 5 10 15 20 25 30 35 40 0 5 10 15 20 25 30
Time (ms) Time (ms)

(c) (d)
20
20 1-3
3-2
15
15
Deceleration (104g)
Deceleration (104g)

10 10

5 5

0 0
Stage 1 Stage 2 Stage 4
-5 Stage 1 Stage 2 Stage 4
-5
Stage 3
Stage 3
-10 -10
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 0 5 10 15 20 25 30 35
Time (ms) Time (ms)

Fig. 3.34 Deceleration-time histories of projectiles in Shot a 1-2, b 1-1, c 3-2, and d 1-3

The related target configuration and ballistic parameters of the above experi-
ments are summarized in Table 3.4. For the target configuration given in the 3rd
column, “300(1)” represents the monolithic 300-mm-thick slab with a 1-mm steel
liner welded on the rear face, and “200 + 100(1)” refers the two spaced slabs with
the thicknesses of 200-mm and 100-mm as well as a 1-mm rear steel liner.
Figure 3.35 presents the posttest photographs of the slab in Shot 1-2, and it is
shown from Fig. 3.35b that the rear concrete crater is not formed although the
distance between the projectile nose and the rear face of the slab is only 46 mm
since the existing of the rear steel liner. The recovered projectiles show only minor
nose abrasions and the mass losses are very small. The more details such as the
damage of the perforated slab can be found in Wu et al. (2015a).
3.8 Deceleration-Time History 85

Table 3.4 Test data

Shot S1 Configuration S2 M (g) V0 V1r V2r
no. (m) (m) (m/s) (m/s) (m/s)
1-2 3.71 300(1) – 428.3 540 DOP = 254 mm
1-1 3.72 300(1) – 429 641.5 272 –
3-2 3.66 200 + 100(1) 0.27 429.9 597 404 270
1-3 3.71 300(1) – 427.1 601 180 –

(a) (b)

Fig. 3.35 Posttest photographs for the slab in Shot 1-2 a front face and b rear face with the steel
liner removed

3.8.2 Accuracy Analysis of Measured Deceleration-Time

Histories

When the recorded deceleration data is retrieved, it is important to verify its

accuracy. The time-resolved deceleration data given in Fig. 3.34 shows large
amplitude structural vibrations superimposed on the rigid-body responses.
However, few researchers discussed the correctness of the deceleration-time data
measured from projectile impact experiments.
Forrestal et al. (2003a) and Frew et al. (2006) performed integrations of the
deceleration-time data to obtain striking velocity and DOP into concrete targets.
And the comparisons were made between the integrated results and the independent
measurements of striking velocity and DOP in order to validate the measured
deceleration. One should be noted that the integrations of raw deceleration signals
also contain the contributions of structural response. The reason why they could
discuss the accuracy of the deceleration data is based on the premise that the
rigid-body deceleration makes a major contribution to the overall deceleration since
the projectiles lost small amounts of mass and experiences relatively small defor-
mations. Actually, the higher frequency responses were not measured since the
deceleration data was limited to 15 kHz, which may be the main reason why the
rigid-body deceleration dominates the raw signal.
Li et al. (2013) and Wu et al. (2015a) compared the striking velocities measured
by electric probes with an integration of their crude deceleration-time data and
made the comparison between the projectile displacement and the displacement by
86 3 Rigid Projectile Penetration

the double integration of the raw signal. However, these discussions on the accu-
racy of deceleration data neglect the influences of the structural response without
any analysis or clarification. Besides, it is less confidence to verify the accuracy of
deceleration test data by comparing integrated results with terminal ballistic
parameters predicted by numerical simulation or formula.

3.8.2.1 Proposed Analysis Approach

In this section, for the validation of the deceleration data taken with the onboard
recorder, an approach is proposed based on the spectral analysis and
deceleration-time data integrations.
Frequency characteristic analyses of the deceleration-time data indicated that the
higher frequency oscillations are associated with structural vibration while the
lower frequency (about 0–2 kHz) responses represent the rigid-body component
(Fan et al. 2012). Taking the Shot 1-2, for example, Fig. 3.36 presents the fre-
quency spectrum for the whole-range deceleration-time data with Fourier transform,
it can be seen that the high-amplitude responses associated with rigid-body
deceleration fall in the frequency domain of about 0–1.4 kHz.
Now let us discuss the contribution of the high-frequency component to the
integration of the raw deceleration-time data. Figure 3.37a shows the 1.4-kHz
low-pass-filtered deceleration-time data of Shot 1-2, and Fig. 3.37b presents the
filtered signal with 1.4 kHz high-pass filter, the sum of these two filtered signals
gives the raw signal of Shot 1-2. Figure 3.38 illustrates the velocity-time histories
from an integration of the above deceleration data for the whole range and pene-
tration stage, respectively. It indicates that the integration of the high-frequency
response almost keep zero except some low-amplitude oscillations during the
penetration into the concrete target, while the integrations of the raw signal and 0–

Fig. 3.36 Frequency 0.6

spectrum for the deceleration
data of Shot 1-2 0.5
Amplitude (10 4g)

0.4

0.3

1.4kHz
0.2

0.1

0.0
0 10 20 30 40 50 60
Frequency (kHz)
3.8 Deceleration-Time History 87

Fig. 3.37 Filtered (a) 15.0

deceleration-time histories of
Shot 1-2 with a 1.4-kHz 12.5
low-pass filter and b 1.4-kHz 10.0
Raw signal
high-pass filter Filterted signal

Deceleration (104g)
7.5 (1.4kHz low-pass)

5.0

2.5

0.0

-2.5

-5.0
0 5 10 15 20 25 30 35
Time (ms)

(b) 8

6
Filterted signal
4 (1.4kHz high-pass)
Deceleration (104g)

-2

-4

-6

-8
0 5 10 15 20 25 30 35
Time (ms)

1.4-kHz filtered deceleration-time data are almost identical. One can be concluded
that the integrations of the rigid-body deceleration make a major contribution to the
velocity and displacement, while the structural vibration has negligible influence.
Thus, integrating the raw deceleration-time data from the onboard accelerometer
and comparing these integrated results with both the independent measured pro-
jectile velocity and displacement is a convincing approach to validate the
deceleration-time test data.

3.8.2.2 Validation of the Proposed Analysis Approach

For the penetration tests, Fig. 3.39 shows single and double integrations of the
whole-range deceleration-time history for Shot 1-2 as well as the measured velocity
and displacement. According to the comparisons between the deceleration-time
history in Fig. 3.34a, velocity and displacement versus time history in Fig. 3.39, as
88 3 Rigid Projectile Penetration

Fig. 3.38 Velocity-time (a) 100

histories from an integration Intergrated from 1.4kHz high-pass signal
of deceleration data for 0
a whole range and
b penetration stage -100

Velocity (m/s)
An intergration from
-200
raw signal
1.4kHz low-pass
-300 1.4kHz high-pass

-400

-500

-600
0 5 10 15 20 25 30 35
Time (s)

(b) 100
Intergrated from 1.4kHz high-pass signal
0

-100
Velocity (m/s)

-200

An intergration from
-300
raw signal
1.4kHz low-pass
-400 1.4kHz high-pass

-500

-600
25.0 25.2 25.4 25.6 25.8 26.0 26.2 26.4 26.6 26.8
Time (s)

well as the fixed propelling distance of the projectile in the barrel

(1.857 + 0.152 m) shown in Fig. 3.32, the following three stages can be identified
during the whole journey of the projectile. They are the Stage 1: interior ballistics
(accelerating in the barrel) during 13.09–18.52 ms; Stage 2: exterior ballistics
(flight in the air from the muzzle to the target face) during 18.52–25.20 ms; Stage 3:
terminal ballistics (decelerating in the target) during 25.20–26.50 ms and then the
projectile stopped.
Figure 3.39a indicates that the projectile velocity from an integration of the
deceleration signal during the Stage 2 has nearly no changes, approximately from
526.1 m/s at 18.52 ms to 519.9 m/s at 25.20 ms, which coincides well with the
striking velocities 540 m/s measured by the electric probes as listed in Table 3.4.
As shown in Fig. 3.39b, the double integration of the deceleration data shows that
the displacement of the projectile is 5.47 m when it striking on the front face of the
target, which is 1.74% less than the actual distance 5.567 m (1.857 + 3.71 m).
3.8 Deceleration-Time History 89

Fig. 3.39 Validation for the (a) 100

deceleration-time data of Shot Stage 3
Stage 1 Stage 2 Stopped
1-2 by a velocity and 0
b displacement -0.52m/s
-100

Velocity (m/s)
An intergration
-200 of signal
Test data
-300

-400
-519.9m/s
-500

-526.1m/s -540m/s
-600
0 5 10 15 20 25 30 35
Time (ms)

(b) 1 Stage 3
Stage 1 Stage 2
0

-1 (13.09,0) (18.52, -1.98)

Displacement (m)

-2

-3 Double intergration
of signal
-4 Test data
(26.50, -5.712)
-5 -5.567m

-6
(25.20, -5.470) -5.821m
-7
0 5 10 15 20 25 30 35
Time (ms)

Besides, at the Stage 3, DOP obtained from double integration of the

deceleration-time data is 0.242 m, which is 4.72% less than the actual 0.254 m. So
far, the accuracy of the measured deceleration-time data for Shot 1-2 is validated.
For the perforation tests, Figs. 3.40 and 3.41 present the whole-range projectile
velocity and displacement versus time history from the integrations of the decel-
eration data for Shots 1-1 and 3-2. Obviously, four stages during the whole journey
of the projectile can be identified, the first three stages are the same with Shot 1-2
and the Stage 4 is the residual-free flight stage. The specific time of each stage can
be found in Figs. 3.40b and 3.41b. It is indicated that the results by the single or
double integrations show good agreements with the measured striking velocities,
residual velocities, and displacement listed in Table 3.4. Thus, the deceleration-
time data of Shots 1-1 and 3-2 is also validated.
90 3 Rigid Projectile Penetration

Fig. 3.40 Validation for the (a) 100

deceleration-time data of Shot Stage 3
Stage 1 Stage 2 Stage 4
1-1 by a velocity and 0
b displacement -100
278.80m/s

Velocity (m/s)
-200
An intergration
-300 of signal
Test data -272m/s 256.5m/s
-400

-500
-641.5m/s -650.6m/s
-600

-700 -650.1m/s
0 5 10 15 20 25 30
Time (ms)

(b) 1
Stage 3
Stage 1 Stage 2 Stage 4
0

-1
(9.54,0)
Displacement (m)

-2

-3 (14.38,2.03)

-4 Double intergration
of signal (20.50,5.87)
-5 Test data -5.577m

-6
(19.82,5.57)
-7
-5.877m
-8
0 5 10 15 20 25 30
Time (ms)

3.8.2.3 Correction of Less Accurate Measured Data

If the integrations of the whole deceleration-time data are inconsistent with the
independent measured striking velocity, residual velocity and displacement as
discussed in Sect. 3.8.2.2, it may be indicated that the measured data is less
accurate or invalid. For example, the deceleration signal in Shot 1-3 is the data of
this kind. Since the projectile deceleration in the concrete target is mainly con-
cerned, the following discussion focuses on the period of 22.68–25.21 ms (perfo-
ration and after flight stage for Shot 1-3). Figure 3.42a gives the deceleration-time
data of Shot 1-3. It is indicated that the 300-mm concrete slab is perforated by the
projectile with the initial velocity of 601 m/s. The residual velocity of 180 m/s is
listed in Table 3.4. Figure 3.42b illustrates an integration of the deceleration signal.
The change of the projectile velocity before and after perforation is only 321 m/s,
while the actual decrease of the velocity is 421 m/s. This inconsistency indicates
that the data is less accurate.
3.8 Deceleration-Time History 91

Fig. 3.41 Validation for the (a) 100 Stage 3

deceleration-time data of Shot Stage 1 Stage 2
0 An intergration
3-2 by a velocity and of signal
b displacement -100 Test data

-200 -267.1m/s

Velocity (m/s)
-300 -404m/s
-270m/s -274m/s
-400

-500 -585.6m/s

-600
-597m/s Stage 4
-600.5m/s
-700
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0 22.5
Time (ms)

(b) 1
Stage 1 Double intergration
0 of signal
Stage 2
Test data
-1 (6.21, -2.05)
(0.988, 0)
-2 Stage 3
Displacement (m)

-3
Stage 4
-5.517m
-4

-5 (13.42, -6.045)

-6 (12.03, -5.474)

-7
-6.087m
-8
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
Time (ms)

For the uncertainties in the measurements as well as the high cost of the onboard
deceleration recorder, this defective data should be corrected. As stated in Warren
et al. (2014), there are no available calibration procedures to determine the
accelerometer transducer sensitivity with an internationally accepted standard.
Currently, the accuracy of measurements with ultra-high g accelerometer is
examined by the performance evaluation tests with Hopkinson pressure bars
(Forrestal et al. 2003b, Foster et al. 2012), in which the measured accelerations are
compared with other transducers such as strain gauges and quartz stress gauges. In
this section, based on the methodology proposed by Warren et al. (2014), a method
is proposed to correct the defective data so that an integration of the measured
deceleration data would agree with the independent measured striking velocity V0
and residual velocity Vr. If the double integration of the corrected deceleration data
consists well with the actual projectile displacement, the revised deceleration data is
regarded as correct.
From the recorded deceleration data, the decrease of the projectile velocity
during perforation can be expressed as
92 3 Rigid Projectile Penetration

Fig. 3.42 Time histories of (a) 15

projectile in Shot 1-3 1-3
a deceleration and b velocity Raw deceleration data
10 at the interval of 22.68-25.21 ms

Deceleration (104g)
5

-5

22.5 23.0 23.5 24.0 24.5 25.0

Time (ms)

(b) -100
An integration from signal at -180m/s
the interval of 22.68-25.21 ms
-200
Residual velocity

-300
Velocity (m/s)

421 m/s
321 m/s
-400

-500

-600
-601m/s
-700
22.6 22.8 23.0 23.2 23.4 23.6 23.8 24.0 24.2 24.4
Time (ms)

Ztf
DV ¼ am ðtÞdt ð3:79Þ
t0

where t0 is the time when the projectile striking on the front face of the target, tf is
the time at which the projectile perforates the target, and am(t) is measured
deceleration-time data. The deceleration data is corrected as


V0
Vr
aðtÞ ¼ am ðtÞ ð3:80Þ
DV

where V0 is the striking velocity from electric probes and Vr is the residual velocity
captured by high-speed photographs. Thus, the velocity-time history from decel-
eration data can be described as
3.8 Deceleration-Time History 93

Zt
VðtÞ ¼ V0
aðtÞdt ð3:81Þ
t0

Such that V(t0) equals to the striking velocity V0, and V(tf) is consistent with the
residual velocity Vr. In particular, when Vr = 0, the above correction approach is
identical with the method proposed by Warren et al. (2014).

(a) 20 20
Before After

15 15
Orignal Corrected
signal signal
Deceleration (104g)

10 10

5 5

0 0

-5 -5

-10 -10
22.5 23.0 23.5 24.0 24.5 25.0 22.5 23.0 23.5 24.0 24.5 25.0
Time (ms)

(b) -100 0.1

-180m/s
(23.57, -188.0m/s)
-200 0.0
0m

-300 -0.1
Displacement (m)
Velocity (m/s)

Velocity
Displacement
-400 Residual velocity -0.2

-500 -0.3

(23.57, -0.3m)
-600 -0.4
-610m/s
-700 -0.5
22.6 22.8 23.0 23.2 23.4 23.6 23.8 24.0 24.2 24.4
Time (ms)

Fig. 3.43 Correction and validation for Shot 1-3 a deceleration correction and b velocity and
displacement
94 3 Rigid Projectile Penetration

Figure 3.43a shows the correction of the deceleration for Shot 1-3. The cor-
rection factor is 421/321 from Fig. 3.42b, which means the revised signal (red line
at right) is 421/321 times of the original data (black line at left). Figure 3.43b gives
the integrations of the corrected deceleration-time history and they are taken neg-
ative to be consistent with the figures in Sect. 3.8.2.1. When t = 23.57 ms, the
projectile nose arrives at the rear face of the concrete slab, and the velocity after this
time nearly keeps unchanged. Thus, the above correction of the deceleration data is
verified to be convincing.

3.8.3 Derivation of Rigid-Body Deceleration-Time Histories

The rigid-body deceleration of the projectile during penetration or perforation into

targets could provide experimental basis for the theoretical models and numerical
simulations; thus, the high-frequency responses associated with the structural
vibration of the projectile should be removed.
How to determine a critical cutoff frequency to ensure that the structural response
is filtered and the rigid-body deceleration is not damaged? Forrestal and Luk
(1992a) mounted two accelerometers into the projectile to record the deceleration
and processed the measured signals with low-pass filter from high to low fre-
quencies. When filtered to 1.0 kHz, no difference could be detected in the filtered
results from the two accelerometers. Thus, the signal filtered by 1.0-kHz low-pass
filter was the rigid-body deceleration-time history.
For the projectile which contains only one accelerometer, the above processing
approach cannot be applied. For single recorded deceleration, Wang et al. (2007a)
filtered the deceleration data from high to low frequencies. When the cutoff fre-
quencies are high, the striking velocity and DOP integrated from the filtered
deceleration data with different cutoff frequencies have no obvious changes. When
the signal is filtered to a lower frequency, the integrated data suddenly changes
obviously, this frequency was treated as the critical cutoff frequency and the cor-
responding filtered signal was regarded as the rigid-body deceleration. Besides,
Huang et al. (2009) and Fan et al. (2012) proposed a signal processing approach by
combining the frequency analysis and numerical modal analysis. The critical cutoff
frequency was determined by comparing the first-order frequency from numerical
modal analysis and the peak frequency of the first frequency spectrum.
However, as discussed in Sect. 3.8.2.1, the higher frequency has almost no
influence on the integrations, the signal processing approach proposed by Wang
et al. (2007a) which used filter from high to low frequencies and conducted cor-
responding integrations may be tedious. While for the approaches proposed by
Huang et al. (2009) and Fan et al. (2012), the first-order frequency from the
numerical modal analysis is always inconsistent with the actual one from the fre-
quency analysis of the recorded deceleration data.
In this section, based on the works of Fan et al. (2012) and Wang et al. (2007a) as well
as the discussions in Sect. 3.8.2.1, a simple signal processing approach is proposed.
3.8 Deceleration-Time History 95

(a)
20 16
14
Raw test data (1-1) Raw test data (1-2)
15 12
Deceleration (10 g)

Deceleration (10 g)
4

10

4
10 8
6
5 4
2
0
0
-2
-5
19.8 20.0 20.2 20.4 20.6 20.8 21.0 21.2 25.2 25.4 25.6 25.8 26.0 26.2 26.4 26.6 26.8
Time (ms) Time (ms)
(b)
0.9 1-1
1.2
1-2
0.8
1.0
0.7
Amplitude (10 g)

Amplitude (10 g)
4

4
0.6 0.8
0.5
0.6 1439Hz
0.4
2570Hz
1747Hz
0.3 0.4 4831Hz
0.2 3289Hz
0.2
0.1
0.0 0.0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
Frequency (kHz) Frequency (kHz)

(c)
10 1-1
10 1-2

8 3.29kHz low-pass filter 8 4.83kHz low-pass filter

Deceleration (10 g)

Deceleration (10 g)

1.75kHz low-pass filter 2.57kHz low-pass filter

4

1.44kHz low-pass filter

6 6

4 4

2 2

0 0

-2 -2
19.8 20.0 20.2 20.4 20.6 20.8 21.0 21.2 25.2 25.4 25.6 25.8 26.0 26.2 26.4 26.6 26.8 27.0
Time (ms) Time (ms)

Fig. 3.44 Signal processing a raw deceleration data, b frequency spectrums, and c filtered signals

Figure 3.44 illustrates this approach by taking the data in Shots 1-1 and 1-2 for
examples. Figure 3.44a gives the projectile deceleration-time histories of Shots 1-1 and
1-2 after impacting on the target, respectively. Figure 3.44b illustrates the frequency
spectrums of the deceleration signal, and the sampling starts at the initial instant in
Fig. 3.44a. To keep a relatively higher frequency resolution, sampling number is
adopted as 1024 and the corresponding sampling duration time is 9.7185 ms.
It is shown from Fig. 3.44b that the highest-amplitude responses correspond to
very low frequencies and the amplitudes decreases linearly with large slope to the
96 3 Rigid Projectile Penetration

Fig. 3.45 Projectile 600

velocity-time history for Shot
1-2 500

400

Velocity (m/s)
300
An integration of signal 1-2
raw signal
200 2.57kHz low-pass filter
1.44kHz low-pass filter
0.8kHz low-pass filter
100

0
25.2 25.4 25.6 25.8 26.0 26.2 26.4 26.6 26.8 27.0
Time (ms)

first valley. As presented in Huang et al. (2009) and Fan et al. (2012), the first-order
frequency of the structural response of the projectile is associated with a crest
whose frequency is often higher than the first valley’s frequency. Considering that
the amplitudes from the first valley to the crest starting point are much lower and
have relatively slight influences on the filtered signal, we take the frequency of the
first valley as the critical cutoff frequency. Figure 3.44c shows the filtered signals of
both two shots. It indicates that the filters with the first valley frequency (1747 Hz
for Shot 1-1 and 1439 Hz for Shot 1-2) remove almost all the structural vibrations,
while the filtered signals with higher cutoff frequencies than the first valley fre-
quency still remain vibrational.
Meanwhile, it should be ensured that the rigid-body deceleration component is
not damaged. If the integration of the filtered signal has not obviously decreased
comparing with the integration of the raw signal, the above processing approach is
regarded as effective. Figure 3.45 further presents an integration of the raw and
processed signals for Shot 1-2 from Fig. 3.44a. It shows that the two integrations
from the raw deceleration data and the 1.44-kHz filtered signal have negligible
difference. In Fig. 3.45, the integration of the 0.8-kHz filtered deceleration data is
also plotted, the difference with the integration of the raw signal becomes larger, the
corresponding peak value of the deceleration is 6.06  104g, which is much lower
than the 7.33  104g for the 1.44-kHz filtered signal.
Until now, the rigid-body deceleration-time history of the projectile is reliably
obtained. The processing approach can be summarized as the following three steps:
(i) performing frequency spectrum analysis; (ii) low-pass filtering the
deceleration-time data by using the first valley frequency in the spectrum as the
cutoff frequency; (iii) integrating the filtered signal and comparing with the raw
signal integration, the filtered signal is reliable if the comparison shows negligible
difference.
3.8 Deceleration-Time History 97

3.8.4 Predictions of Deceleration-Time Histories

In this section, both theoretical analyses and numerical simulations for the
projectile-time histories during penetration and perforation into concrete targets are
performed. The deceleration-time history model as well as the velocity- and
displacement-time expressions for projectile penetrations are firstly given based on
the works of Forrestal et al. (1994, 2003a). Then, a time-resolved model to predict
the deceleration of the projectile perforating the finite thickness concrete slab is
established. For the numerical simulation, the well-known Holmquist–Johnson–
Cook (HJC) model (Holmquist et al. 1993) is utilized to describe the concrete
material. Finally, the reliable deceleration-time data from the previous sections is
used to evaluate the validations of the theoretical and numerical models in pre-
dicting the instantaneous projectile motions during penetration and perforation.

3.8.4.1 Theoretical Model

(1) Deceleration-time model for projectile penetration

According to Forrestal et al. (1994, 2003a), the drag force acting on the pro-
jectile nose can be taken as

Fx ¼ cx; for 0\x  kd ð3:82aÞ

pd 2
Fx ¼ ðR þ Nq0 V 2 Þ; for kd\x\hpen ð3:82bÞ
4

where c is a constant and x is the projectile displacement measured from the

concrete target surface. Parameters d and V are the shank diameter and the
instantaneous velocity of projectile, respectively. R is the target resistance param-
eter and q0 is the density of the target. For ogive-nosed projectile, N ¼

ð8w
1Þ 24w2 is the projectile nose shape factor, w ¼ s=d and s are the
caliber-radius-head (CRH) and circle radius of the projectile nose, respectively.
Besides, kd is denoted as the front crater depth and k = 2.
From Newton’s second law, Eqs. (3.82a, b) can be written as

d2 x
M ¼
cx; for 0\x  kd ð3:83aÞ
dt2

dV pd 2
M ¼
ðR þ Nq0 V 2 Þ; for kd\x  hpen ð3:83bÞ
dt 4

With the initial conditions x(t = 0) = 0 and V(t = 0) = V0, Eq. (3.83a) has the
following solutions for projectile instantaneous displacement, velocity, and
acceleration
98 3 Rigid Projectile Penetration


V0
xðtÞ ¼ sin xt for 0\x  kd ð3:84aÞ
x

dxðtÞ
VðtÞ ¼ ¼ V0 cos xt for 0\x  kd ð3:84bÞ
dt
aðtÞ ¼
xV0 sin xt for 0\x  kd ð3:84cÞ

x2 ¼ c=M ð3:84dÞ

Denoting t1 and V1 as the time and velocity of projectile at x = kd, it has

x(t = t1) = kd and V(t = t1) = V1; thus, Eq. (3.83b) has the following solutions

0:5 ( "
0:5 # )
R
1 Nq0 pd 2
VðtÞ ¼ tan tan V1
ðRNq0 Þ0:5 ðt
t1 Þ
Nq0 R 4M ð3:85aÞ
for kd\x  hpen

pd 2 R
aðtÞ ¼
n h i o
4M cos2 tan
1 ðNq0 =RÞ0:5 V1
pd 0:5
4M ðRNq0 Þ ðt
t1 Þ
ð3:85bÞ
2

for kd\x  hpen

2 n h io 3

1 0:5

pd
2 0:5
4M cos tan ðNq 0 =RÞ V1 4M ðRNq0 Þ ðt
t1 Þ
xðtÞ ¼ 2 ln4 n h io 5 þ kd
pd Nq0 cos tan
1 ðNq0 =RÞ0:5 V1
for kd\x  hpen
ð3:85cÞ

According to the continuity conditions of the force, velocity, and displacement at

x = kd, V1 and c can be obtained as

4MV02
kpRd 3 pdMðR þ Nq0 V02 Þ

V12 ¼ ; c¼ ð3:86Þ
4M þ kpNq0 d 3 kð4M þ kpNq0 d 3 Þ

The target resistance parameter R can be calculated by Forrestal et al. (2003a)

Nq0 V02
R¼  h 2 i ð3:87Þ
1þ pkd 3 Nq
4M
0
exp pd ðPe2M

kdÞNq0

1

where Pe is the experimental DOP. If there is no available DOP test data, R = Sfc
can be adopted for concrete targets, where fc (unit in Pa) is the compressive strength
of the concrete and S = 82.6(fc/106)−0.544 (Frew et al. 1998).
(2) Deceleration-time model for projectile perforation
3.8 Deceleration-Time History 99

Stage 1: initial cratering stage

Stage 2: stable penetration stage
A Stage 3: decrease stage

B
B1
Deceleration

Stage 1 Stage 2 Stage 3

C C1
x
O kd P X2 H
Displacement

Fig. 3.46 Schematic of the projectile deceleration versus displacement during perforation

Although there are some perforation models (Chen et al. 2008b; Wu et al. 2012a;
Peng et al. 2015a) to predict the projectile’s residual velocity, they could not offer
the deceleration-time equation, as the resistance is only utilized to calculate the
ballistic limit and the residual velocity is finally obtained from kinetic energy
balance in these models.
As follows, a deceleration-time history model for projectile perforating the finite
thick concrete slab is proposed. When the projectile reaches a certain distance from
the rear surface, the deceleration is linearly decreasing with the distance away from
the rear surface until the projectile’s nose emerges out of the concrete slab
(Rosenberg and Kositski 2016). This distance is defined as H2, and the corre-
sponding projectile displacement is X2. The deceleration-time expression before the
deceleration decreasing stage is identical with the above penetration model.
Figure 3.46 illustrates the projectile deceleration–displacement relationship during
perforation as O-A-B1-C1, while the curve O-A-B-C describes the deceleration
during penetration.
The motion equation of a projectile during the deceleration linearly decreasing
stage is developed and the resistance acting on the projectile takes the form as

d2 x
F ¼
M ¼ c2 ðH
xÞ; for X2 \x  H ð3:88Þ
dt2

where c2 is a constant and H is the thickness of the concrete slab. With the interface
conditions x(t = t2) = X2 and V(t = t2) = V2, where t2 and V2 are defined as the time
and projectile velocity at x = X2, Eq. (3.88) has the following solutions
100 3 Rigid Projectile Penetration

xðtÞ ¼ A exp½x2 ðt
t2 Þ þ B exp½
x2 ðt
t2 Þ þ H for X2 \x  H ð3:89aÞ

VðtÞ ¼ Ax2 exp½x2 ðt
t2 Þ
Bx2 exp½
x2 ðt
t2 Þ for X2 \x  H ð3:89bÞ

aðtÞ ¼ Ax22 exp½x2 ðt
t2 Þ þ Bx22 exp½
x2 ðt
t2 Þ for X2 \x  H ð3:89cÞ

X2 H V2 X2 H V2
A¼
þ ; B¼

ð3:89dÞ
2 2 2x2 2 2 2x2

x22 ¼ c2 =M; c2 ¼ Ma2 =ðH
X2 Þ ð3:89eÞ

Based on the continuity conditions of the deceleration, velocity, and displace-

ment at x = X2, the time t2 can be solved from Eq. (3.85c), then V2 and a2 can be
obtained from Eq. (3.85a) through V(t2) and Eq. (3.85b) with a(t2), respectively.
R = Sfc is adopted as the lack of Pe data from perforation tests. When x(t) = H,
which means that the projectile emerges from the back face of the concrete slab, the
corresponding velocity computed by Eq. (3.89b) is the residual velocity of the
projectile.
When the distance H2 is set as d, the calculated residual velocity is in close
agreement with the measured residual velocity; thus, H2 = d is adopted. It should be
noted that as the effect of a rear steel linear on the decrease of residual velocity is
not considered in the model, the calculated projectile residual velocities are little
lower than the experimental values for the concrete/steel composite target. Besides,
the value of H2 should be further discussed for various concrete slabs.

3.8.4.2 Numerical Simulation

Numerical simulation using finite element program LS-DYNA (Hallquist 2007) is

carried out. Because of the structural symmetry, one quarter of the geometry is
modeled to reduce the computational time, Fig. 3.47 gives the finite element model
of the projectile and concrete slab with a rear steel linear. The projectile, concrete
slab, and steel linear are all discretized by 8-node reduced integration solid hexa-
hedron elements. The mesh grid is refined in the local region of the striking point to
guarantee the requirements of computational precision and save computational cost
simultaneously. The projectile adopts the *MAT_RIGID (*MAT_20) material
model, the concrete target is described by the *MAT_JOHNSON_
HOLMQUIST_CONCRETE (*MAT_111) model, and the steel linear uses the
*MAT_PLASTIC_KINEMATIC (*MAT_3) model.
*MAT_111 (HJC) model is originally proposed by Holmquist et al. (1993),
and it is validated as an effective concrete material model for projectile impact
problems. Material parameters of the HJC model for the concrete slab are listed in
Table 3.5, most of them are identical with which given in the original reference
except for the basic mechanical parameters such as the compressive strength and
shear modulus. Table 3.6 presents the material parameters of the steel linear. The
3.8 Deceleration-Time History 101

Fig. 3.47 Finite element

model

Table 3.5 Material parameters of concrete

Parameter Value Parameter Value
Density q0 (kg/m3) 2300 Crushing pressure Pcrush 13.67
(MPa)
Shear modulus G (GPa) 13.67 Crushing volumetric strain 0.0081
Ucrush
Normalized cohesive strength A 0.79 Locking pressure Plock 0.8
(GPa)
Normalized pressure hardening B 1.6 Locking volumetric strain 0.1
Ulock
Strain rate coefficient C 0.007 Damage constant D1 0.04
Pressure hardening exponent N 0.61 Damage constant D2 1
Uniaxial compressive strength fc (MPa) 41 Pressure constant K1(GPa) 85
Maximum tensile hydrostatic pressure 3.6 Pressure constant K2 (GPa) −171
T (MPa)
Amount of plastic strain before fracture 0.01 Pressure constant K3 (GPa) 208
EFMIN
Normalized maximum strength SFMAX 7

Table 3.6 Material parameters of steel linear

Parameter Value Parameter Value
Density qs (kg/m3) 7800 Tangent modulus ETAN 0
Young’s modulus E (GPa) 210 Hardening parameter BETA 0
Poisson’s ratio PR 0.29 Failure strain for eroding elements FS 0.8
Yield stress SIGY (MPa) 400
102 3 Rigid Projectile Penetration

projectile is modeled by a solid rigid body, and its density is equivalent to

6385 kg/m3. The “CONTACT_ERODING_SURFACE_TO_SURFACE” algo-
rithm is employed between the projectile and target interface, and the contact
stiffness-scale factor is assigned as 2. The erosion algorithm *MAT_ADD_
EROSION is brought into solve the excessive element distortion problem, and the
erosion criterion is defined by the maximum principal strain of 0.8 in the present
simulation through trial-and-error method.

(a) 16
Shot 1-2
14 Raw test data
Filtered test data
12 (1439Hz low-pass )
Deceleration (10 g)

10 Theoretical model
4

Numeirical simulation
8

-2

25.2 25.4 25.6 25.8 26.0 26.2 26.4 26.6 26.8

Time (ms)

(b) 100 0.05

Shot 1-2
0 0.00
Velocity
-100 -0.05
Displacement (m)
Velocity (m/s)

-200 -0.10
Deceleration signal intergration
-300 Theoretical model -0.15
Numerical simulation
-400 -0.20

-500 -0.25
Displacement
-600 -0.30
25.2 25.4 25.6 25.8 26.0 26.2 26.4 26.6 26.8
Time (ms)

Fig. 3.48 Comparisons between the test data, theoretical prediction, and numerical simulation for
Shot 1-2 a deceleration-time history and b velocity and displacement comparisons
3.8 Deceleration-Time History 103

(a) 20
Shot 1-1

15 Raw test data

Filtered test data

Deceleration (10 g)
(1747Hz low-pass)
4
10 Theoretical model
Numeirical simulation

-5
19.8 20.0 20.2 20.4 20.6 20.8 21.0 21.2
Time (ms)

(b)
-200 0.1
Deceleration signal intergration
Shot 1-1
-250 Theoretical model
Numerical simulation
-300 0.0
-350

Displacement (m)
Velocity (m/s)

-400 -0.1
Velocity
-450
-500 -0.2
-550
300mm slab
-600 -0.3
Displacement
-650
-700 -0.4
19.8 20.0 20.2 20.4 20.6 20.8
Time (ms)

Fig. 3.49 Comparisons between the test data, theoretical prediction, and numerical simulation for
Shot 1-1 a deceleration-time history and b velocity and displacement comparisons

3.8.4.3 Comparison with Test Data

Figure 3.48 shows the experimental deceleration-, velocity-, and displacement-time

histories in Shot 1-2 as well as the corresponding curves predicted by the proposed
theoretical model and numerical simulation. It should be noted that the resistance
parameter R = 499 MPa is used in the model predictions. Figure 3.48a shows that
both the theoretical model prediction and the numerical simulation result consist
well with the experimental peak value of the rigid-body deceleration. Figure 3.48b
indicates that the velocity- and displacement-time histories predicted by the
104 3 Rigid Projectile Penetration

(a) 20
Raw test data Shot 3-2
Filtered test data
15 (2650Hz low-pass)
Theoretical model
Deceleration (10 g)
4
10 Numeirical simulation

-5

-10
12.0 12.3 12.6 12.9 13.2 13.5 13.8
Time (ms)

(b) -150 Deceleration signal intergration

0.2
Shot 3-2
-200 Theoretical model
-250 Numerical simulation 0.0
-300 200mm slab

Displacement (m)
Velocity (m/s)

-350 -0.2
-400 270mm
-450 -0.4
-500
100mm slab
-550 -0.6
-600
Velocity Displacement
-650 -0.8
12.0 12.3 12.6 12.9 13.2 13.5 13.8 14.1
Time (ms)

Fig. 3.50 Comparisons between the test data, theoretical prediction, and numerical simulation for
Shot 3-2 a deceleration-time history and b velocity and displacement comparisons

integration approach and the theoretical model as well as the numerical simulations
are in good agreements.
Figures 3.49, 3.50, and 3.51 illustrate the comparisons of the measured decel-
eration-, velocity-, and displacement-time histories during the perforations with the
predictions by the theoretical model and numerical simulation. It can be seen that
good agreements are obtained for Shots 1-1, 1-3, and 3-2. Thus, the capability of
the theoretical model and the numerical simulation in predicting the projectile’s
motion time histories is verified.
3.9 Summary 105

(a) 20
Shot 1-3
15 Raw test data
Filtered test data
Deceleration (104g)
(1953Hz low-pass )
10
Theoretical model
Numeirical simulation
5

-5

-10
22.6 22.8 23.0 23.2 23.4 23.6 23.8 24.0 24.2
Time (ms)

(b) -100 0.1

Shot 1-3
-200 0.0
Velocity

Displacement (m)
-300 -0.1
Deceleration signal intergration
Velocity (m/s)

Theoretical model
-400 Numerical simulation -0.2

300mm slab
-500 -0.3

Displacement
-600 -0.4

-700 -0.5
22.6 22.8 23.0 23.2 23.4 23.6 23.8 24.0 24.2 24.4
Time (ms)

Fig. 3.51 Comparisons between the test data, theoretical prediction, and numerical simulation for
Shot 1-3 a deceleration-time history and b velocity and displacement comparisons

3.9 Summary

In this chapter, after briefly reviewing the existing studies on the rigid projectile
penetration and perforation into concrete targets, as well as the classical cavity
expansion theory, the following five parts are introduced.
(1) In Sect. 3.3, an extended dynamic spherical cavity expansion model is firstly
proposed by using a hyperbolic yield criterion and Murnaghan equation of state
to describe the plastic behavior of concrete material under projectile penetra-
tions. A one-dimensional resistance function of concrete targets to projectile
penetrations is then formulated with the consideration of projectile nose shape
106 3 Rigid Projectile Penetration

influences. By combining with the one-dimensional momentum theorem

equation, the proposed rigid projectile penetration model gives better predic-
tions of DOP than the previous model based on the linear yield criterion and
equation of state. Furthermore, a set of practical parameters for the extended
dynamic cavity expansion model are also proposed and verified.
(2) In Sects. 3.4 and 3.5, for relatively thick concrete slabs, the concept of the
mean resistance of the target is firstly introduced based on the deceleration-time
history during projectile penetration. A simple and efficient equation to predict
the DOP of diversified targets struck normally by rigid projectiles with different
nose shapes is then proposed. The related mean resistance coefficient l = 0.4 is
obtained and verified by amounts of test data with various projectile/target
combinations. Besides, according to large amounts of test data, it is found that
the height of rear crater is about 2.5 times of the projectile diameter and the
velocity of rear concrete plugging fragments is 20% of the residual velocity of
projectiles, a simple projectile perforation model for thick concrete slabs is
obtained and validated.
(3) In Sect. 3.6, for relatively thin concrete slabs under projectile perforations, the
rigid projectile perforation test on the thin UHP-SFRC slabs is conducted and a
projectile perforation model for thin concrete slabs is established to predict the
residual velocity. The performed perforation test on thin concrete slabs and the
corresponding validated prediction model expand the scope of existing studies
on projectile perforations.
(4) In Sect. 3.7, based on the works in Sects. 3.5 and 3.6, a unified projectile
perforation model for both thick concrete slabs (H/d > 5) and thin concrete
slabs (H/d  5) is proposed and validated by comparing with the projectile
perforation test data on segmented concrete slabs.
(5) In Sect. 3.8, the rigid projectile deceleration-time histories measured in tests are
analyzed in detail. Firstly, the accuracy validation approach for the recorded
ultra-high g deceleration-time data is proposed, and a method to correct the less
accurate deceleration-time data is presented and validated. Then, a signal
processing approach is given to obtain the rigid-body projectile deceleration,
and the first valley frequency from the spectrum analysis is determined as the
critical cutoff frequency. Finally, theoretical models for the deceleration-time
histories during the projectile penetration/perforation are given and the
numerical simulations are also performed. The capability of the theoretical
models and numerical simulations in predicting the projectile’s motion time
histories is validated by comparing with the related test data.
Chapter 4
Mass Abrasive Projectile Penetration

4.1 Introduction

For the four regimes of kinetic energy projectiles penetrating into concrete targets
introduced in Chap. 1, the rigid projectile penetration is discussed in Chap. 3, while
this chapter discusses the second penetration regime-mass abrasive projectile
penetration.
With the rapid development of high-strength alloy materials and the propellant
techniques (such as scramjet, rocket, and secondary acceleration), the striking
velocity of advanced earth penetration weapon (AEPW) is possible to achieve a
higher striking velocity, e.g., 1.5 km/s, aiming to attack deep buried hardened
military facilities. At such high striking velocity, the projectile usually has signif-
icant mass loss and nose blunting when penetrating into concrete targets (Fig. 4.1a),
which may lead to its terminal ballistic instabilities, such as the structural
destruction (buckling, bending, fracture, etc., Fig. 4.1b) and ballistic trajectory
deviation (Fig. 4.1c), so as to reduce the penetration efficiency distinctly.
Investigations on the structural and terminal ballistic trajectory stabilities of the
projectile high-speed penetrating into concrete targets are helpful for the structural
design of hypersonic AEPW as well as the design of protective structures, such as
the material selection, geometry optimization, attacking gesture control of the
penetrator and the structural configuration, and yaw-inducing approaches of the
target.
In this chapter, provided that the work of the friction between the projectile and
target is totally transformed into the heat to melt penetrator material at its nose
surface, an engineering model for the mass loss and nose-blunting of the
ogive-nosed projectile is established. Furthermore, considering the symmetrical or
asymmetrical nose abrasion as well as the initial oblique and attacking angles, the
structural and terminal ballistic trajectory stabilities of the mass abrasive projectile
are discussed.

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 107
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_4
108 4 Mass Abrasive Projectile Penetration

Fig. 4.1 Structural and ballistic trajectory instabilities of projectile high-speed penetrating into
concrete targets Wu et al. (2012b, c), Wang et al. (2013, 2014), Mu and Zhang. (2011), and He
et al. (2010a)

4.2 Mass Abrasion Analysis

Since 1990s, Forrestal et al. (1996), Frew et al. (1998), He et al. (2010a), Mu and
Zhang. (2011), Wu et al. (2012b, c) conducted a series of penetration tests of
small-scaled ogive-nosed projectiles into concrete targets with maximum striking
4.2 Mass Abrasion Analysis 109

velocity up to 1682 m/s. In the above tests, the mass losses of the penetrators were
recorded and the maximum relative mass loss reached 12%. In addition, it was
observed that the majority of the mass loss occurred on the noses of the projectiles
and the abrasive noses were approaching to a hemispherical shape with increasing
the striking velocity.
Silling and Forrestal (2007) found a linearly dependence of the mass loss on the
initial kinetic energy of projectiles for the striking velocities below 1.0 km/s, by
summarizing the test data in Forrestal et al. (1996) and Frew et al. (1998). Based on
the same test data, Chen et al. (2010) presented an engineering model of mass
abrasion to predict the nose shape and mass loss of the residual projectile after
high-speed penetration into concrete targets. It was found that the linear relationship
between the mass loss and initial kinetic energy might be normalized by consid-
ering the coarse aggregate hardness. Considering the yield strength of projectile
material and the Moh’s hardness of the coarse aggregate, Wen et al. (2010) pro-
posed an empirical prediction formula for the relative mass loss, which had a power
function relation with the initial striking velocity of the projectile.
According to the metallographic observation of the recovered projectiles after
penetration, it is commonly thought that the mass loss mainly comes from the
peeling of the molten surface layer on the projectile nose and the melting heat is
totally transformed from the frictional work between the target and projectile. Based
on the dynamic cavity expansion theory and Mohr–Coulomb friction law, Jones
et al. (2002) and Davis et al. (2004) obtained the frictional work by assuming the
frictional coefficient equals to zero and decays linearly with the instant velocity of
the projectile, respectively. Further, He et al. (2010b) proposed a
velocity-dependent prediction formula for the mass loss coefficient by introducing
the Moh’s hardness of coarse aggregate of concrete. Regarding the nose-blunting
effects, Zhao et al. (2010a, 2012) proposed that the variation of the nose factor of
ogive-nosed projectiles before and after the penetration tests was linearly dependent
on the square of the initial striking velocity of projectiles and further obtained the
empirical proportional coefficient through fitting the experimental data of Forrestal
et al. (1996) and Frew et al. (1998).
It can be concluded that the majority of above-mentioned studies are empirical,
and the derived formulae commonly have the characteristics of velocity depen-
dency, poor prediction accuracy, limited physical basis, and application range. In
this section, with referring to Jones et al. (2002) and He et al. (2010b), the
dimensionless expression for the relative mass loss of projectiles is formulated by
introducing the impact function I0 and the geometry function N of projectiles. An
engineering model for the mass loss and nose blunting of the ogive-nosed projectile
is established. Comparatively, the proposed model has clear physical meaning of
transformation between frictional work and melting heat. Furthermore, the ana-
lytical expressions for the mass loss and nose-blunting coefficients are proposed and
compared with the 11 sets of existing experimental data.
110 4 Mass Abrasive Projectile Penetration

4.2.1 Mass Loss Coefficient

Based on the empirical fitting of the test data of concrete targets in Forrestal et al.
(1996) and Frew et al. (1998), Silling and Forrestal (2007) found that the mass loss
of projectiles is proportional to its initial kinetic energy in case of striking velocity
V0 less than 1.0 km/s; thus, the relative mass loss d can be expressed as

DM M0  Mr
d¼ ¼ ¼ CV02 =2 ð4:1Þ
M0 M0

where DM is the mass loss, M0 is the mass of the unfired projectile, and Mr is the
residual mass of the recovered projectile after test. C is the mass loss coefficient in
s2/km2 with respect to the striking velocity V0 in km/s. Equation (4.1) is no longer
applicable when the striking velocity exceeds 1.0 km/s.
Introducing the Moh’s hardness of the coarse aggregate in concrete targets HM,
the density of the concrete target q0 and the yield strength of the projectile material
Yp, Wen et al. (2010) proposed an empirical prediction formula for the relative mass
loss as


q0 2 0:805
d ¼ 0:0162 HM  V ð4:2Þ
2Yp 0

Since the melting and spalling of the penetrator surface are thought as the
primary causes of the projectile mass loss and the heat to melt projectile material are
totally transformed from the frictional work between the projectile and target, the
key point is to derive the amount of the frictional work. Supposed that the friction
on the interface between the penetrator and target satisfies the Mohr–Coulomb law
and the frictional coefficient equals to zero, Jones et al. (2002) obtained the fric-
tional work wt by integrating the friction along the nose surface during the whole
penetration process based on the dynamic cavity expansion theory

pd 2 s0 N0a hpen
wt ¼ ð4:3Þ
4

where s0 is the shear strength of concrete. For the concrete brittle materials, s0 ¼
pffiffiffi
fc = 3 satisfies based on the Tresca criterion, and fc is the unconfined compressive
strength of concrete. N0a is the dimensionless longitudinal section area of the unfired
penetrator’s nose. For a general nose shape shown in Fig. 3.4, it can be expressed
as

Zl0
8
N ¼ 2
a
yð xÞdx ð4:4aÞ
d
0
4.2 Mass Abrasion Analysis 111

where l0 is the length of penetrator nose, yðxÞ is the nose generatrix function in the
illustrated coordinate. In particular, Eq. (4.4a) can be simplified as for an
ogive-nosed projectile


1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N a ¼ 4w2 cos1 1   ð2w  1Þ 4w  1 ð4:4bÞ
2w

where the dimensionless nose parameter w ¼ s=d is the caliber-radius-head (CRH),

and s is the curvature radius of the ogival nose. Denoting the initial CRH of the
intact projectile as w0 ; N0a can be obtained by substituting w0 into Eq. (4.4b).
Supposing the frictional work is totally transformed into the heat to melt the
surface material at penetrator nose, the mass loss can be expressed as (Jones et al.
2002)

wt pd 2 s0 N0a hpen
DM ¼ ¼ ð4:5Þ
jQ 4jQ
R
where j = 4.18 J/cal is the mechanical equivalent of heat, Q ¼ T Cp dT is the unit
mass melting heat for the penetrator material, Cp is the specific heat of penetrator
material at constant pressure, and the integral of temperature T is from the ambient
temperature to the melting point of projectile material. The parameter of pure iron
may be taken as Q ¼ 3:0258  105 cal/kg for simplicity.
Based on the conclusions of Chen et al. (2010), and taking the Moh’s hardness
of the coarse aggregate in concrete matrix into consideration, He et al. (2010b)
introduced the dimensionless hardness of the coarse aggregate g ¼ HM =HM0 into
Eq. (4.5) and derived the mass loss as

pd 2 s0 N0a hpen
DM ¼ g ð4:6Þ
4kQ

where HM0 = 7 is the reference value for Moh’s hardness of the coarse aggregate.
When the initial striking velocity of the projectile is relatively low, it can be
considered that the mass loss has little influence on the DOP of the penetrator.
Therefore, we can firstly replace the penetration depth hpen in Eq. (4.6) with the
DOP of the rigid penetrator and then make some corrections. Chen and Li (2002)
proposed a formula for predicting the DOP of the rigid projectile as


2M0 q0 N0 V02
hpen ¼ ln 1 þ ð4:7Þ
pd 2 q0 N0 Sfc

where S ¼ 72:0fc0:5 is a dimensionless empirical constant only depending on fc

(unit in MPa). As illustrated in Fig. 3.4, the dimensionless nose factor N  can be
expressed as follows for any arbitrary nose geometry
112 4 Mass Abrasive Projectile Penetration

Zl0
 8 yy03
N ¼ 2 dx ð4:8aÞ
d 1 þ y02
0

In particular, for the ogive-nosed projectiles, Eq. (4.8a) can be simplified as

8w  1 1
N ¼ ; 0\N  \ ð4:8bÞ
24w2 2

The initial nose factor has the expression of N0 ¼ ð8w0  1Þ=24w20 . Substituting
Eq. (4.7) into Eq. (4.6) and introducing the dimensionless impact function I0 ¼
M0 V02 =Sfc d 3 and the geometry function N ¼ M0 =N0 q0 d 3 of the projectile proposed
by Chen and Li (2002), the dimensionless prediction formula for the relative mass
loss can be obtained as



gs0 N0a I0 gs0 N0a N0 q0 2
d¼ ln 1 þ or d ¼ ln 1 þ V ð4:9Þ
2q0 N0 jQ N 2q0 N0 jQ Sfc 0

Compared with Eq. (4.1), Eq. (4.9) takes into account the influences of the
initial impact velocity, projectile material and geometry, compressive strength and
the density of the target, hardness of coarse aggregate mixed in the concrete targets
on the mass loss of the projectile, and the above related parameters can be all
obtained before tests. Besides, Eq. (4.9) has a wider application range and clearer
theoretical basis than the work of Silling and Forrestal (2007) and Wen et al.
(2010). It is also shown from Eq. (4.9) that the mass loss increases nonlinearly with
the dimensionless parameters I0/N, which can reasonably interpret that the growth
rate of the mass loss tends to reduce gradually when the impact velocity exceeds
1.0 km/s in the tests. The test data of Forrestal et al. (1996) and Frew et al.
(1998) supports this viewpoint. However, Silling and Forrestal (2007) and Chen
et al. (2010) assumed that the mass loss of the projectile keeps constant when the
impact velocity exceeds 1.0 km/s.
Furthermore, the mass abrasion of penetrator nose should affect the DOP and
thus Eq. (4.7) should take the influence of mass loss into account, especially in case
of high-speed penetration. In this case, the mass loss of penetrator during pene-
tration, i.e., Eq. (4.9) should be evaluated iteratively, which can be referred to He
and Chen (2011). It is believed that the influence of mass loss on DOP is of second
order (Zhao et al. 2010a). Thus, the prediction of Eq. (4.9) on the mass loss of
projectiles in case of high-speed penetration, e.g., exceeding 0.9 km/s, essentially
indicates that the mass loss rises as the impact velocity increases. It should be noted
that the transformation between heat and frictional work is only considered here, as
supposing that melting and spalling of the penetrator surface materials are the
dominant causes of the mass loss. If other mechanisms of the mass abrasion are
taken into account, analyses will become much more complicated.
4.2 Mass Abrasion Analysis 113

Next, we will discuss relative low-velocity penetration in the following analysis.

He et al. (2010a), Mu and Zhang (2011), Silling and Forrestal (2007) indicated that
the mass loss of the penetrator increases linearly with the initial impact kinetic
energy of the projectile when the striking velocity is lower than a critical value,
such as shown in Eq. (4.1).
The Taylor series of Eq. (4.9) has the form as


gN0a V02 1 N0 q0 2
d ¼ pffiffiffi 1 V þ  ð4:10Þ
2 3jQS 2 Sfc 0

The mass loss coefficient C can be derived by comparing Eq. (4.10) with
Eq. (4.1)


gN a 1 N0 q0 2
C ¼ pffiffiffi 0 1 V0 þ    ð4:11Þ
3jQS 2 Sfc

Assuming that the deviation of the mass loss is less than 20%, we can employ
the first-order Taylor expansion of Eq. (4.11) to approximate Eq. (4.9), which
requests N0 q0 V02 =2Sfc \1=5: Thus, an upper critical impact velocity V0c can be
derived as
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
V0c ¼ 2Sfc =5N0 q0 ð4:12Þ

Apart from the initial striking velocity, the factors affecting the deviation of the
mass loss also include the assumption of rigid projectiles and the incompleteness of
actual transformation between the frictional work and melting heat. Generally, the
prediction of the mass loss coefficient C obtained from Eq. (4.11) is relatively larger
than the true value. With considering the general deviation of the mass loss is less
than 20% in Eq. (4.11), we can derive the corrected value of C. It means that, when
the initial striking velocity of projectiles satisfies V0  V0c ; the mass loss of the
penetrator increases linearly with the initial kinetic energy of the projectile
increasing, and the analytical expression of the mass loss coefficient C can be
obtained as

0:8gN a
C ¼ pffiffiffi 0 ð4:13Þ
3jQS

In the case of relatively low-velocity penetration, Eqs. (4.12, 4.13) give the
analytical formulae of the upper critical impact velocity and the mass loss coeffi-
cient, respectively. Thus, the physical basis to support the empirical conclusions of
Silling and Forrestal (2007) is reasonably constructed.
When the inequality V0  V0c is satisfied, it can be found from Eq. (4.13) that the
mass loss coefficient of the projectile is related to the compressive strength of
concrete, the Moh’s hardness of the coarse aggregate, and the initial nose geometry
114 4 Mass Abrasive Projectile Penetration

Fig. 4.2 Amplifications of 2.8

C versus the corresponding 2.6

Amplifications of C
influential parameter 2.4
2.2
2.0
1.8 HM
1.6 fc
1.4 ψ0
1.2
1.0
1 2 3 4 5 6 7
Amplifications of influential parameters

as well as the unit mass melting heat of the penetrator material. All the
above-mentioned parameters can be obtained before penetration tests, and thus,
Eq. (4.13) is able to predict the mass loss of the projectile and guide the design of
projectiles and protective structures. For an ogive-nosed projectile, adopting high
melting heat penetrator material and properly reducing the CRH value of the pro-
jectile nose will decrease the mass loss in the projectile penetration. Regarding the
design of protective structures, the penetration efficiency of the penetrator will be
reduced by enhancing the compressive strength of the target and the Moh’s hard-
ness of the coarse aggregate in concrete. Figure 4.2 shows the effects of the amplifi-
cations of the influential parameters HM, fc and w0 on the amplifications of C. The
reference parameters are HM = 3, fc = 30 MPa, and w0 = 2, and the variation
ranges of the parameters are 3  HM  7; 30 MPa  fc  200 MPa and
2  w0  9:25; respectively. It indicates that the mass loss coefficient C increases
with increasing HM, fc, and w0 ; respectively. C is mostly sensitive to HM, while the
influencing degrees of w0 and fc are almost the same.

4.2.2 Nose-Blunting Coefficient

During the penetration process, the projectile nose is blunted continuously. By

fitting the test data in Forrestal et al. (1996) and Frew et al. (1998), Zhao et al.
(2010a) obtained that the variation of the nose factor N  of the ogive-nosed pro-
jectiles before and after penetration test has a linear relationship with the square of
the initial impact velocity, which can be expressed as

Nr ¼ N0 þ k0 V02 ð4:14Þ

where k0 is the nose-blunting coefficient of the projectile in s2/km2 with respect to

the striking velocity V0 in km/s. N0 is the initial nose factor of the unfired
projectile before penetration test, shown in Eqs. (4.8a, b). Nr is the residual nose
factor of the recovered projectile after penetration test, which has the expression
4.2 Mass Abrasion Analysis 115

of Nr ¼ ð8wr  1Þ=24w2r if the residual CRH of the recovered projectile nose is
denoted as wr :
Furthermore, Zhao et al. (2012) gave an empirical prediction of k0 as

fc N  HM
k 0 ¼ 0:336   0  ð4:15Þ
62:8 0:106 7

The existing high-speed penetration tests of the ogive-nosed projectiles indicate

that the majority of the mass loss occurs on the projectile nose, and the abrasive
nose is keeping ogival and approaching to a hemispherical shape with increasing
the striking velocity. The variation of the nose geometry is closely related to the
mass loss of the penetrator. Denoting DM and DMn are the mass losses of the whole
projectile and the nose part, respectively, it should have DM = DMn : Assuming that
the density of the projectile keeps unchanged during the penetration, based on
Eq. (4.1), we have


1 DM DMn Mn V ðwr Þ Mn
CV 2 ¼ ¼  ¼ 1  ð4:16Þ
2 0 M0 Mn M0 V ðw0 Þ M0

where M0 and Mn are the masses of the entire and the nose part of the unfired
projectile, respectively. V ðw0 Þ and V ðwr Þ are the nose volumes of the unfired and
recovered projectiles, respectively. For an ogive-nosed projectile, it has

V ðwÞ
¼ f ðw Þ
pd 3 "sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi




#
1 1 1 1 1 1 1
¼w 3
1 1 þ  1 cos 1
w 4w 3w 12w2 2w 2w
ð4:17Þ

Substituting Eq. (4.17) into Eq. (4.16), we have

CM0
f ðwr Þ ¼ f ðw0 Þ  f ðw0 ÞV02 ð4:18Þ
2Mn

Figure 4.3 shows the relationship between f ðwÞ and nose CRH w in solid line,
where CRH w varies within our interesting range of 3  w  6: The nearly linear
relationship between f ðwÞ and w is observed, and thus, we adopt the dashed straight
line through the points (3, f(3)) and (6, f(6)) in Fig. 4.3 to approximate the
dimensionless function of the nose volume f ðwÞ: Therein Eq. (4.18) can be sim-
plified as

CM0 2
wr ¼ w0  ðw0 þ 4Þ V ð4:19Þ
2Mn 0
116 4 Mass Abrasive Projectile Penetration

Fig. 4.3 Original and the 0.34

approximation curves for Original curve
0.32
f ðwÞ Approximation curve

0.30

f (ψ )
0.28

0.26

0.24

0.22

2 3 4 5 6 7
ψ

Substituting Eqs. (4.19, 4.11) into Eq. (4.14), and comparing the coefficients of
V02on both sides of the equation, the expression of the nose-blunting coefficient k0
can be obtained as


0 ðw0 þ 4ÞM0 gN0a ðw0 þ 4ÞM0 1 N0 q0 2
k ¼C ¼ pffiffiffi 1 V þ  ð4:20Þ
6Mn w20 6 3jQSMn w20 2 Sfc 0

Since the above approximation of Eq. (4.19) might lead to underestimate the
mass loss and the value of k 0 ; we only adopt the first term in Eq. (4.20) to represent
the value of k 0 : Thus, the analytical form of the nose-blunting coefficient k 0 can be
drawn as

gN a ðw þ 4ÞM0
k 0 ¼ p0ffiffiffi 0 ð4:21Þ
6 3jQSMn w20

Similar to the mass loss coefficient C in Eq. (4.13), for the given projectile and
concrete target, all the related parameters of the nose-blunting coefficient k0 in
Eq. (4.21) can be confirmed before the penetration tests. Therefore, the residual
nose geometry of an ogive-nosed projectile after penetration test can be predicted
beforehand by combining Eqs. (4.14, 4.21).
Equation (4.21) presents the analytical formulae of the nose-blunting coefficient
k0 : Comparing with Eq. (4.15), it reasonably constructs the physical basis to support
the empirical conclusions of Zhao et al. (2010a, 2012). It should be noted that
Eq. (4.21) is only available in the range of V0 \V0c as the linearity of d ¼ CV02 =2
exists in this range.

4.2.3 Prediction of Mass Loss in All Striking Velocity Range

This section mainly aims to present a prediction of the mass loss of the projectile at
all striking velocity range and Eq. (4.9) will be employed. Table 4.1 collects the
4.2 Mass Abrasion Analysis 117

basic parameters of projectiles and targets in total 11 sets of high-speed penetration

tests, in which the striking velocity varies from 345 to 1475 m/s. Symbol L in
Table 4.1 is the total length of the projectile. According to the penetration test data
in Table 4.1, Fig. 4.4 shows the comparisons of the relative mass loss d between
the experimental data and the theoretical predictions by Eq. (4.9). It should be
noted that the abscissa axis in the coordinates is the dimensionless parameter I0/N.
Based on Eq. (4.9), we may further compare the relative mass loss of a projectile
under different impact conditions of projectiles and concrete targets. Equation (4.9)
distinctly indicates that d–I/N dependence is influenced by HM, fc ; N0a ; N0 , and q0 if
the unit mass melting heat for the pure iron Q ¼ 3:0258  105 cal/kg is adopted.
The density q0 of concrete is limited within a narrow variation range, and thus, its
influence on the d–I/N curve is neglected. Therein, just for an example, referring to
the parameters of Test 3 (HM = 7, fc = 62.8 MPa, N0a = 4.5 and N0 = 0.106), the
other 10 sets of test data in Table 4.1 can be simply normalized. That is to say, if
the measured relative mass loss is d for any test with parameters HM, fc ; N0a and N0 ;
its modified relative mass loss based on Eq. (4.9) will be changed as d0 ¼
   
d  ð7=HM Þ  ð62:8=fc Þ  4:5=N0a  N0 =0:106 : Via the above approach, the com-
parisons between the test data of d0  I0 =N and the predicted curve by Eq. (4.9) are
illustrated in Fig. 4.5.
The following four issues should be noted by analyzing Figs. 4.4 and 4.5.
(i) For all striking velocity ranging from low at 300 m/s to high at 1475 m/s,
Eq. (4.9) reasonably predicts the mass loss of the various projectiles during
penetration into concrete targets. In particular, Eq. (4.9) gives a better
description on the tendency of the mass loss at high-speed penetration, i.e.,
its growth rate obviously becomes slower with comparing to the scenario of
low-velocity penetration.
(ii) The majority of the experimental data of the mass loss in Fig. 4.4 is smaller
than the theoretical predictions by Eq. (4.9), especially when the striking
velocity is larger, i.e., in case of I0 =N [ 0:4 in Fig. 4.4. The possible reason
is due to the adoption of pure iron parameter and the assumption of rigid
projectile to calculate DOP, which generally tends to magnify the theoretical
prediction of the mass loss. In addition, as mentioned previously, only the
melting and spalling of the penetrator surface material is considered as the
dominant causes of the mass loss whereas no other mechanisms of the mass
abrasion are taken into account, and it is absolutely the extreme case.
(iii) Regarding the Tests 1 and 2, the experimental results are larger than the
theoretical predictions obtained by Eq. (4.9), which may be caused by the
measuring deviation of the mass loss in the tests. Since the projectile mass is
only 64 g, the maximum value of the mass loss is about 7.9% of the pro-
jectile mass and it equals to about 5.1 g. In that case, it is quite difficult to
weigh the projectile mass accurately. For example, if a little concrete clay is
included, the absolute measuring deviation will zoom evidently.
 118

Table 4.1 Parameters of targets and projectiles in the high-speed penetration tests
Ref. Forrestal et al. (1996) Frew et al. He et al. (2010a) Wu et al. (2012b)
(1998)
No. 1 2 3 4 5 6 7 8 9 10 11
Target Grout Concrete Concrete
fc (MPa) 13.5 21.6 62.8 51 58.4 34.8 48.6 61.3 76.4 27.3
q0 (kg/m3) 2000 2300 2320 2300 2300
Aggregate Quartz Limestone Quartz
HM 7 3 7
Projectile material 4340 steel 4340 60Si2Mn 60Si2Mn 60Si2Mn 60Si2Mn 30CrMnSi
steel/AerMet /TC4 /20#steel /45#steel /35CrMnSi /45#steel
100
Yp (MPa) 1481 1481/1820 1300/1030 1300/450 1300/680 1300/1540 1130/835
Solid or Hollow Hollow Solid Solid/Hollow
M0 (kg) 0.064 0.478 1.6 0.478 1.62 0.154 0.127/0.101
M0/Mn 5.7/4.6 10.2 8.55 6.64/5.35
d (mm) 12.9 20.3 30.5 20.3 30.5 14 15
L/d 6.88 10 7
w0 3/4.25 3 4.25 3
4 Mass Abrasive Projectile Penetration
4.2 Mass Abrasion Analysis 119

(a) 0.10 Forrestal et al. (1996) CRH=3

(b) 0.14 Forrestal (1996) CRH=3
Eq. (4.9) CRH=3 0.12 Eq. (4.9) CRH=3
0.08 Forrestal et al. (1996) CRH=4.25 Forrestal (1996) CRH=4.25
Eq. (4.9) CRH=4.25 0.10 Eq. (4.9) CRH=4.25
0.06 0.08

0.06
0.04
0.04
0.02
0.02
0.00 0.00
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.2 0.4 0.6 0.8 1.0
I 0 /N I 0 /N

(c) 0.12 (d) 0.16

Forrestal et al. (1996) Forrestal (1996)
0.14 Eq. (4.9)
0.10 Eq. (4.9)
0.12
0.08
0.10
0.06 0.08
0.06
0.04
0.04
0.02
0.02
0.00 0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
I 0 /N I 0 /N

(e) 0.05 (f) 0.06

Frew et al. (1998) 4340 steel Frew et al. (1998) 4340 steel
Frew et al. (1998) AerMet 100 0.05 Frew et al. (1998) AerMet 100
0.04 Eq. (4.9) Eq. (4.9)
0.04
0.03
0.03
0.02
0.02

0.01 0.01

0.00 0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7

I 0 /N I 0 /N

(g) 0.08 (h)

He et al. (2010a) 60Si2Mn 0.10
0.07 He et al. (2010a) Tc4 He et al. (2010a) 60Si2Mn
Eq. (4.9) He et al. (2010a) 20# Steel
0.06 0.08 Eq. (4.9)
0.05
0.06
0.04
0.03 0.04

0.02
0.02
0.01
0.00 0.00
0.0 0.2 0.4 0.6 0.8 1.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
I 0 /N I 0 /N

Fig. 4.4 Comparison of predicted d–I0/N with experimental data from a Test 1, b Test 2, c Test 3,
d Test 4, e Test 5, f Test 6, g Test 7, h Test 8, i Test 9, j Test 10, and k Test 11
120 4 Mass Abrasive Projectile Penetration

(i) 0.12 (j) 0.12

He et al. (2010a) 60Si2Mn He et al.(2010a) 60Si2Mn
0.10 He et al. (2010a) 45# Steel 0.10 He et al.(2010a) 35CrMnSi
Eq. (4.9) Eq. (4.9)
0.08 0.08

0.06 0.06

0.04 0.04

0.02 0.02

0.00 0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.0 0.1 0.2 0.3 0.4 0.5 0.6
I 0 /N I 0 /N

(k) 0.20
Wu et al. (2012b) 30CrMnSi Solid
Wu et al. (2012b) 45# steel Hollow
0.16 Eq. (4.9)

0.12

0.08

0.04

0.00
0.0 0.2 0.4 0.6 0.8 1.0
I 0 /N

Fig. 4.4 (continued)

Fig. 4.5 Modified 0.4

Modified Test 1 Modified Test 2
experimental data and Test 3 Modified Test 4
predicted curve of d0 –I0/N Modified Test 5 Modified Test 6
0.3 Modified Test 7 Modified Test 8
Modified Test 9 Modified Test 10
Modified Test 11 Eq. (4.9)
0.2

0.1

0.0
0.0 0.2 0.4 0.6 0.8 1.0
I0 /N

(iv) Equation (4.9) can be applied to predict the mass loss for any nose-shaped
projectiles theoretically. However, since the existing tests are all limited to
the ogive-nosed penetrators, the validity of Eq. (4.9) for other nose-shaped
penetrator needs to be further verified.
4.2 Mass Abrasion Analysis 121

(a) 0.06 (b) 0.08 Forrestal et al. (1996) CRH=3

Forrestal et al. (1996) CRH=3 0.07 Forrestal et al. (1996) CRH=4.25
Forrestal et al. (1996) CRH=4.25 Eq. (4.13) CRH=3
0.05
Eq. (4.13) CRH=3 0.06 Eq. (4.13) CRH=4.25
Eq. (4.13) CRH=4.25
0.04 0.05

0.03 0.04
C =0.13
0.03
0.02 C =0.1
0.02
0.01 C=0.08 C=0.11
0.01
V c0=0.705 V c0=0.833 V c0=0.792 V c0=0.937
0.00 0.00
0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4 0.5
V 20 /2 (km/s) 2 V 20 /2 (km/s) 2

(c) 0.14 (d) 0.08

Forrestal et al. (1996) Forrestal et al. (1996)
0.12 Eq. (4.13) Eq. (4.13)
0.06
0.10
C =0.16
0.08
C =0.18 0.04
0.06

0.04 0.02
0.02
V c0=0.965 V c0=0.965
0.00 0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.0 0.1 0.2 0.3 0.4 0.5
V 20 /2 (km/s) 2 V 20 /2 (km/s) 2

(e) 0.05 (f) 0.05

Frew et al. (1998) 4340 steel Frew et al. (1998) 4340 steel
Frew et al. (1998) AerMet 100 Eq. (4.13)
0.04 Eq. (4.13) 0.04

0.03 0.03
C =0.074
0.02 0.02

C =0.074 0.01
0.01

V c0=0.943 V c0=0.943
0.00
0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.1 0.2 0.3 0.4 0.5

V 20 /2 (km/s) 2 V 20 /2 (km/s) 2

Fig. 4.6 Experimental data and the predicted d  0:5V02 dependence curves for V0 \V0c a Test 1,
b Test 2, c Test 3, d Test 4, e Test 5, f Test 6, g Test 7, h Test 8, i Test 9, j Test 10, and k Test 11

4.2.4 Validation of Mass Loss Coefficient

In this section, in case of the initial striking velocity lower than the upper critical
impact velocity V0c in Eq. (4.12), the validity of the analytical expression of the
mass loss coefficient C in Eq. (4.13) will be further verified experimentally. Based
122 4 Mass Abrasive Projectile Penetration

(g) 0.07 (h) 0.07

He et al. (2010a) 60Si2Mn Heet al.(2010a) 60Si2Mn
0.06 Heet al.(2010a) Tc4 0.06 Heet al.(2010a) 20# Steel
Eq. (4.13) Eq. (4.13)
0.05 0.05
0.04 0.04
0.03 0.03
C=0.08
0.02 0.02
C=0.07
0.01 0.01
V c0=0.985 V c0=1.07
0.00 0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
V 20 /2 (km/s) 2 V 20 /2 (km/s) 2

(i) 0.08 (j) 0.10

Heet al.(2010a) 60Si2Mn He et al.(2010a) 60Si2Mn
0.07 0.09
Heet al.(2010a) 45# Steel Heet al.(2010a) 35CrMnSi
Eq. (4.13) 0.08 Eq. (4.13)
0.06
0.07
0.05 0.06
0.04 0.05
C =0.1
0.03 C =0.088 0.04
0.03
0.02
0.02
0.01 0.01
V c0=1.135 V c0=1.199
0.00 0.00
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8
V 20 /2 (km/s) 2 V 20 /2 (km/s) 2

(k) 0.08 Wuet al.(2012b) 30CrMnSi Solid

0.07 Wuet al.(2012b) 45# steel Hollow
Eq. (4.13)
0.06
0.05
0.04
0.03
C =0.12
0.02
0.01
V c0=0.834
0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
V 20 /2 (km/s) 2

Fig. 4.6 (continued)

on the 11 sets of test data in Table 4.1, Fig. 4.6 shows the experimental results of
d  0:5V02 and the corresponding predictions by Eq. (4.13), in which the test data
exceeding V0c is ignored. The values of the V0c (km/s) and C (s2/km2) predicted by
Eqs. (4.12, 4.13) are both shown out.
Similar to Sect. 4.2.3, referring to the parameters of the test 3 (HM = 7,
fc ¼ 62:8 MPa, N0a ¼ 4:5), the other 10 sets of test data in Table 4.1 can be simply
4.2 Mass Abrasion Analysis 123

Fig. 4.7 Modified 0.16

Modified Test 1 Modified Test 2 Test 3
experimental data and the 0.14 Modified Test 4 Modified Test 5 Modified Test 6
predicted curves of Modified Test 7 Modified Test 8 Modified Test 9
0.12 Modified Test 10 Modified Test 11
d00  0:5V02 for V0  V0c
0.10
0.08 linear fit C =0.2
0.06
0.04
0.02 Eq. (4.13), C=0.18
V c0 =0.965
0.00
0.0 0.1 0.2 0.3 0.4 0.5
V 20 /2 (km/s)2

normalized. That is to say, for any test other than the Test 3 in Table 4.1, according
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi 
00
to Eq. (4.13), its modified relative mass loss is d ¼ d  ð7=HM Þ  62:8=fc 
 
4:5=N0a : Via this approach, the derived test data of d00  0:5V02 is illustrated in
Fig. 4.7. The solid straight line (C = 0.18) in Fig. 4.7 is the predicted curve by
Eq. (4.13). Comparatively, the dashed straight line (C = 0.2) in Fig. 4.7 is the
linear fitting of the test data by means of the least squares method, and the relative
deviation is 10%.
Some remarks are given from Figs. 4.6 and 4.7 as follows.
(i) Regarding all 11 sets of penetration tests, the predicted upper critical impact
velocities V0c by Eq. (4.12) range from 0.705 to 1.199 km/s, and the average
value is about 0.95 km/s. Thus, the applicable upper critical impact velocity
can be approximated as V0c ¼ 1:0 km/s, which is the same as the empirical
conclusion drawn by Silling and Forrestal (2007).
(ii) The predicted mass loss coefficients agree well with the test data. The
validity of the linear dependence of d  0:5V02 is verified when the initial
striking velocities are less than 1.0 km/s. Therefore, the theoretical basis of
the empirical conclusions drawn by Silling and Forrestal (2007) is
constructed.
(iii) The validities of Eqs. (4.12, 4.13) for other nose-shaped projectiles need to
be further verified by high-speed penetration tests.

4.2.5 Validation of Nose-Blunting Coefficient

In this section, the validity of Eq. (4.21) in predicting the residual nose factor of the
projectile after the penetration is verified by comparing with the test data in
Table 4.1. Figure 4.8 gives the test data of Nr  V02 as well as the predicted curves
by Eq. (4.21).
124 4 Mass Abrasive Projectile Penetration

(a) 0.4
Forrestal et al. (1996) CRH=3 N*
(b) 0.4
0=0.106 Forrestal et al. (1996) CRH=3 N*0=0.106
Forrestal et al. (1996) CRH=4.25 N*
0=0.076 Forrestal et al. (1996) CRH=4.25 N*0=0.076
Eq. (4.21) CRH=3 Eq. (4.21) CRH=3
0.3 Eq. (4.21) CRH=4.25 0.3
Eq. (4.21) CRH=4.25

k'=0.077

N*
N *r

r
0.2 0.2 k'=0.097

0.1 0.1
k'=0.055
k'=0.044

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
V 20 (km/s) 2 V 20 (km/s) 2

(c) 0.8 (d) 0.8

Forrestal et al. (1996) N* Forrestal et al. (1996) N*
0=0.106
0.7 0=0.106 0.7
Eq. (4.21) Eq. (4.21)
0.6 0.6
0.5 0.5
N *r

N *r
0.4 k'=0.3 0.4
0.3 k'=0.27
0.3
0.2 0.2

0.1 0.1
0.0
0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6

V 2 (km/s) 2 V 20 (km/s) 2
0

(e) 0.4 (f) 0.4

Frew et al. (1998) 4340 steel N*0=0.106 Frew et al. (1998) 4340 steel N*0=0.106
Frew et al. (1998) AerMet 100 N*0=0.106 Frew et al. (1998) AerMet 100 N*0=0.106
Eq. (4.21) Eq. (4.21)
0.3 0.3
N*

N*
r

0.2 0.2
k'=0.12 k'=0.12
0.1 0.1

0.0 0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
V 20 (km/s) 2 V 20 (km/s) 2

Fig. 4.8 Experimental and predicted Nr –V02 curves a Test 1, b Test 2, c Test 3, d Test 4, e Test 5,
f Test 6, g Test 7, h Test 8, i Test 9, j Test 10, and k Test 11

Figure 4.8 illustrates the following: (i) the analytical expression of Eq. (4.21)
has a good theoretical prediction of the nose-blunting coefficient compared to the
test data. Regarding to the completely empirical determination of the nose-blunting
coefficient in Zhao et al. (2010a, 2012), the present analysis shows its physical
characteristic and also guarantees its prediction accuracy; (ii) although the
4.2 Mass Abrasion Analysis 125

(g) 0.4 (h) 0.4

He et al. (2010a) 60Si2Mn N* 0=0.076 He et al. (2010a) 60Si2Mn N* 0=0.076
He et al. (2010a) Tc4 N*0=0.076 He et al. (2010a) 20# Steel N*
0=0.076
0.3 Eq. (4.21) Eq. (4.21)
0.3

N*
N *r

r
0.2 0.2

0.1 k'=0.056 0.1 k'=0.066

0.0 0.0
0.0 0.4 0.8 1.2 1.6 2.0 0.0 0.4 0.8 1.2 1.6 2.0
V 20 (km/s)2 V 20 (km/s)2

(i) 0.4 (j) 0.4

He et al. (2010a) 60Si2Mn N*0=0.076 He et al. (2010a) 60Si2Mn N*0=0.076
He et al. (2010a) 45# Steel N*0=0.076 He et al. (2010a) 35CrMnSi N*0=0.076
Eq. (4.21) 0.3 Eq. (4.21)
0.3
N*

N*
r

0.2 0.2

0.1 0.1
k'=0.074 k'=0.083

0.0 0.0
0.0 0.4 0.8 1.2 1.6 2.0 0.0 0.4 0.8 1.2 1.6 2.0
V 20 (km/s)2 V 20 (km/s)2

(k) 0.5
Wu et al. (2012b) 30CrMnSi Solid N*0=0.106
Wu et al. (2012b) 45# steel Hollow N *0=0.106
0.4 Eq. (4.21) 30CrMnSi Solid
Eq. (4.21) 45# steel Hollow
0.3 k'=0.128
N*
r

0.2 k'=0.1

0.1

0.0
0.0 0.4 0.8 1.2 1.6 2.0
V 20 (km/s)2

Fig. 4.8 (continued)

formulation of Eq. (4.21) is restricted in the range of V0 \V0c as the linearity of

d ¼ CV02 =2 exists in this range, it is found from Fig. 4.8 that the analytical deter-
mination of the nose-blunting coefficient may extend to a larger velocity range. As
shown in Fig. 4.8, it basically covers all the tested velocities up to 1475 m/s. The
reason may be caused by the weak linearity of d ¼ CV02 =2 even when the initial
striking velocity exceeds the critical value, i.e., V0 [ V0c ; as shown in Fig. 4.4.
126 4 Mass Abrasive Projectile Penetration

4.3 Structural Stability Analysis

With the increasing of the impact velocity, the high pressure and temperature at the
projectile/target interface, the initial oblique and attacking angles, as well as the
friction and indentation of the randomly distributed coarse aggregate in the concrete
target, lead to the non-axisymmetric interaction between the target and the projectile
nose; thus, the symmetrical or asymmetrical nose abrasions occur inevitably (Mu
and Zhang 2011; Wu et al. 2012b; Erengil and Cargile 2002; Liang et al. 2008). For
instance, as shown in Fig. 4.9, Chen et al. (2010) depicted the actual (black lines) as
well as the approximated symmetrical abrasive (red lines) envelope contours of the
recovered projectile noses, according to the penetration test conducted by Forrestal
et al. (1996). The initial CRH values of the unfired projectiles are 3 and 4.25 in
Fig. 4.9a and b, and the initial striking velocity ranges are 492–1192 and 473–
1190 m/s, respectively.
When the initial striking velocity increases up to a critical value, which we call it
as the limit striking velocity (LSV), the structural destruction of the projectile takes
place (Lundgren 1994; Zhao et al. 2010b, 2012), and the projectile is bended or
fractured, which leads the penetration efficiency decreasing distinctly. Generally,
the projectile is under the combined action of the axial and transverse drag forces
attributed by the symmetrical or asymmetrical nose abrasion as well as the initial
oblique and attacking angles. Regarding the structural stability of the projectile
penetrating into concrete targets, Chen (2005), Pi and Huang (2007), and Wang
et al. (2010a) analyzed the drag forces acting on the projectile under normal pen-
etration, oblique penetration as well as the penetration with a small attacking angles
(0°–1°), respectively. The variation ranges of the initial oblique angle and the

Fig. 4.9 Actual and predicted envelope contours of the recovered projectile noses from
penetration tests (Chen et al. 2010)
4.3 Structural Stability Analysis 127

cartridge thickness that insuring the projectile maintains its structural integral sta-
bility were discussed. Furthermore, by considering the influence of the nose
abrasion on the structural response of the projectile, Zhao et al. (2011, 2012)
discussed the LSVs of the normal and oblique striking projectiles, respectively.
However, the above descriptions on the mass loss and nose-blunting laws are
derived by test data fitting, which have little physical basis; thus, their application
ranges are limited. Besides, the detailed discussions on the combined influences of
the initial oblique and attacking angles as well as the projectile’s nose abrasions on
LSV are limited.
In this section, based on the engineering model for the mass loss and
nose-blunting of the ogive-nose projectile established in Sect. 4.2, with considering
the asymmetrical nose abrasion, the initial oblique and attacking angles, both the
axial and the transverse drag forces acting on the projectile are derived. Furthermore,
the expression of LSV is obtained based on the ideal elasto-plastic yield criterion of
the projectile. The regressive expressions in particular penetration scenarios are
consistent with the previous studies. Finally, the experimental validation and para-
metric influential analysis of the proposed approaches are conducted.

4.3.1 Drag Forces

As illustrated in Fig. 4.10a, for the most unfavorable condition, we assume that the
plane of oblique angle coincides with the plane of attacking angle. The original
point of coordinate system x-z is fixed to the projectile centroid CG, and the z-axis
is along the projectile axis. The instantaneous velocity of the projectile centroid is
denoted by v, and the initial oblique angle (the angle between the normal line of the
horizontal target surface and the projectile axis) and attacking angle (the angle
between the CG velocity and the projectile axis) of the projectile are denoted as a
and b, respectively.
For convenience, it is assumed that the drag force acting on the nose tip and the
projectile pose is unchangeable during the penetration until yielding, i.e., the
oblique and attacking angles keep constant. As shown in Fig. 4.10a, the axial and
transverse components of the drag force acting on the projectile are denoted as Faxis
and Fvert, respectively.
Figure 4.10b shows the 3D schematic diagram and the longitudinal section
profile of a projectile nose, where rr stands for the normal stress acting on an
arbitrary surface point A of the projectile, and g is the angle between the penetrator
axis and the tangential line across point A in the longitudinal section. r is the radius
of the transverse cross section crossing the point A, and U is the angle between the
x-axis and the projection of rr on the above transverse cross section.
Regarding the asymmetrical nose abrasion, supposing that the two half sides of
the projectile nose (I and II) are blunted with different degrees, and thus, the side
I (p=2  U\p=2) and the side II (p=2  U\3p=2) have different CRHs (denoted
by wI and wII ) after asymmetrical mass abrasion. The instantaneous transverse cross
128 4 Mass Abrasive Projectile Penetration

section of the projectile nose is given in Fig. 4.10c, where the solid and dashed
contours are corresponding to the original and asymmetrical abrasive projectile,
respectively.
As given in Fig. 4.10, only rigid-body movement and axial rotation of the pro-
jectile occur before yielding, and the velocity components of the central of the mass
are (vx;CG ; vy;CG ; vz;CG ); thus, the velocity components of the point A (x, y, z) can be
expressed as

vx ¼ vx;CG þ zh_ y þ r sin Uh_ z

vy ¼ vy;CG þ zh_ x þ r cos Uh_ z ð4:22Þ
vz ¼ vz;CG þ xh_ y þ yh_ x

where h_ y ; h_ x , and h_ z are the angular velocity of the projectile surrounding the axes y,
x, and z, respectively. Based on the above assumption of the unchangeable pene-
tration pose, the rotations of the projectile surround its axis are neglected, which has
h_ x ¼ h_ y ¼ h_ z ¼ 0: Also, from Fig. 4.11a, we can derive that vx;CG ¼ v sin b;
vy;CG ¼ 0 and vz;CG ¼ v cos b:
Based on Fig. 4.10b, the normal expansion velocity of an arbitrary surface point
A of the projectile Vc can be expressed as

Vc ¼ v cos b sin g þ v sin b cos g cos U ð4:23Þ

The normal stress rr on the projectile nose is derived from the spherical cavity
expansion theory and expressed by Forrestal et al. (1994) as

rr ¼ Sfc þ q0 Vc2 ; S ¼ 72:0 fc0:5 ð4:24Þ

(a)

α x
(b) (c)
CG

β
r A r x
v z r Φ x x II I
Φ
A
η
A
F ax

I ert
Fv
is

II z z y

Fig. 4.10 Illustration of non-normal projectile penetration

4.3 Structural Stability Analysis 129

(a) 2000 (b) 1600

fc =30MPa
λ 1=0.5 λ 1=0.2 λ 1=0.15
1800 λ 1=0.125 λ 1=0.1 fc =40MPa
1400
1600 fc =50MPa
V0,lim (m/s)

fc =100MPa

V0,lim (m/s)
1200
1400
1200 1000
1000
800
800
600
600
400 400
30 40 50 60 70 80 90 100 0.08 0.10 0.12 0.14 0.16 0.18 0.20
fc (MPa) λ1

Fig. 4.11 Dependence curves of LSV with a strength of concrete targets, and b dimensionless
thickness of projectile cartridges

Equation (4.24) is obtained by the regression of the test data based on the
spherical cavity expansion theory. In general, the tangential frictional force acted on
the projectile can be neglected when employing Eq. (4.24) in the drag force
analyses.
During the penetration process, both axial and transverse drag forces Faxis and
Fvert are acting on the projectile. Based on Fig. 4.10, the axial resistance Faxis has
the form as
ZZ
Faxis ¼ rr sin gdA ð4:25aÞ
R

where R denotes the lateral area of the ogival part of the projectile. The transverse
1
drag force Fvert consists of two parts, one is the transverse resistance Fvert attributed
from the initial oblique and attacking angles, and the other is the transverse
2
resistance Fvert attributed from the asymmetrical nose abrasion and attacking angles
during the penetration process. Here, we consider that Fvert ¼ Fvert 1
þ Fvert
2
for the
most unfavorable condition.
The penetration damages of the semi-infinite target are commonly a conical
crater followed by a tunnel with the diameter equals to that of the projectile shank.
1
After the projectile entering the target, Fvert usually increases to the maximum and
1
then drops to zero at the tunneling stage. The average value of the Fvert can be
expressed as (Chen 2005; Zhao et al. 2011)

1
1
Fvert ¼ Faxis;max sin a ð4:25bÞ
2
130 4 Mass Abrasive Projectile Penetration

where Faxis;max is the maximum of the axial drag force Faxis with considering the
attacking angle and the mass abrasion. It should be noted that for a rigid projectile,
Faxis;max usually occurs at the transition between the cratering and the tunneling
stages. When the projectile nose abrasion is considered, Faxis;max usually occurs at
the tunneling stage, and the detailed discussions can be referred to He and Chen
(2011).
2
Based on Fig. 4.10, the transverse resistance Fvert can be expressed by
ZZ
2
Fvert ¼ rr cos g cos UdA ð4:25cÞ
R

Substituting Eqs. (4.23, 4.24) into Eqs. (4.25a, c) gives that



1 2  1 
 2 1 b
Faxis ¼ pd Sfc þ q0 v Ne cos b þ 1  Ne sin b þ Ne sin 2b
2 2
4 2 p
ð4:26aÞ



d2 a 2 a  2 p 

2
Fvert ¼ Ne Sfc þ q0 v Ne cos b þ Ne  Ne sin b þ
2 b 2 b
1  Ne sin 2b
4 3 2
ð4:26bÞ

where d denotes the shank diameter of the projectile. The asymmetrical nosed
geometrical factors in the above equations can be expressed as
 
Ne ¼ NI þ NII =2; Nea ¼ NIa  NIIa ; Neb ¼ NIb  NIIb ð4:26cÞ

For the ogive-nosed projectile, Eq. (4.26c) can be obtained by substituting the
asymmetrical CRHs of side I and II (wI and wII ) into the following nosed geo-
metrical coefficients N  ; N a and N b :

8w  1
N ¼ ;
24w2


1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N a ¼ 4w2 cos1 1   ð2w  1Þ 4w  1;
2w

 ð4:26dÞ
p 1 2
N b ¼ w2  /0  sinð4/0 Þ  sinð2/0 Þ ;
2 12 3


1 1
/0 ¼ arcsin 1  ; w
2w 2

where the dimensionless nose parameter w ¼ s=d is the caliber-radius-head (CRH),

and s is the curvature radius of the ogival nose.
4.3 Structural Stability Analysis 131

4.3.2 Limit Striking Velocity

Based on the analyses of the test data in Forrestal et al. (1996) and Frew et al.
(1998), Zhao et al. (2010a) proposed that the variation of the nose factor N  of the
ogive-nose projectile during the penetration process satisfies
 
N  ¼ N0 þ k0 V02  v2 ð4:27Þ

where V0 and v denote the initial striking velocity and the instantaneous velocity of
the projectile, respectively. Denoting the initial CRH of the intact projectile as w0 ,
the initial nose factor N0 can be obtained by substituting w0 into the expression of
N  given in Eq. (4.26d). k 0 is the nose-blunting coefficient of the projectile in s2/
km2 with respect to the striking velocity V0 in km/s. In Sect. 4.2, the analytical
expression of nose-blunting coefficient is derived as

gN a ðw þ 4ÞM0
k 0 ¼ p0ffiffiffi 0 ð4:28Þ
6 3jQSMn w20

Shown in Sect. 4.2.5, Eq. (4.28) is validated by total 11 sets of different pene-
tration test data (the initial striking velocity reached as high as 1682 m/s).
From Eqs. (4.25a–c, 4.26a–d), it can be drawn that, for the strength analysis of
the projectile, the instantaneous velocity of the projectile vm when the drag forces
reach the maximum should be firstly confirmed. The axial and transverse maximal
resistances Faxis;max and Fvert;max are then obtained. Based on the work of Zhao et al.
(2012), for the most conservative estimation, we assume that the maximum
asymmetrical nose abrasion initiates at instant of maximum axial drag force.
Meanwhile, the discrepancy of the CRHs on both sides of the projectile nose and
the transverse drag force also reach its maximum. Assuming that vm of the
non-normal penetrating projectile with the asymmetrical nose abrasion is equal to
which of the normal penetrating projectile with the symmetrical nose abrasion
discussed in Zhao et al. (2012), where Ne ¼ N  ; a ¼ b ¼ 0: The axial drag force
Faxis is a quadratic function of v2 for the given initial striking velocity V0 according
to Eqs. (4.26a, 4.27). By substituting Eq. (4.27) into Eq. (4.26a) and differentiating
@Faxis =@ ðv2 Þ ¼ 0, we can get the velocity which corresponds to the maximum axial
drag force as

N0 þ k0 V02
v2m ¼ ð4:29Þ
2k0

Substituting above equation into Eq. (4.27) derives the maximal value of nosed
geometrical coefficients N  when the axial drag force reaches the maximum, which
is denoted as Nm and expressed as
132 4 Mass Abrasive Projectile Penetration

N0 þ k0 V02
Nm ¼ ð4:30Þ
2

Regarding the CRHs of the both sides as wI and wII when the drag forces reach
maximum, Zhao et al. (2012) obtained wI by assuming V0 ¼ 1 km/s in Eq. (4.30),
which may underestimate the abrasion of the projectile nose. The results are
inclined to more dangerous for the structural design of the projectile. The initial
striking velocity when the projectile begins to bend or break usually varies nearly
from 1 to 1.5 km/s in the existing high-speed penetration tests. Therefore, Nm is
derived by substituting V0 ¼ 1:5 km/s and Eq. (4.28) into Eq. (4.30) for conser-
vation. Thus, wI is obtained from Eqs. (4.26d, 4.30).
Regarding the asymmetrical nose abrasion, we may assume that the side II is
abraded more serious than side I, and thus, we have wI [ wII : Since the
time-dependent variation of CRHs of both sides of the projectile nose cannot be
measured based on the existing experimental technique, the discrepancy of the
asymmetrical nose abrasion wI [ wII is only discussed here. Therefore, for the
given initial striking velocity V0 ; we can obtain the expression of the maximum
drag forces Faxis;max and Fvert;max by substituting and, as well as Eq. (4.29) into
Eqs. (4.25a–c, 4.26a–d). In particular, wI  wII ¼ 0 corresponds to the symmetrical
nose abrasion.
Assuming that the projectile mass distributes uniformly along the shank, when
the maximal concentrated axial and transverse drag forces act on the projectile nose,
the expressions of equivalent axial force Nz and equivalent bending moment Mz at
the location z0 along the shank axis can be given as (Zhao et al. 2010b)

Nz ¼ Faxis;max ð1  z0 =LÞ; Mz ¼ Fvert;max z0 ð1  z0 =LÞ

2
ð4:31Þ

where z0 is the distance from the nose tip to the location of the concerned cross
section and L denotes the whole length of the projectile.
Pi and Huang (2007) formulated an elastic-plastic yielding function for the
projectile under the combination of compression and bending as follows:

jNz j jMz j
ne ¼ þ 1 ð4:32Þ
NY MY

where NY and MY are the yielding limits in the cases of individual compression and
bending. Equation (4.32) is a conservative criterion to identify the dynamic
structural stability of a slender projectile during penetration.
Regarding the cross section of a circular hollow tube projectile, it has
p 2  p  4 
NY ¼ d  d12 Yp ; MY ¼ d  d14 Yp ð4:33Þ
4 32d
4.3 Structural Stability Analysis 133

where Yp is the yielding strength of projectile material, d1 is the inner diameter of

the projectile tube. In particular, d1 = 0 is satisfied for the solid projectile.
Substituting Eqs. (4.25a–c, 4.26a–d, and 4.31) together into Eq. (4.32), we can
obtain the expression of yielding function as



z0 0 z0 2
ne ¼ A 1  þ Bz 1  1 ð4:34aÞ
L L

where



pd 2 1  1
A¼ Sfc þ q0 v2 Ne cos2 b þ 1  Ne sin2 b þ Neb sin 2b ;
4NY 2 p
8

9
> 1   2 1 b >
>
> p sin a Sf þ q v2
N 
cos 2
b þ 1  N 
sin b þ N sin 2b >
>
d2 < =
c 0 e
2 e
p e
B¼


8MY >> 2  p  >

>
>
: þ 2 Nea Sfc þ q0 v2 Neb cos2 b þ Nea  Neb sin2 b þ 1  Ne sin 2b > ;
3 2
ð4:34bÞ

After confirming the CRH values of the asymmetrical mass abrasion, the critical
location z0m of the cross section of projectile shank corresponding to the maximum
yielding function can be drawn by dne =dz ¼ 0 as well as d2 ne =dz2 \0 from
Eq. (4.34a), which derives
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
z0m 2 1 3A
¼  1þ ð4:35Þ
L 3 3 BL

Equation (4.35) indicates that the critical yielding cross section only locates
between the root of nose and the location of L/3 distance from the projectile nose
tip, which is consistent with the conclusions drawn by Pi and Huang (2007). For
conservative analysis, the critical yielding cross section is assumed at the root of
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
nose, which satisfies z0m =L ¼ w0  1=4=k2 , where k2 ¼ L=d is the
length-to-diameter ratio of the projectile.
So far, with the initial projectile nosed geometry and striking velocity V0 given,
substituting Eqs. (4.29, 4.33, 4.34b, 4.35) into Eq. (4.34a), we can obtain the
yielding function of the critical cross section as follows
 
Sfc q0 N0 þ k0 V02
ne ¼ D þ ðE þ F Þ 1 ð4:36aÞ
Yp 2k0 Yp

where
134 4 Mass Abrasive Projectile Penetration




z0 z0 z0 2 1 Na
D ¼ C1 1  m þ C2 m k2 1  m sin a þ e ;
L L L 2 p



z 0
1   N b sin 2b
E ¼ C1 1  m Ne cos2 b þ 1  Ne sin2 b þ e ;
L 2 p
2
3
sin a  1 
 2 1 b

 N cos 2
b þ 1  N sin b þ N sin 2b
z0 2 6 p e 7
e e
z0 6 2 2 7
F ¼ C2 k 2 m 1  m 6
 7;
L L 4 1 2 a  2 p 
 5
þ Ne cos b þ Ne  Ne sin b þ
b 2 b
1  Ne sin 2b
p 3 2
1 1
C1 ¼ ; C2 ¼
4k1  4k21 2k41 þ 4k31  3k21 þ k1
ð4:36bÞ

in which k1 ¼ ht =d is the dimensionless thickness of projectile cartridge, and ht is

the thickness of projectile cartridge. In particular, k1 ¼ 0:5 corresponds to a solid
projectile.
When Eq. (4.36a) satisfies ne ¼ 0;, we can obtain the expression of the limit
striking velocity (LSV) of the projectile as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  ffi
2 Yp  DSfc N0
V0;lim ¼  ð4:37Þ
ðE þ F Þq0 k

Equation (4.37) gives the critical velocity that no yielding failure occurs, and
thus, the projectile maintains its integral structural stability. It should be pointed out
that the above equation takes the initial oblique and attacking angles as well as the
asymmetrical nose abrasion into account.

4.3.3 Normal Penetration

4.3.3.1 Without Nose Abrasion

Regarding the projectile normal penetrating into concrete targets without any nose
abrasion, it has a ¼ b = 0°, k0 ¼ 0 and Ne ¼ N0 , only the axial drag force acts on
the projectile. Thus, we can obtain that the dimensionless thickness of projectile
cartridge should satisfy the following expression when ne  0 in Eqs. (4.36a, b).
4.3 Structural Stability Analysis 135

 

Sfc þ N0 q0 V02 z0
k1  k21  1 m ð4:38Þ
4Yp L

Limited by the backfill ratio of the internal ammunition and other electric
components in the projectile, the dimensionless thickness of projectile cartridge
usually satisfies 0:08  k1  0:2. Thus, the second-order quantity k21 is neglected,
and Eq. (4.38) is simplified as
 

Sfc þ N0 q0 V02 z0
k1  1 m ð4:39Þ
4Yp L

Furthermore, assuming the critical cross section located at the nose tip and, thus,
z0m = 0, and then, Eq. (4.39) can be further simplified as

Sfc þ N0 qV02

k1  ð4:40Þ
4Yp

Equations (4.39) and (4.40) are consistent with the conclusions of Pi and Huang
(2007) and Chen (2005) for the normal penetration without considering the pro-
jectile nose abrasion (Eq. (24) in Pi and Huang (2007) and Eq. (14) in Chen
(2005)). Therefore, the proposed approach can be regressed to the normal pene-
tration without projectile nose abrasion.
Based on Eq. (4.39), when the projectile cartridge is fixed, the LSV of the
normal penetrating projectile without mass abrasion is
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 4Yp k1
u    c
Sf
V0;lim ¼t ð4:41Þ
0 q0
 z0m N
N q 1 0 0 L

4.3.3.2 With Symmetrical Nose Abrasion

Regarding the projectile normal penetrating into the concrete target with symmet-
rical nose abrasion, it has a ¼ b = 0°, wI ¼ wII and Ne ¼ Nm ; and only the axial
1
drag force acts on the projectile. Fvert ¼ 0 and Fvert
2
¼ 0 can be drawn from
Eqs. (4.25a–c, 4.26a–d). Substituting Eq. (4.30) into Eqs. (4.36a, b, 4.37), the LSV
of the normal penetrating projectile with symmetrical nose abrasion is given as
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u N0
u Yp
 
Sfc
2
V0;lim ¼ 2t   ð4:42Þ
k 0 q C1 1  m
z0 k 0 q0 k 0
0 L
136 4 Mass Abrasive Projectile Penetration

In particular, for the solid projectile, which has k1 ¼ 1=2 and C1 ¼ 1, Eq. (4.42)
is simplified as
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u N
u Y
 p 0   0 c  00
Sf
2
V0;lim ¼ 2t ð4:43Þ
k0 q 1  m
z k q0 k
0 L

Equations (4.42, 4.43) are consistent with the conclusions of Zhao et al. (2011,
2012) for the normal penetration with symmetrical nose abrasion (Eq. (23) in Zhao
et al. (2012) and Eqs. (14, 15) in Zhao et al. (2011)). Therefore, the above-proposed
approach can be regressed to the normal penetration with symmetrical nose
abrasion.
Figure 4.11 shows the dependence curves of LSV with the strength of concrete
targets, and the dimensionless thickness of projectile cartridges based on Eq. (4.42)
for the normal penetration with symmetrical nose abrasion. The calculation
parameters are chosen from Forrestal et al. (1996), i.e., q0 ¼ 2300 kg/m3 , HM = 7,
Yp = 1481 MPa, d = 20.3 mm, k2 ¼ 10 and w0 ¼ 3. The detailed parameters are
listed in Table 4.1. It can be drawn that, for a given dimensionless thickness of
projectile cartridge k1 , LSV of the projectile decreases with increasing the concrete
strength, and this decreasing tendency is aggravated with the reduction of k1 . When
the concrete strength keeps constant, LSV of the projectile increases nonlinearly
with strengthening the projectile cartridge. The safe dimensionless thickness of
projectile cartridges should satisfy k1  0:11 for the projectile penetrating into
concrete targets with the conditions of V0 ¼ 1 km/s and fc ¼ 40 MPa.

4.3.3.3 With Asymmetrical Nose Abrasion

Regarding the projectile normal penetrating into concrete targets with asymmetrical
nose abrasion, it has a ¼ b ¼ 0°. Both the axial and transverse drag forces act on
the projectile. We can obtain the expression of LSV from Eq. (4.37) as

   2

z0 z0 N a z0
2 Yp  C1 1  Lm þ C2 Lm pe k2 1  Lm Sfc
N0
2
V0;lim ¼    
 ð4:44Þ
z0 N b z0 z0
2 k0
q0 C1 Ne 1  Lm þ C2 k2 pe Lm 1  Lm

Equation (4.44) is consistent with the conclusions of Zhao et al. (2011, 2012) for
the normal penetration with asymmetrical nose abrasion (Eq. (22) in Zhao et al.
(2012) and Eq. (19) in Zhao et al. (2011)). Therefore, the above-proposed approach
can also be regressed to the normal penetration with asymmetrical nose abrasion.
4.3 Structural Stability Analysis 137

(a) 2200 (b) 2000

No abrasion
2000 1800 ΨI-ΨII =0
ΨI-ΨII =0.1
1800 1600 ΨI-ΨII =0.2

V0,lim (m/s)
V0,lim (m/s)

1600
1400
1400
1200
1200
No abrasion 1000
1000 ΨI-ΨII=0
800 ΨI-ΨII=0.1 800
ΨI-ΨII=0.2
600 600
30 40 50 60 70 80 90 100 0.10 0.12 0.14 0.16 0.18 0.20
fc (MPa) λ1

Fig. 4.12 Dependence curves of LSV with the discrepancy of asymmetrical nose abrasion as well
as a strength of concrete targets, and b dimensionless thickness of projectile cartridges

Figure 4.12 gives the dependence curves of LSV with the discrepancy of the
asymmetrical nose abrasion, the strength of concrete targets and the dimensionless
thickness of projectile cartridge based on Eq. (4.44), respectively. The calculation
parameters are the same as which used above, except that k1 ¼ 0:2 in Fig. 4.12a
and fc ¼ 40MPa in Fig. 4.12b. The solid lines in Fig. 4.12 are the LSVs of the
normal penetrating projectile without mass abrasion given in Eq. (4.41). It is clear
that, the discrepancy of asymmetrical nose abrasion induced by the inhomogeneous
concrete targets has great influence on the LSV of the projectile. Even if the
dimensionless thickness of projectile cartridge or concrete strength is unchanged,
the LSV of the projectile still decreases tremendously with minor rise of the dis-
crepancy of asymmetrical nose abrasion. In particular, when the discrepancy of
asymmetrical nose abrasion is Dw ¼ wI  wII ¼ 0:1, the safe dimensionless
thickness of projectile cartridge should satisfy k1  0:18 for the projectile pene-
trating into concrete targets with the conditions of V0 ¼ 1 km/s and fc ¼ 40 MPa.
Compared with Fig. 4.11, much thicker projectile cartridge is needed in the case of
the asymmetrical nose abrasion.
The asymmetrical nose abrasion depends on the projectile alloy material and
concrete strength as well as the hardness, characteristic dimension, and the distri-
butions of the coarse aggregate, etc. Therefore, for protective structures, enhancing
the concrete strength and hardness of the coarse aggregate is the effective approach
to promote its antipenetration ability (such as the corundum block or rock-rubble
bursting layer and high-strength granite coarse aggregate). It also indicates that, for
the design of the AEPW, using high-strength alloy (such as tungsten and depleted
uranium) is helpful to reduce the symmetrical or asymmetrical nose abrasion, to
guarantee the integrity of the projectile and enhance the penetration efficiency.
138 4 Mass Abrasive Projectile Penetration

4.3.4 Oblique Penetration

4.3.4.1 Without Nose Abrasion

When the projectile oblique penetrates into concrete targets without nose abrasion,
which has b ¼ 0°, k0 ¼ 0 and Ne ¼ N0 : Both the axial and transverse drag forces
act on the projectile. Thus, the initial striking oblique angle of the projectile should
satisfy the following expression when ne \0 in Eqs. (4.36a, b),
 
Yp z0
Sfc þ N0 q0 V02
 C1 1  Lm
sin a\  2 ð4:45Þ
z0m z0m
1
2 C 2 k 2 L 1  L

If k21 and even higher order quantity of k1 are neglected, it has C2 ¼ 4C1 ¼ k1
1 ,
and Eq. (4.45) is simplified as
 
Yp z0
 4k11 1  Lm
Sfc þ N0 q0 V02
sin a\  2 ð4:46Þ
k2 z0m z0m
2k1 L 1  L

Conversely, if the initial striking oblique angle of the projectile is fixed, the safe
dimensionless thickness of projectile cartridge should satisfy
"

#
k2 z0m z0m 2 1 z0m Sfc þ N0 q0 V02
k1 [ 1 sin a þ 1 ð4:47Þ
2 L L 4 L Yp

If the oblique angle satisfies a ¼ 0°, Eq. (4.47) is same as Eq. (4.39).
Equations (4.45, 4.47) are consistent with the conclusions of Pi and Huang (2007)
for the oblique penetration without considering the mass abrasion (Eqs. (22–23) in
Pi and Huang (2007)).
If pure bending analysis is conducted, it definitely has z0m =L ¼ 1=3; thus, the
initial striking oblique angle of the projectile should satisfy the following expres-
sion when Mz  MY from Eqs. (4.31–4.34a, b).

27Yp
sin a    ð4:48Þ
2C2 k2 Sfc þ N0 q0 V02

For the solid projectile, i.e., k1 ¼ 0:5, it can be deduced from the above equation
that

27Yp
sin a    ð4:49Þ
16k2 Sfc þ N0 qV02
4.3 Structural Stability Analysis 139

Fig. 4.13 Recovered projectiles with a a ¼ 0 , b a ¼ 20 , c a ¼ 30 (Chen et al. 2006)

Equations (4.48, 4.49) agree well with Chen (2005) for the pure bending anal-
ysis of the oblique penetration without considering the mass abrasion (Eqs. (23–25)
in Chen (2005)). Therefore, the proposed approach is regressed to the oblique
penetration without projectile nose abrasion.
Chen et al. (2006) conducted the oblique penetration tests (a ¼ 0°, 20°, 30°) on
the concrete targets with the different cartridge thicknesses (k1 ¼ 0:1, 0.15) and
length–diameter ratios (k2 = 6, 8, 10) of projectiles. Some soft recovered projectiles
from the penetration tests are shown in Fig. 4.13. The mass losses of the projectiles
varied from 2% to 4%, and thus, the mass losses are neglected in the present
analysis. Figure 4.14 shows the comparisons of prediction curves of the critical
oblique angles based on Eq. (4.46) and the test results. Obviously, the theoretical
prediction agrees well with the test data, and thus, the validation of Eq. (4.46) on
the prediction of critical oblique angle of non-abrasive projectile under the com-
bined compression and bending is verified.

4.3.4.2 With Asymmetrical Nose Abrasion

When the projectile oblique penetrates into concrete targets with asymmetrical nose
abrasion, it has b ¼ 0°. Equation (4.37) is simplified as
(
" 

# )
z0m z0m z0m 2 1 Nea
2
¼ 2 Yp  C1 1 
V0;lim þ C 2 k2 1  sin a þ Sfc
L L L 2 p
(


)1
1  z0m z0m z0m 2 sin a  Neb N
q0 C1 Ne 1  þ C 2 k2 1 Ne þ  00
L L L 2 p k
ð4:50Þ

Equation (4.50) is the same as Eq. (23) in Zhao et al. (2011) for the oblique
penetrating projectile with asymmetrical nose abrasion. Therefore, the proposed
approach is regressed to the oblique penetration scenario with asymmetrical nose
abrasion.
140 4 Mass Abrasive Projectile Penetration

(a) 35 (b) 80
Eq. (4.46)
30 Intact
60
25

20
40
15

10 Eq. (4.46) 20
Intact
5 Little bending
0 0
200 400 600 800 1000 200 400 600 800 1000
V0 (m/s) V0 (m/s)

(c) 35 (d) 50
Eq. (4.46)
Intact
45
30
Little bending 40
25 Bending 35
20 30
25
15
20
10 15
Eq. (4.46)
10 Intact
5
5 Little bending
0 0
200 400 600 800 1000 200 400 600 800 1000
V0 (m/s) V0 (m/s)

(e) 35 (f) 40
Eq. (4.46)
30 Intact 35
Little bending
25 Bending 30
25
20
20
15
15
10 Eq. (4.46)
10 Intact
5 Little bending
5
Bending
0 0
200 400 600 800 1000 200 400 600 800 1000
V0 (m/s) V0 (m/s)

Fig. 4.14 Critical oblique angles for bending of projectiles and tests results a k1 ¼ 0:1; k2 ¼ 6,
b k1 ¼ 0:15; k2 ¼ 6, c k1 ¼ 0:1; k2 ¼ 8, d k1 ¼ 0:15; k2 ¼ 8, e k1 ¼ 0:1; k2 ¼ 10,
f k1 ¼ 0:15; k2 ¼ 10

Figure 4.15 shows the dependence curves of LSV based on Eq. (4.50) with the
concrete strength, discrepancy of asymmetrical nose abrasion, and the projectile
cartridge thickness as well as the oblique angle. The calculation parameters are the
same as which used above, except that k1 ¼ 0:2 and wI  wII ¼ 0:1 in Fig. 4.15a,
k1 ¼ 0:2 and a ¼ 2° in Fig. 4.15b, fc ¼ 40 MPa and a ¼ 2° in Fig. 4.15c.
4.3 Structural Stability Analysis 141

(a) 1500
(b) 1400
α=0 fc=30MPa
1400 1300
α=1 fc=40MPa
1300 1200 fc=50MPa
α=2

V0,lim (m/s)
V0,lim (m/s)

1200 α=3 1100

1100 α=4
1000
1000
900
900
800 800

700 700
30 40 50 60 70 0.00 0.05 0.10 0.15 0.20 0.25 0.30
fc (MPa) ΨI -ΨII

(c) 1200
1100
ΨI-ΨII =0
ΨI-ΨII =0.1
1000 ΨI-ΨII =0.2
V0,lim (m/s)

900
800
700
600
500

0.10 0.12 0.14 0.16 0.18 0.20

λ1

Fig. 4.15 Dependence curves of LSV with the oblique angle, discrepancy of asymmetrical nose
abrasion, strength of concrete targets, and the dimensionless thickness of projectile cartridges

It is shown in Fig. 4.15 that LSV of the projectile decreases sharply with
increasing the initial striking oblique angle, concrete strength, and the discrepancy
of asymmetrical nose abrasion. In addition, LSV rises with the increase of the
dimensionless thickness of projectile cartridge. From Fig. 4.15a, if we keep k1 , Dw
and a unchangeable, LSV of the projectile decreases linearly with the increasing of
concrete strength, and the slopes under different oblique angles are almost
unchanged.
From Fig. 4.15, by comparison with the oblique penetration discussed in
Sect. 4.3.4.1, much thicker projectile cartridge is needed when the asymmetrical
nose abrasion is considered. For example, when the discrepancy of asymmetrical
nose abrasion is Dw ¼ wI  wII ¼ 0:1, and the initial striking oblique angle a ¼ 2°,
the safe dimensionless thickness of projectile cartridge should satisfy k1  0:2 for
the projectile penetrating into concrete targets with V0 ¼ 1 km/s and fc ¼ 40 MPa.
Conversely, aiming to guarantee the structural integrity of the projectile, the initial
striking oblique angle should be controlled rigorously. For example, under the
conditions of k1 ¼ 0:2 and wI  wII ¼ 0:1, the initial striking oblique angle should
142 4 Mass Abrasive Projectile Penetration

Fig. 4.16 Dependence 2000

curves of LSV with the a =0°
oblique angle and length– 1800 a =5°
diameter ratio of projectile a =10°
1600

V0,lim (m/s)
1400

1200

1000

800

600
6 7 8 9 10 11 12
λ2

satisfy a  2° for the projectile penetrating into concrete targets with the V0 ¼
1 km/s and fc ¼ 40 MPa from Fig. 4.15a.
It must be pointed that, Zhao et al. (2011) derived the initial striking oblique
angle should satisfy a  4°, and the differences lie in the conservative analysis for
wI from Eq. (4.30) and the deviation of the nose-blunting coefficient predicted by
Eq. (4.28). Comparatively, the nose-blunting coefficient used in Zhao et al.
(2011) is empirically fitted by test data.
Figure 4.16 shows the dependence curves of LSV predicted by Eq. (4.50) with
the oblique angle and length–diameter ratio of projectile considering the asym-
metrical nose abrasion. The calculation parameters are the same as which used in
Sect. 4.3.3.2, except that fc ¼ 30 MPa, wI  wII ¼ 0:1 and k1 ¼ 0:2.
As shown in Fig. 4.16, when the asymmetrical projectile nose abrasion is
considered, the LSV of the projectile decreases with the rise of the length–diameter
ratio for different oblique angles. Similarly, when the length–diameter ratio of the
projectile is fixed, the LSV of the projectile decreases with the increasing of oblique
angle. All these are consistent with the conclusions drawn from the penetration tests
by Chen et al. (2006).

4.3.5 Penetration with Attacking Angle and Asymmetrical

Abrasion

Affected by the launching condition of the projectile, another important factor, the
initial attacking angle b exists inevitably in the penetration. When we only consider
the attacking angle and asymmetrical nose abrasion, i.e., a ¼ 0°, Eq. (4.37) can be
simplified as
4.3 Structural Stability Analysis 143

(a) (b)
1500 1150
1400 1100 ΨI-ΨII=0
β=0
β=1 1050 ΨI-ΨII=0.1
1300
β=2 1000 ΨI-ΨII=0.2

V0,lim (m/s)
V0,lim (m/s)

1200
950
1100
900
1000
850
900
800
800 750
700 700
600 650
500 600
30 35 40 45 50 55 60 30 35 40 45 50 55 60
fc (MPa) fc (MPa)

(c) 1200
ΨI-ΨII=0
1100 ΨI-ΨII=0.1
ΨI-ΨII=0.2
V0,lim (m/s)

1000

900

800

700

600
0.10 0.12 0.14 0.16 0.18 0.20 0.22
λ1

Fig. 4.17 Dependence curves of LSV with the attacking angle, discrepancy of asymmetrical nose
abrasion, strength of concrete targets, and the dimensionless thickness of projectile cartridges

8

 91
> z0m  1 
 2 Neb sin 2b >
>
> C 1  N cos 2
b þ 1  N sin b þ þ >
>
2< =
1
L e
2 e
p
2
V0;lim ¼


>
q0 > 1 z0 z0 2
2  p  >
>
>
: C2 k2 m 1  m Neb cos2 b þ Nea  Neb sin2 b þ 1  Ne sin 2b >;
pL L 3 2
( "

 # )
z0 z0 N a z0 2 N
 Y p  C1 1  m þ C2 m e k2 1  m Sfc  00
L L p L k

ð4:51Þ

Figure 4.17 shows the dependence curves of LSV based on Eq. (4.51) with the
attacking angle, discrepancy of asymmetrical nose abrasion, and strength of con-
crete targets as well as the dimensionless thickness of projectile cartridge. The
calculation parameters are the same as which used in Sect. 4.3.3.2, except that
k1 ¼ 0:2 and wI  wII ¼ 0:1 in Fig. 4.17a, k1 ¼ 0:2 and b ¼ 1° in Fig. 4.17b,
fc ¼ 40MPa and b ¼ 1° in Fig. 4.17c.
It is shown in Fig. 4.17 that LSV of the projectile decreases sharply with the
increases of the initial striking attacking angle, concrete strength, and the
144 4 Mass Abrasive Projectile Penetration

discrepancy of asymmetrical nose abrasion, while it rises with strengthening the

projectile cartridge thickness. It should be pointed out that even minor increase of
the attacking angle can tremendously decrease the LSV of the high-speed pene-
trating penetrator; thus, the initial attacking angle of the AEPW should be con-
trolled rigorously. The conservative variation range of the initial striking attacking
angle is 0  b  1°. Furthermore, from the Fig. 4.17a, if we keep k1 , wI  wII , and
b unchangeable, LSV of the projectile decreases almost linearly with the rising of
concrete strength gradually.
It can be found from Eq. (4.37) that the LSV of the projectile is dependent on the
compressive strength of concrete, the initial oblique and attacking angles, the dis-
crepancy of asymmetrical mass abrasion, the dimensionless cartridge thickness,
length–diameter ratio, and initial CRH of the projectile nose. Limited by the backfill
ratio of the internal ammunition and other electric components, the dimensionless
thickness of projectile cartridge is usually confined within a narrow range. The
parametrical analysis of the other factors on the LSV is conducted as follows.
Figure 4.18 shows the comparisons among the amplifications of LSV and which
of the influential parameters fc, wI  wII , a, b, k2 , and w0 . The reference parameters
are fc ¼ 30 MPa, Yp = 1481 MPa, HM = 7, q0 ¼ 2300 kg/m3 , d = 20.3 mm,
w0 ¼ 3, k1 ¼ 0:2, k2 ¼ 6, a ¼ b ¼ 1°, wI  wII ¼ 0:1, respectively. The corre-
sponding variation ranges of above parameters are 1  a  7 , 30 MPa 
fc  90 MPa,1  b  3 , 6  k2  12, 0:1  wI  wII  0:5 and 3  w0  8,
respectively.
Figure 4.18 indicates that when the other parameters are fixed, the LSV
decreases with the increase of the discrepancy of asymmetrical mass abrasion,
length–diameter ratio, and concrete strength as well as the oblique and attacking
angles. In addition, the LSV rises with the increase of the CRH. Furthermore, the
higher influential degree of length–diameter ratio, attacking angle, concrete
strength, and the CRH are almost the same, while the less influential degree of the

Fig. 4.18 Amplifications of 1.1

LSV versus the corresponding fc
influential parameter 1.0 ΨI -ΨII
Amplifications of V0,lim

0.9 a
β
0.8 λ2
0.7 1/Ψ0

0.6

0.5

0.4

0.3
1 2 3 4 5 6 7 8
Amplifications of influential parameters
4.3 Structural Stability Analysis 145

discrepancy of asymmetrical mass abrasion and the oblique angle are almost the
same.
Therefore, the proper length–diameter ratio and CRH, high-strength alloy, and
the rigorous control of the initial impact gesture are the key issues for the design of
a high-speed AEPW. In the contrast, increasing the concrete strength is an efficient
and direct approach to enhance the defense capability of protective structures.

4.3.6 Validation

The structural destruction of the projectile (e.g., bending and fracture) occurred in
the high-speed penetration tests conducted by Forrestal et al. (1996), He et al.
(2010a), and Wu et al. (2012b). Some projectiles recovered in the above penetration
tests and the corresponding initial striking velocities are given in Fig. 4.19.
In this section, the proposed Eq. (4.37) for predicting LSV is verified by the
above-cited tests.
Chen and Li (2002) proposed a formula for predicting the DOP of rigid pro-
jectile as


2M0 q0 N0 V02
hpen ¼ ln 1 þ ð4:52Þ
pd 2 q0 N0 Sfc

Zhao et al. (2010a) proposed the depth of penetration of projectile with con-
sidering the mass loss and nose blunting as

Fig. 4.19 Bending and fracture of the projectiles recovered in high-speed penetration tests a He
et al. (2010a), b Wu et al. (2012b)
146 4 Mass Abrasive Projectile Penetration

(a) (b)
1.6 2.00
Forrestal et al. (1996) Forrestal et al. (1996)
1.4 1.75
1.2 1.50
1.0 1.25

hpen (m)
V+0,lim=1478m/s
hpen (m)

+
V 0,lim=1771m/s
0.8 V0,lim=1386m/s 1.00
0.6 0.75
ΔΨ =0.1
0.4 ΔΨ =1 projectile 0.50
bented
0.2 projectile fractured 0.25 V0,lim=1198m/s projectile
at the bottom fractured
0.0 0.00
0 400 800 1200 1600 2000 0 200 400 600 800 1000 1200 1400 1600
(c) V0 (m/s) (d) V0 (m/s)
3.0 2.5
He et al. (2010) He et al. (2010)
2.5 2.0

2.0 +
+
V 0,lim=2060m/s
V 0,lim=2091m/s 1.5
hpen (m)
hpen (m)

1.5
1.0
1.0
projectile projectile
bented 0.5 bented
0.5
V0,lim=1165m/s ΔΨ =0.9 V0,lim=1386m/s ΔΨ =0.6
0.0 0.0
0 250 500 750 1000 1250 1500 1750 2000 0 400 800 1200 1600 2000
V0 (m/s)
V0 (m/s) (f)
(e)
3.5 2.5
He et al. (2010) He et al. (2010)
3.0
2.0
2.5
1.5
hpen (m)
hpen (m)

2.0 +
V 0,lim=2034m/s
+
V 0,lim=2082m/s
1.5
1.0
1.0 projectile
projectile bented
0.5 bented 0.5
V0,lim=1376m/s ΔΨ =0.5 V0,lim=1287m/s ΔΨ =0.35
0.0 0.0
0 400 800 1200 1600 2000
0 500 1000 1500 2000
V0 (m/s)
V0 (m/s)

Fig. 4.20 Comparisons between the predicted DOP and LSV with the test data

8    
9
>
> q0 N0 þ k0 V02 q0 N0  k 0 V02 > >
>
<  arctan h
arctan h >
=
M0 v v
hpen ¼

2pk0 d 2 q0 v >
>    q NV 2 > >
:  8k 0 q0 þ 2q0 C N0  k 0 V02  Cv ln 1 þ 0 0 0 >
> ;
Sfc
ð4:53Þ
4.3 Structural Stability Analysis 147

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  2ffi
0 0
where v ¼ 4Sfc q0 k þ q0 N0 þ k V0 . The mass loss coefficient C is given in
2 2

Eq. (4.13). The nose-blunting coefficient k 0 is determined by Eq. (4.21).

Figure 4.20 shows the comparison of the predicted DOP and LSV with the test
data (Figs. 4.20a–f corresponds to the normal penetration Tests 2–3 and Tests 7–10
in Table 4.1, respectively), where the solid and dashed curves are the DOP of the
projectiles predicted by Eqs. (4.52, 4.53), respectively. Compared with the rigid
projectile prediction Eq. (4.52), in the case of higher striking velocity, the DOP
predicted by Eq. (4.53) agrees better with the tests results. It further confirms that
the proposed mass loss and nose-blunting coefficients of the ogive-nose projectile
are applicable.
More importantly, due to the effect of the mass loss and symmetrical nose
blunting, there may exist a theoretical critical striking velocity at which the pro-
þ þ
jectile achieves its maximum DOP (denoted as V0;lim given in Fig. 4.20). V0;lim can
 2 þ
be obtained by differentiating Eq. (4.53) dhpen =d V0 ¼ 0. The expression of V0;lim
was also given by Zhao et al. (2010a).
Actually, due to the random distribution of the coarse aggregate in the concrete
target, the asymmetrical abrasion occurs inevitably on the projectile nose during
high-speed penetration. When the initial penetration velocity of projectile exceeds
than which predicted by Eq. (4.37), depending on the discrepancy of asymmetrical
nose abrasion, the projectile might yield under the combining action of the axial
resistance and transverse forces. Figure 4.20 also shows the predicted LSV by
Eq. (4.37) and the corresponding discrepancy of the asymmetrical mass abrasion
Dw ¼ wI  wII . It can be drawn that, at high striking velocity ranging from 1 to
þ
1.5 km/s, before the projectile reach its theoretical critical striking velocity V0;lim ,
the structural destruction failure already occurs due to the asymmetrical nose
abrasion as well as initial oblique and attacking angles, which agrees well with the
measurements in the tests.
Wu et al. (2012b) and Wang et al. (2010a) conducted the penetration tests on
concrete targets with limestone aggregate (HM = 3), and the projectiles were
machined by the alloy 30GrMnSiNi2A with the yielding strength of
Yp = 1413 MPa and the dimensionless thickness of projectile cartridge of
k1 = 0.17. The other parameters can be referred to the Tests 11 in Table 4.1. Some
recovered projectiles after the tests with different striking velocities are shown in
Fig. 4.19b. During the test, the projectile maintained its integrity when the initial
striking velocity and the attacking angle are V0 ¼ 1031 m/s and b ¼ 0 . And the
bending or fracture of the projectile occurred when V0 ¼ 1263 m/s and b ¼ 0:17 ,
V0 ¼ 1308 m/s and b ¼ 0:28 , V0 ¼ 1425 m/s and b ¼ 0:19 , respectively. Based
on the Eq. (4.37), the predicted LSVs for the above three tests are 1215 m/s,
1116 m/s, and 1196 m/s under the discrepancy of asymmetrical nose abrasion
Dw ¼ wI  wII ¼ 0:3, respectively.
It should be noted that the actual time-dependent variation of asymmetrical nose
abrasion could not be measured as the absence of reliable measuring techniques in
the high-speed penetration tests on concrete targets. In this section, by assuming the
148 4 Mass Abrasive Projectile Penetration

discrepancy of asymmetrical nose abrasion Dw at the instant when the cross section
of the root of the projectile nose is yielding, the predicting approach for the LSV is
proposed and validated. The experimental phenomenon that the projectiles lose its
þ
structural stability before reaching the theoretical critical striking velocity V0;lim is
verified. Besides, the high-speed (1:0 km/s  V0  1:5 km/s) penetration tests on
concrete targets with the larger initial oblique and attacking angles are very limited
due to the incapable launching capacity as well as the measurement accuracy. Thus,
the verifications of the proposed formula Eq. (4.37) by the non-normal penetration
tests with larger initial oblique and attacking angles need further efforts in the
future.

4.4 Stability Analyses of Terminal Ballistic Trajectory

As stated previously in this chapter, with increasing the penetration velocity, due to
the higher pressure and temperature occurred at the interface between the projectile
and the target, as well as the friction and indentation of the randomly distributed
coarse aggregate in the concrete target, the serious symmetrical or asymmetric nose
abrasions may occur inevitably. Under the combined action of the asymmetric nose
abrasion, as well as the oblique and attacking angles, etc., the projectile experiences
unbalanced resistances and the terminal ballistic trajectory deviates obviously,
shown in Figs. 4.1c and 4.21.
Investigations on the terminal ballistic trajectory stability are most helpful for the
structural design of high-speed penetrators and protective structures. However, at

Fig. 4.21 Projectile penetration tests on a finite thickness concrete target (http://club.china.com/
data/thread/1013/2729/27/96/4_1.html, 2011.08.03), and b semi-infinite thickness concrete target
(Erengil and Cargile 2002)
4.4 Stability Analyses of Terminal Ballistic Trajectory 149

least three challenges should be conquered: (i) the abrasion law of the projectile
during the high-speed penetration process; (ii) the influence of the target surface on
the drag force which acts on the non-normal penetrating projectile (oblique and
attacking angles exist); and (iii) the prediction approach of the terminal ballistic
trajectory of projectile high-speed penetrating into the concrete target with con-
sidering the above effects.
Regarding the first aspect, in Sect. 4.2, the analytical models for the mass loss
and nose blunting of projectiles are established and validated by large amounts of
test data.
Regarding the second aspect, for the non-normal striking projectile with the
initial oblique and attacking angles, asymmetric drag force would act on the pro-
jectile induced by the front surface of the target. Such is the free-surface effect,
which should be concerned in the structural design of projectiles. Considering the
target is an incompressible and damaged Mohr–Coulomb material, Macek and
Duffey (2000) suggested a spherical cavity expansion forcing function to account
for the free-surface effects during projectile penetration into rock and frozen soil
targets. Introducing the inertia effects into the free-surface effect model following
Longcope et al. (1999), Warren and Poormon (2001) and Warren et al. (2004b)
introduced a decaying function to account for the asymmetric drag forces based on
the dynamic spherical cavity expansion theory with considering incompressible
perfectly plastic and Mohr–Coulomb target material, respectively. The predicted
terminal ballistic trajectories of the projectiles by the above free-surface effect
model agree well with the oblique penetration tests on aluminum (Warren and
Poormon 2001) and limestone (Warren et al. 2004b) targets, respectively. The
existing investigations indicate that the free-surface effect extends to larger distance
as the instantaneous penetrating velocity increasing, which has significant effect on
the terminal ballistic performance of projectiles, especially for the non-normal
high-speed penetration into concrete-like brittle targets.
Regarding the last aspect, by utilizing the differential area force law (DAFL)
method proposed by AVCO Corporation in the early 1970s (Heuze 1990), the
researchers from US Army Waterways Experiment Station and Sandia National
Laboratory developed 2D (PENCO2D) and 3D (PENCRV3D, PRONTO3D) pro-
grams to predict the terminal ballistic trajectory, where the projectile was modeled
by an explicit, transient dynamic, finite element code, and the target was repre-
sented by an analytical forcing function obtained from the dynamic spherical cavity
expansion theory (Warren and Poormon 2001; Warren et al. 2004b; Bernard and
Creighton 1978, 1979; Adley and Moxley 1996; Taylor and Flanagan 1989;
Attaway et al. 1998; Warren and Tabbara 1997). Youch (2006) proposed an inte-
grated force law (IFL) method to predict the terminal ballistic trajectory of MK
series bombs. However, the expression of the forcing function acted on the pro-
jectile surface was complicated and the related coefficients were empirically
obtained by fitting the test data. Li and Flores-Johnson (2011) discussed the ballistic
trajectory and the reverse velocity (the critical velocity that the J-shape trajectory
formed during the penetration) of the projectile high-speed penetrating into soil
targets, by combining the semiempirical resistance of a target with the finite element
150 4 Mass Abrasive Projectile Penetration

modeling of a projectile. Comparing with FE simulation, the discretizing of the

target and the contact algorithm between the target and the projectile are not needed
in both DAFL and IFL methods. Therefore, the computing time and the memory
requirements are reduced tremendously. The existing investigations are mainly
concerned with the homogeneous targets (e.g., ductile metallic or soil targets) and
the mass abrasions of the projectiles are usually neglected in the predictions of the
terminal ballistic trajectory.
In this section, based on the proposed model for the mass loss and nose blunting
of the ogive-nosed projectile established in Sect. 4.2, DAFL method is employed to
construct the finite differential approach for predicting the terminal ballistic tra-
jectory of a projectile high-speed penetrating into concrete targets. The code cor-
responding to the above approach is named as PENTRA2D, which accounts for the
free-surface effect of the target attributed by oblique and attacking angles, as well as
the asymmetric nose abrasions. The terminal ballistic trajectories of the projectile
non-normal penetrating into the concrete target are analyzed and compared with the
available test data. In addition, the related parametric influential analyses are
conducted.

4.4.1 Drag Forces

Similar to Sect. 4.3.1, we assume that the plane of oblique angle coincides with the
plane of attacking angle for convenience of discussion. The axial longitudinal
section of a projectile is shown in Fig. 4.22a. Two coordinate systems are used and
depicted, where the XZ-coordinate is fixed in the target, and the xz-coordinate is
fixed on the projectile axis. The origin of xz-coordinate locates at the mass central
CG and the z-axis is along the symmetrical axis of the projectile. The instantaneous
rotation angle of the projectile is denoted as hy illustrated in Fig. 4.22a, and the
positive value is defined as counterclockwise. The instantaneous velocity of the
projectile CG is denoted by v, the initial oblique angle a is ranged from 0° to 90°,

G
v x,C
x,
(a) d
O
X

θy
z,
v z,

Z
CG

x
a (b) (c)
CG

β
v
L=

r A
r x
λ2

x II
d

r Φ x I
L3

A
=λ

I
3
d

η
L4

II
=λ

z A
4
d

z z y

Fig. 4.22 Schemes of non-normal projectile penetration

4.4 Stability Analyses of Terminal Ballistic Trajectory 151

and the attacking angle of the projectile b is ranged from −90° to 90°. Positive
attacking angle is chosen when the velocity component along x-axis is positive. The
length and the shank diameter are denoted, respectively, as L ¼ k2 d and d, k2 is the
aspect ratio of the projectile. The distance of CG from the nose tip of the projectile
is expressed as L3 ¼ k3 d. The length of the projectile nose is L4 ¼ k4 d. In particular
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
for an ogive-nosed projectile, it has k4 ¼ w  1=4.
Figure 4.10b is illustrated again in Fig. 4.22b, which shows the 3D schematic
diagram and the longitudinal section profile of projectile nose in x-z plane crossing
an arbitrary surface point A of the projectile, respectively. rr is the normal stress
acted on point A, and g is the angle between penetrator axis and the tangential line
crossing point A in the longitudinal section, r is the radius of the transverse cross
section through the point A, U is the angle between the x-axis and the projection of
rr on the above transverse cross section. Regarding the description of the asym-
metric nose abrasion described in Sect. 4.3, we suppose that the two half sides of
the projectile nose (I and II) are blunted at different levels. Here, we assume that the
side II (p=2  U\3p=2) is blunted more severely than the side
I (p=2  U\p=2). The instantaneous transverse cross section of projectile nose is
given in Fig. 4.22c, similar with Fig. 4.10c, where the solid and dashed contours
are corresponding to the original and asymmetrical abrasive projectile, respectively.
Assuming that only rigid-body translation and axial rotation of the projectile
occur, and the velocity components of the center of mass are (vx;CG ; vy;CG ; vz;CG );
thus, the velocity components of the point A can be expressed by Eq. (4.22).
Since the 2D scenario is considered, only the translation of the projectile in the
x-z plane and the rotation of the projectile about y-axis are considered, which has
h_ x ¼ h_ z ¼ 0 and vy;CG ¼ 0. Also, from Fig. 4.22a, we can derive that vx;CG ¼ v sin b
and vz;CG ¼ v cos b. Thus, the velocity components of an arbitrary surface point A of
the projectile are further expressed as

vx ¼ v sin b þ zh_ y ; vz ¼ v cos b þ xh_ y ð4:54Þ

Based on Fig. 4.22b and Eq. (4.54), the normal velocity of an arbitrary surface
point A of the projectile Vc is expressed as
   
Vc ¼ v cos b þ xh_ y sin g þ v sin b þ zh_ y cos g cos U ð4:55Þ

It should be noted that the influence of angular velocity about y-axis h_ y on the
velocity vz was neglected in Bernard and Creighton (1978).
Regarding the concrete target, a semiempirical expression of normal stress rr
acted on the projectile was derived from the spherical cavity expansion theory by
Forrestal et al. (1994) as
152 4 Mass Abrasive Projectile Penetration

rr ¼ Sfc þ q0 Vc2 ð4:56Þ

Equation (4.56) is obtained by the regression of the test data based on the
spherical cavity expansion theory. It should be pointed that the effects of the
constitutive behavior of the target and the frictional resistance are all lumped into
Eq. (4.56). Thus, the tangential frictional force acted on the projectile can be
neglected when using Eq. (4.56) in the drag force analysis.

4.4.2 Free-Surface Effect

Regarding a projectile high-speed penetrating into concrete targets non-normally,

apart from the asymmetric nose abrasion, the asymmetric loadings can be attributed
by the initial oblique and attacking angles due to the existence of the free surface.
The influence of the front surface of the semi-infinite depth target on the drag force
acted on the projectile is usually considered. While for a finite thickness target, the
influences of both the front and rear surfaces of the target should be taken into
account simultaneously.
Equation (4.56) is acquired based on the infinite dynamically expanding
spherical cavity model. However, when the projectile impacts the target
non-normally, there must be some correction with Eq. (4.56) because it will be
affected by the free-surface. Here, we employ the free-surface effect model pro-
posed by Warren and Poormon (2001) and Warren et al. (2004b), shown in the
Fig. 4.23, where the resistance acting on the surface of a penetrator depends on the
distance rd ¼ ra þ d  , where d  is taken from a surface node to the free surface of
the target along a normal line to the penetrator surface and ra is a radius depending
on the local penetrator geometry at the surface node.

Fig. 4.23 Sketch of the

free-surface effect model

Free surface
CG

Target
4.4 Stability Analyses of Terminal Ballistic Trajectory 153

By constructing a decay function f ðrd ; ra ; Vc Þ to account for the free-surface

effect, Warren and Poormon (2001) and Warren et al. (2004b) proposed that the
normal stress rr acted on the projectile can be expressed as
 
rr ¼ f ðrd ; ra ; Vc Þrr ¼ f ðrd ; ra ; Vc Þ Sfc þ q0 Vc2 ð4:57Þ

The above decay function is based on the solution of a dynamically expanding

spherical cavity in a finite sphere of the target that is assumed to be incompressible
and Mohr–Coulomb material, and the decay function is summarized as follows

rr ðra Þ
f ðrd ; ra ; Vc Þ ¼ ð4:58Þ
rr ðra Þrd !1

where rr ðra Þ and rr ðra Þrd !1 are the radial stresses at cavity surface when the
free-surface effect of the target is considered or not, respectively.
Based on the dynamic spherical cavity expansion theory, during the penetration,
a spherically symmetrical cavity forms and expands from zero initial radius to
radius ra with the expansion velocity Vc , which is given in Eq. (4.55). For the brittle
target material, e.g., concrete and rock, the cavity expansion usually produces
comminuted, cracked, plastic, elastic, and intact regions radially outward. For
convenience, here we assume that the cavity expansion produces plastic, elastic,
and intact regions, as shown in Fig. 4.24.
It should be noted that the expansion of cavity might produce both plastic and
elastic regions when rd is relatively large, e.g., rd  rp shown in Fig. 4.24a. While
the expansion of cavity only produces plastic region when rd is relatively small,
which means that the plastic region reaches the free surface, e.g., rd \rp shown in
Fig. 4.24b. The plastic region is bounded by the radii ra and rp , where rp is the
interface between the plastic and elastic response regions. Similarly, the elastic
region is bounded by rp and rd , where rd corresponds to the position of the elastic

(a) (b)

Intact
Elastic Plastic
Plastic Plastic
Cavity Cavity Free-surface

ra rp rd ra rd rp

Fig. 4.24 Response regions for the cavity expansion problem

154 4 Mass Abrasive Projectile Penetration

dilatation wave front in Fig. 4.24a or free surface in Fig. 4.24b where the radial
stress diminishes to zero.
By employing the momentum and mass conservation laws, rr ðra Þ can be
obtained as follows:
For rd  rp ,


ak=3 ( "
3 #
1 1 s 2Ec 2s rp
rr ðra Þ ¼
2q0 Vc2   þ 1
1k  4 1k  1 k 3s 3 rd
"
1=3

4=3

 # )
1k 3s 1 1k 3s ra 1 ra 4 s
þ 2q0 Vc
2
  þ þ
1k  1 2Ec 4 1k  1 2Ec rd 4 rd k
ð4:59aÞ

For rd \rp , it is considered that the concrete material in the plastic zone offers no
resistance on the projectile; thus,

rr ðra Þ ¼ 0 ð4:59bÞ

where 1 ¼ 6=ð3 þ 2kÞ, k and s are the pressure hardening coefficient and the
cohesive force in the Mohr-Coulomb yield criterion, respectively. For a concrete
target, k ¼ 0:67 and s ¼ ð3  kÞfc =3 are adopted in the following calculation by
referring to Forrestal and Tzou (1997). The symbol Ec is the elastic modulus of
concrete, and the distance rp ¼ ra ð2Ec =3sÞ1=3 is derived by the continuity condition
of the stress at the interface of plastic and elastic zones (Warren et al. 2004b).
rr ðra Þrd !1 in Eq. (4.58) is derived by letting rd ! 1 in Eq. (4.59a).
Therefore, for a projectile non-normally penetrating into a semi-infinite depth
concrete target, by comparing the values of the distances rd and rp , the normal stress
acted on the upper side of the projectile rr is obtained based on Eqs. (4.56–4.59a, b).

4.4.3 Mass Abrasion Model of Projectiles

Regarding the projectile high-speed penetrating into concrete targets, the mass loss
and nose blunting of the abrasive projectile occur obviously. As introduced in
Sect. 4.2, Zhao et al. (2010a) proposed that the instantaneous mass of the projectile
M during the penetration satisfies

1  
M ¼ M0  M0 C V02  v2 ð4:60Þ
2

where M0 is the mass of the unfired projectile, V0 is the initial striking velocity of
the projectile, C is the mass loss coefficient in s2/km2 with respect to the striking
velocity V0 in km/s, and v is the instantaneous velocity of the projectile.
4.4 Stability Analyses of Terminal Ballistic Trajectory 155

For the nose-blunting effect, Zhao et al. (2010a, 2012) proposed that the
instantaneous nose factor of the ogive-nosed projectile during the penetration
process N  satisfies
 
N  ¼ N0 þ k0 V02  v2 ð4:61Þ

In Sect. 4.2, the analytical expressions of the mass loss and nose-blunting
coefficients are given as follows

0:8gN a gN a ðw þ 4ÞM0
C ¼ pffiffiffi 0 ; k0 ¼ p0ffiffiffi 0 ð4:62aÞ
3jQS 6 3jQSMn w20


1
1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
N ¼ 4w cos
a 2
1  ð2w  1Þ 4w  1 ð4:62bÞ
2w

Limited by the existing measuring techniques in the high-speed penetration tests

on concrete targets, the time-dependent asymmetric nose abrasion could not be
realized. Regarding the asymmetric nose abrasion described in Sect. 4.4.1, we
suppose that the side I of the projectile nose is blunted regularly and the corre-
sponding nose-blunting coefficient kI0 is obtained by Eq. (4.62a), while the side II is
blunted more severely than side I, the related nose-blunting coefficient kII0 [ kI0 .
As mentioned previously, the nosed geometry of the abrasive projectile still
keeps ogival shape during the penetration and approaches to a hemispherical shape
gradually with increasing the initial striking velocity. Considering the most extre-
mely circumstance and assuming that the side II of the projectile is finally blunted
to hemispherical shape (w ¼ 0:5) when the initial striking velocity is 1 km/s, the
nose-blunting coefficient kII0 can be derived from Eq. (4.61) as

kII0 ¼ 0:5  N0 ð4:63Þ

It should be noted that kII0 derived from Eq. (4.63) is the upper bound of the
nose-blunting coefficient, and the nose-blunting coefficients kI0 and kII0 are corre-
sponding to the most distinct abrasion circumstance. The discrepancy of the
nose-blunting coefficients of both two sides of the projectile nose achieves the
maximum. Therefore, the deviation of the ballistic trajectory also reaches the
maximum.

4.4.4 Prediction of Terminal Ballistic Trajectory

As shown in Fig. 4.22a, the matrix to transform the CG velocity and the forces
acting on the projectile from the projectile xz-coordinate system to the target XZ-
coordinate system is given by
156 4 Mass Abrasive Projectile Penetration

Fig. 4.25 Scheme of the L=λ2d

longitudinal section of the L3=λ3d
projectile
L4=λ 4 d
x

d
o

s
Θ1
Θ2 Θ


cos hy sin hy
T¼ ð4:64Þ
 sin hy cos hy

Denoting angles H1 and H2 as shown in Fig. 4.25, which can be expressed by

pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi!

w  0:25 k2  k4
H1 ¼ arcsin ; H2 ¼ arctan ð4:65Þ
w w

Therefore, the nose and shank of the projectile are corresponding to the
arcsinð1  1=2wÞ  H\p=2 and p=2  H  p=2 þ H2 , respectively. The projectile
is discretized longitudinal and transversely by dividing the projectile H1 , H2 , and U
equally into I, J, and K parts, respectively. The coordinate of an arbitrary surface
point located at the projectile nose and shank can be expressed as following,
respectively.
For the projectile nose,
8        
>
> x ¼ s sin arcsin 1  1
þ i
H  s þ d
cos k
2p
>
<    2w
 I 1
 2
  K

y ¼ s sin arcsin 1  2w 1
þ Ii H1  s þ d2 sin Kk 2p ð4:66aÞ
>
>    
>
: z ¼ ðk  k Þd þ s cos arcsin 1  1 þ i H
3 4 2w I 1

For the projectile shank,

 
x ¼ d2 cos Kk 2p
 
y ¼ d2 sin Kk 2p ð4:66bÞ
j 
z ¼ ðk3  k4 Þd  s cot J H2

Based on Eqs. (4.66a, b), the area of an arbitrary surface element located at the
projectile nose and shank can be expressed as following, respectively.
For the projectile nose,
4.4 Stability Analyses of Terminal Ballistic Trajectory 157

 
p H1 1 i
dA1 ¼ s s sin arcsin 1  þ H1
K I 2w I


 
ð4:67aÞ
1 i1
þ s sin arcsin 1  þ H1  2s þ d
2w I

For the projectile shank,




p j1 j
dA2 ¼ sd cot H2  cot H2 ð4:67bÞ
K J J

From the Fig. 4.22b and regarding Eq. (4.57), the components of the normal
stress acting on the projectile with accounting for the free-surface effect can be
given by

rx ¼ rr cos U cos g; rz ¼ rr sin g ð4:68Þ

The forces and moments acting on the surface of the projectile are deduced from
the rigid-body dynamics, and the motion equations of the projectile are given by
(    R R 
Mt X €t ; Z€t T ¼ Tt  rx dA  rz dA T
 R R R R  t ð4:69Þ
I00 €hy;t ¼  R jzjrx dA þ R j xjrz dA t

where j xj and jzj denote the normal distance of an arbitrary point to the z-axis and x-
axis for obtaining the moments, respectively. R is the total area of the embedded
€ and Z€ are the translation acceleration components of the CG
part of the projectile. X
in the target XZ-coordinate system. €hy is denoted as the angular acceleration of the
projectile centerline surrounding y-axis. I00 is the transverse mass moment of inertia
about the y-axis, which is difficult to be determined for the complicated inner
structure of the projectile, and it is assumed as I00 ¼ M0 L2 =12 for simplicity.
Denoting the time increment as Dt and assuming that the CG of the projectile is
decelerated uniformly during each time step when Dt ! 0, based on Eqs. (4.68,
4.69), the translation and rotation velocities of the projectile in the target XZ-
coordinate system can be expressed as
( T  T  R R T
X_ t þ 1 ; Z_ t þ 1 ¼ X_ t ; Z_ t þ Tt  R rx dA  R rz dA t Dt=Mt
 R R  ð4:70Þ
h_ y;t þ 1 ¼ h_ y;t þ  R zrx dA þ R xrz dA t Dt=I00

Furthermore, the displacement and rotation angle velocity of the projectile can
be obtained as
158 4 Mass Abrasive Projectile Penetration

Fig. 4.26 Scheme of stress p3

acted on the projectile surface 3
element

4 p4

2 p2

1 p1

(  T  R R T
ðXt þ 1 ; Zt þ 1 ÞT ¼ ðXt ; Zt ÞT þ X_ t ; Z_ t Dt þ Tt  R rx dA  R rz dA t Dt2 =ð2Mt Þ
 R R  2 0
hy;t þ 1 ¼ hy;t þ h_ y;t Dt þ  R zrx dA þ R xrz dA t Dt =2I0
ð4:71Þ

Based on the above approach, a mesh at the projectile surface is firstly formed by
Eqs. (4.66a, b) at each time step. Then, judge the locations of the surface points of
the projectile and apply the normal stresses from Eqs. (4.56, 4.57, 4.68) on the
surface nodes of the projectile, as illustrated in Fig. 4.26. According to the averaged
nodal stresses of P1–P4 and element area given by Eqs. (4.67a, b), the forces and
moments acting on the surface of the projectile in Eq. (4.69) are derived. Finally,
the projectile motion Eqs. (4.69–4.71) can be solved, and the motion parameters of
the projectile such as velocity, acceleration, and rotation angle at the end of this
time step are obtained and output. If the global velocity of the projectile is positive,
the calculation enters into the next step.
During the next time step, the abrasive mass and asymmetric blunted nose shape
of the projectile is firstly updated by Eqs. (4.60, 4.61), and the mass loss coefficient
and the asymmetric CRHs of the two sides of projectile nose (wI and wII ) are
obtained from Eqs. (4.62a, b, 4.63), respectively. Then the calculation procedure is
repeated until the global CG velocity of the projectile equals to zero.
Based on the detailed influential analyses of element size and time step on the
penetration depth of the projectile, the parameters I = J = K = 150 and Dt ¼ 105 s
are adopted. The flowchart of the above calculation procedure is given in Fig. 4.27,
which is programmed by the software MATLAB and the program is named as
PENTRA2D.

4.4.5 Verifications of PENTRA2D

Different from the metallic target, few test data on terminal ballistic trajectory of the
high-speed projectile penetration into concrete targets was publicly reported. The
validation of PENTRA2D is verified by comparing the terminal rest or exit loca-
tions of the projectiles in the high-speed non-normal penetration test on the concrete
4.4 Stability Analyses of Terminal Ballistic Trajectory 159

START

Initial parameters
M0,V0, ψ0, α, β, k′I , k′II …

Projectile surface meshing

N Projectile surface node

enters into target or not

Y Update the asymmetric

σ r* = 0 N CRHs Eqs. (4.61)~(4.63)
Vc>0
Y
Update the mass of
f =1 σ from Eqs. (4.56~4.58)
*
r
projectile Eq. (4.60)

Solving Eqs. (4.69~4.71) of

t=t +Δt
motion and output X , Z , X , Z

X < 0 and Z <0

N
Y

END

Fig. 4.27 Flowchart of the program PENTRA2D

target conducted by Forrestal et al. (1996). The related parameters of the four shots
are listed in Table 4.2.
Figure 4.28 illustrates the terminal rest or exit locations of the projectiles (the
square solid symbol) in the four shots given in Table 4.2, as well as the trajectories
predicted by PENTRA2D, which considers the asymmetric nose abrasion and
corresponds to the most unfavorable condition. The predicted trajectories (solid
curves) in Fig. 4.28 are the most serious deviated trajectories. The shadow part in
each figure stands for the half of the axial longitudinal section of the concrete target.
The origin is the striking point of the projectile, where the vertical coordinate
represents the depth of penetration, and the horizontal coordinate demonstrates the
trajectory deviation along the radial direction of the concrete target.
As shown in Fig. 4.28, regarding the four shots of projectile high-speed normal
penetrating into the concrete targets with small attacking angles (0°–2°), initially
the straight path along the incident impact direction are maintained with the dis-
tance of nearly the length of the projectile. The straight trajectories are all about
0.1 m and the lengths of the projectiles are L = 0.0889 m, which is also consistent
with the experimental measurements by He et al. (2010a). After that, under the
combined effect of the initial attacking angle and the asymmetric nose abrasion, the
160 4 Mass Abrasive Projectile Penetration

Table 4.2 Parameters of the target and projectile in the high-speed penetration tests (Forrestal
et al. 1996)
Parameters Shot 1 Shot 2 Shot 3 Shot 3
fc (MPa) 19.5 19.5 19.5 19.5
Target diameter (m) 0.48 0.48 0.48 0.48
w0 3 4.25 4.25 4.25
H (m) 1.82 1.82 1.82 1.82
d (m) 0.0129 0.0129 0.0129 0.0129
q0 (kg/m3) 2000 2000 2000 2000
HM 7 7 7 7
M0 (kg) 0.064 0.064 0.064 0.064
M0/Mn 5.7 4.6 4.6 4.6
kI0 (s2/km2) 0.1 0.06 0.06 0.06
kII0 (s2/km2) 0.394 0.424 0.424 0.424
V0 (m/s) 1356 1311 1359 1430
a (°) 0 0 0 0
b (°) 1 1.8 0.5 0.8

(a) 0.00 0.05 0.10 0.15 0.20 0.25 0.30

(b) 0.00 0.05 0.10 0.15 0.20 0.25 0.30
0.0 0.0
0.2 impact point Target edge 0.2 Target edge
Front face of the target Front face of the target
impact point
Displacement Z (m)
Displacement Z (m)

0.4 0.4
0.6 0.6
0.8 0.8
1.0 Symmetric axis of target 1.0 V0=1311m/s
Symmetric axis of target
1.2 (incident straight line) 1.2 (incident straight line)
V0=1356m/s
1.4 1.4
1.6 Rear face of the target 1.6 Rear face of the target
1.8 1.8
2.0 2.0
Displacement X (m) Displacement X (m)

(c) (d)
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.00 0.05 0.10 0.15 0.20 0.25 0.30
0.0 0.0
impact point impact point
0.2 Front face of the target
0.2 Front face of the target Target edge
Target edge
Displacement Z (m)
Displacement Z (m)

0.4 0.4
0.6 0.6
0.8 0.8
Symmetric axis of target
1.0 Symmetric axis of target V0=1430m/s
1.0 (incident straight line)
(incident straight line)
1.2 1.2 V0=1430m/s
1.4 1.4
1.6 Rear face of the target 1.6 Rear face of the target

1.8 1.8
2.0 2.0
Displacement X (m) Displacement X (m)

Fig. 4.28 Terminal location or exist point of the projectile and the predicted most unfavorable
ballistic trajectory a Shot 1, b Shot 2, c Shot 3, and d Shot 4
4.4 Stability Analyses of Terminal Ballistic Trajectory 161

following ballistic trajectory tends to a J-shape curve from a straight line. Since the
predictions are the most unfavorable scenarios of deviated trajectory, all the ter-
minal rest or exit locations of the projectile in the four tests are located between the
incident straight and predicted curved trajectories. It confirms the validation of
PENTRA2D for predicting the terminal ballistic trajectory to some extent. Actually,
more high-speed non-normal penetration tests on concrete targets for the terminal
ballistic trajectories are urgently needed.

4.4.6 Parametric Influential Analyses

In this section, the parametric influential analyses on the terminal ballistic trajectory
are conducted, where the initial striking velocity is V0 = 1 km/s, and the other
related calculation parameters are from the Shot 1 in Table 4.2.
Figure 4.29a shows the predicted trajectories of the projectiles with the oblique
angles of a ¼ 15 , 30 , and 45 , and the attacking angle b ¼ 0 , respectively. The
straight arrows represent the incident directions. Figure 4.29b illustrates the tra-
jectories of the projectiles with the initial attacking angles b ¼ 1 –3 referred from
Goldsmith (1999) and the initial oblique angle a ¼ 0 . Figure 4.29c illustrates the
deviated trajectories of the normal penetrating projectile in the case of the

(a) 0.0 0.1 0.2 0.3 0.4 0.5

(b) 0.00 0.05 0.10 0.15 0.20
0.0 0.0
a=15 β=1
Displacement Z (m)
Displacement Z (m)

0.1 a=30 0.1 β=2

a=45 β=3

0.2 0.2

0.3 0.3

0.4 0.4

0.5 0.5
Displacement X (m) Displacement X (m)

(c) 0.00 0.01 0.02 0.03 0.04 0.05 0.06

(d)
0.00 0.01 0.02 0.03 0.04 0.05 0.06
0.0 0.0
k'I=0.1, k'II=0.2 3/ 2=0.5
0.1 0.1
Displacement Z (m)

3/ 2=0.55
Displacement Z (m)

k'I=0.1, k'II=0.3
3/ 2=0.6
0.2 k'I=0.1, k'II=0.4 0.2

0.3 0.3

0.4 0.4

0.5 0.5

0.6 0.6
Displacement X (m) Displacement X (m)

Fig. 4.29 Parametric influential analyses on the terminal ballistic trajectory

162 4 Mass Abrasive Projectile Penetration

asymmetric nose abrasion with different discrepancies of the nose-blunting coeffi-

cients, where a ¼ b ¼ 0 . Figure 4.29d depicts the trajectories of the normal
penetrating projectile with different relative CG locations, where a ¼ b ¼ 0 and
the relative CG locations are referred to Li and Flores-Johnson (2011).
In general, a projectile always firstly penetrates along the incident direction to a
certain distance after entering into the target. As shown in Fig. 4.29a, this straight
distance (represented with the arrow) increases with the rising of the oblique angle.
Under the combined action of initial oblique and attacking angles as well as the
asymmetric nose abrasion, the trajectories then deviate gradually and finally form
“J-shaped” curves toward the target surface, even ricochet of the projectile would
occur under extreme conditions. The deviation of the terminal ballistic trajectory is
aggravated with the increasing of the incident and attacking angles, the discrepancy
of the asymmetric nose abrasion as well as the relative location of the mass center
of the projectile. The above influential laws of the oblique and attacking angles are
consistent with the conclusions by Warren et al. (2004b) and Forrestal et al. (1996),
which drawn from the high-speed non-normal penetration tests on the limestone
and concrete targets with different incident and attacking angles.
Within the discussion range in Fig. 4.29, the most influential factor for the ter-
minal ballistic trajectory is the attacking angle and subsequently followed by the
oblique angle, the discrepancy of asymmetric nose abrasion and the location of mass
center of the projectile. Therefore, for the large aspect-ratio projectile high-speed
penetration into concrete targets, the rigorous control of the incident gesture espe-
cially the attacking angle is important for the terminal ballistic stability. Besides,
utilization of high-strength alloy is helpful to reduce the symmetrical or asymmetric
mass abrasion of the projectile nose to guarantee the ballistic stability and ballistic
performance of the projectile. On the contrary, from the view of designing protective
structures, the asymmetric abrasions of the projectile could be aggravated by uti-
lizing the high-strength concrete and the harder coarse aggregate, such as granite and
corundum; thus, the terminal ballistic trajectory could be deviated and the pene-
tration performances of the projectiles could be weakened tremendously.

4.5 Summary

In this chapter, for the projectile penetrating into concrete targets with the striking
velocity nearly between 0.9 and 1.5 km/s, the mass abrasion law as well as the
structural and terminal ballistic trajectory stabilities of the mass abrasive projectile
are discussed. The main conclusions are as follows:
(1) In Sect. 4.2, the dimensionless equation for relative mass loss of a projectile is
formulated, and the analytical expression of the upper critical striking velocity
for the mass loss proportion to the initial impact kinetic energy of the pro-
jectile is derived, which is confirmed as about 1.0 km/s based on a large
amount of penetration tests. The engineering model for the mass abrasion of
4.5 Summary 163

projectile high-speed penetrating into concrete targets is established, which

provides the physical characteristics of the mass abrasion of the projectile and
also guarantees the prediction accuracy comparing to the completely empirical
approaches in the previous works. Since the foundation of the discussion is the
impact function and the geometry function of a projectile, it is hopeful that the
proposed engineering model may be applied for any nose-shaped projectile.
(2) In Sect. 4.3, the structural stability of mass abrasive projectile high-speed
non-normal penetrating into concrete targets and the related parametrical
influences are analyzed. By deriving the axial and the transverse drag forces
acting on the projectile as well as the ideal elasto-plastic yield criterion, an
approach for predicting the limit striking velocity (LSV) is proposed and
validated by the several high-speed penetration tests. The special penetration
scenarios are discussed in detail and the regression forms of the proposed
formulae are consistent with the existing works. Regarding projectile
high-speed penetration into concrete targets, the future work should be con-
centrated on the non-normal penetration tests with larger initial oblique and
attacking angles, as well as the experimental innovations of real-time mea-
surement techniques for the terminal ballistic mass abrasion of the projectile.
(3) In Sect. 4.4, DAFL method is used to construct the finite differential approach to
predict the terminal ballistic trajectory of the high-speed projectile. It simulta-
neously accounts for the free-surface effect of the target attributed by oblique and
attacking angles, as well as the asymmetric nose abrasions. The validation of the
proposed program PENTRA2D is verified by comparing with the available test
data. The parametric analyses show that, the most influential factor for the
stability of the terminal ballistic trajectory is attacking angle and subsequently
are the oblique angle, the discrepancy of asymmetric nose abrasion, and the
location of mass center of projectile. Besides, the deviation of the terminal
ballistic trajectory is aggravated with the increases of the above four parameters.
Additionally, great efforts should be paid to the following three aspects in the
future: (i) by means of advanced measurement technique, e.g., the flash X-ray and
high-speed photography, the experimental data of the projectile penetrating into
concrete targets in different scenarios, especially for the non-normal and high-speed
(0.9 km/s < V0  1.5 km/s) penetrations, should be further renewed and accumu-
lated; (ii) during the projectile high-speed penetration into concrete targets, the
deviation of the terminal ballistic trajectory is always companied by the integral
structural instability of the projectile, such as bending and fracture. Therefore, except
for the influential factors discussed in this chapter, the comprehensive analysis on
terminal ballistic trajectory needs further considering the structural response of the
projectile simultaneously; (iii) PENTRA2D proposed in this chapter mainly concerns
with the ogive-nosed projectile as well as the 2D scenario that the planes of oblique
and attacking angles coincide with each other. Thus, the predicted approach for the 3D
trajectory of the projectile with arbitrary nosed shape needs to be further developed.
Chapter 5
Eroding Projectile and Shaped
Charge Jet Penetrations

5.1 Introduction

In this chapter, the penetration performances of eroding projectiles and shaped

charge (SC) jet are discussed.
When the striking velocity of a projectile is large enough or the projectile
material is relatively “soft,” the pressure at the projectile/target interface is much
higher than the yield strength of the projectile material. Consequently, the projectile
begins to erode during the penetration process, such as the projectile penetrating
into concrete targets shown in Fig. 5.1.
Existing experimental and theoretical works for the eroding projectile penetra-
tions are mainly established for metallic targets, and a comprehensive summary of
the related experimental data was conducted by Anderson et al. (1992b). However,
for the deep penetration of the eroding projectile into a semi-infinite concrete target,
the available penetration test data is very limited. Gold et al. (1996) and Nia et al.
(2014) conducted the tantalum and copper projectile penetration tests on concrete
targets with the compressive strengths varied from 37.4 to 65.6 MPa, where the
corresponding striking velocities were ranged from 1475 to 1875 m/s and 660 to
1134 m/s, respectively. The projectiles are found to be eroded after the penetration
tests, which indicated that the penetration entered into the eroding stage for these
projectile/target combinations during the above velocity ranges. Furthermore, due
to the lack of the experimental data, the verified analytical model is also limited.
Besides, the SC jet is widely applied in the military and extraction of oil or gas
fields, e.g., the tandem warhead of earth penetration weapon and oil well perforator,
while the existing studies of the SC jet penetrating into concrete material are
limited. The typical SC is composed of a cylindrical explosive with a conical cavity

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 165
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_5
166 5 Eroding Projectile and Shaped Charge Jet Penetrations

V0=710m/s V0=825m/s V0=920m/s V0=950m/s V0=956m/s

Unfired

Fig. 5.1 Unfired and post-test photographs of projectiles (Nia et al. 2014)

adhered to a thin metal liner on one side and a booster on the other side. When the
explosive is detonated by the booster, a spherical detonation wave propagates
outward. Simultaneously, the metallic liner material is accelerated normally under
the high-detonation pressure and forms a jet with high velocity. Consequently, the
target is damaged due to the extremely high kinetic energy of a high-speed jet.
The SC jet can be roughly classified into JPC (jetting projectile charge) and EFP
(explosively formed projectile) with the angle of metallic liner varied within 70°–
120° and 120°–160°, respectively. JPC possesses obvious velocity gradient and
better penetration performance, while EFP exhibits nearly none velocity gradient
and good stability at larger standoff distance. In this book, the performance of JPC
penetrating into concrete targets is mainly concerned.
In this chapter, in Sect. 5.2, we focus on the experimental and analytical studies
of the eroding projectile penetration into concrete targets. Firstly, the existing
eroding long-rod penetration models for metallic targets are revisited to check
whether they can be applied to the concrete targets directly. Secondly, a series of
eroding projectile penetration experiments into mortar targets are conducted and the
damages of the projectiles and targets are assessed. Finally, an analytical model for
the eroding projectile penetration into concrete targets is established and compared
with the test data. In Sect. 5.3, the existing studies on the performance of SC jet
penetrating into the concrete targets are reviewed firstly. Then, the whole processes
of SC jet penetrating into concrete targets are numerically reproduced. The simu-
lated striking velocities of the jet, as well as the damages of the targets, are com-
pared with the available related test data. Finally, the influences of standoff
distance, liner material, and explosive type of SC, as well as the compressive
strength of concrete targets on the diameters and depth of the SC jet penetration
borehole, are further discussed.
5.2 Eroding Projectile Penetration 167

5.2 Eroding Projectile Penetration

5.2.1 A Review of the Existing Studies

The methods used to understand the basic mechanism of an eroding projectile into a
semi-infinite target are classified by three categories, i.e., terminal ballistic exper-
iments, idealized analytical models, and numerical simulations. Among the ana-
lytical models, Alekseevskii–Tate (A-T) model proposed by Tate (1967, 1969) and
Alekseevskii (1966) independently has been considered as a classical
one-dimensional model for the eroding projectile penetration, which is also called
semi-hydrodynamic model as it further includes the influences of both the dynamic
strength of the projectile (Yp) and the target resistance (Rt) in each side of the
Bernoulli’s equation. It is generally accepted that Yp approaches to the elastic limit
under one-dimensional strain condition [Hugoniot elastic limit (HEL)], while Rt is
several times of the dynamic strength of the target due to the confinement contri-
bution of the surrounding target material to the one-dimensional resistance. In the
original A-T model, Rt must be obtained by fitting the experimental data, which
reduces the predictive capability of the model. In a further study, Tate (1986)
presented formulae for Yp and Rt as functions of material property parameters.
However, A-T model cannot give good predictions of the penetration depth for both
low and high velocities simultaneously, no matter how the values of Yp and Rt are
chosen, which indicates the inherent deficiency of A-T model. Tate (1986)
attempted to consider the contribution of the transient and “after-flow” phases to the
discrepancy between the experimental results and model predictions under high
impact velocities, while the convincing explanations about the discrepancy are not
given.
For the existing analytical models concerned in this chapter, studies have been
attempted to improve the A-T model. For example, the cavity expansion model has
been used in several attempts to give the analytical prediction of the constant Rt
(e.g., Rosenberg et al. 1990; Forrestal and Longcope 1990), which, however, is
inconsistent with the study by Anderson et al. (1992a). Anderson et al. (1992a)
found that the constant Rt obtained by fitting the A-T model predictions to the
experimental results is dependent on the impact velocity, and Rt actually varies
during the penetration process according to numerical simulations. With consid-
ering the non-uniform pressure distribution on the projectile/target interface,
Rosenberg et al. (1990) introduced the ratio of the effective cross-sectional areas of
the mushroomed and rigid part of the projectile to modify the A-T model based on
the consideration of the force balance. Lan and Wen (2010) made an extension to
the semi-hydrodynamic model, assuming that the deformed target consists of three
distinct regions, i.e., fluid-like region, plastic region, and elastic region. Several
assumptions, such as the particle velocity and pressure profiles, were made in the
model where a number of empirical constants should be given, i.e., the critical
penetration velocity when the fluid-like region appears, as well as the relationship
between the interface particle velocity and the penetration velocity. Nevertheless,
168 5 Eroding Projectile and Shaped Charge Jet Penetrations

satisfactory predictions were reported for the projectile-target systems used in their
study.
An approach to establish a more fundamental eroding projectile penetration
model was presented by Walker and Anderson (1995) based on the momentum
equation along the centerline of the penetration. In order to integrate the momentum
equation, several assumptions, such as the velocity profile along the centerline in
the projectile and target and the shear behavior of the target material, were made.
This model gave very good predictions for both rigid and eroding long-rod pene-
trations. However, as discussed by Satapathy (1997), this model still has some
limitations. Firstly, the cavity radius was obtained by fitting the experimental data,
which restricts the predictive capability of the model. Secondly, the assumed
velocity profile of the target along the centerline was scaled in such a way that the
radial velocity became zero at elastic-plastic interface, which is inconsistent with
their numerical simulation results. Additionally, a number of differential equations
need to be solved numerically; thus, the complexity of the model is not suitable for
engineering applications. Satapathy (1997) assumed that there is a damaged zone in
the target during eroding projectile penetration, where the target material is
deformed intensively and the shear flow is limited to this zone. The pressure profile
in the damaged zone was obtained by integrating the momentum equation along the
centerline using the same method proposed by Walker and Anderson (1995). The
pressure profile in the surrounding plastic and elastic zones was determined by the
classical dynamic spherical cavity expansion model proposed by Forrestal and Luk
(1988a), where an empirical constant (the diameter ratio of damaged zone to cavity)
was also introduced. The predicted results by the above model showed good
agreement with experimental results.
The existing works have the following limitations: (i) For the experimental
studies, the range of striking velocities for each projectile and target combination is
relatively narrow, and only one penetration regime, i.e., either rigid or eroding
penetration, is observed for each test. DOP test into concrete targets with a broad
range of striking velocities in which different penetration regimes can occur is still
scarce in the literatures. Besides, very limited test data is available for the eroding
projectile penetration into concrete targets; (ii) the target resistance is vital for the
study of penetration process, while the existing dynamic cavity expansion models
applied to determine the target resistance are based on linear yield criterion and
EOS, which cannot well describe the plastic behavior of concrete material subjected
to high-speed projectile impact; and (iii) for eroding projectile penetration, the
classical A-T model cannot be applied to the concrete targets directly.

5.2.2 Comparison Between the Existing Models

This section firstly introduces the typical eroding projectile penetration models for
metallic targets briefly. Then, attempts are made to apply these models for concrete
5.2 Eroding Projectile Penetration 169

targets directly. Readers are referred to the original publications of these models for
detailed descriptions.

5.2.2.1 Existing Models

(i) A-T model

Assuming that an infinitesimal fluid region exists around the projectile/target
interface during the steady-state penetration, Tate (1967, 1969) and Alekseevskii
(1966) independently modified the Bernoulli’s equation with the considerations of
both the projectile and target strengths as

qp ðv
uÞ2 =2 þ Yp ¼ q0 u2 =2 þ Rt ð5:1Þ

The deceleration and erosion rate of the projectile are

v_ ¼
Yp =qp l ð5:2Þ

_l ¼ u
v ð5:3Þ

where Yp and Rt are the dynamic strength of the projectile and one-dimensional
target resistance, respectively. qp and q0 are the projectile and target densities,
respectively. The superscript dot represents the derivative of time. v is the speed of
the rear rigid portion of the projectile, u is the penetration speed, and l is the
instantaneous projectile length, as shown in Fig. 5.2.
(ii) RMM model
Considering the force balance at the projectile/target interface, Rosenberg et al.
(1990) modified the A-T model as



1 1
Ap qp ðv
uÞ2 þ Yp ¼ At q0 u2 þ Rt ð5:4Þ
2 2

where Ap and At are the effective areas of the projectile rigid portion and mush-
roomed portion, respectively. Rosenberg et al. (1990) suggested that At/Ap = 2. Yp

Fig. 5.2 Diagram of A-T

model
l

v u
170 5 Eroding Projectile and Shaped Charge Jet Penetrations

is determined as the HEL of the projectile material, i.e., Yp ¼ ryp ð1
mÞ=ð1
2mÞ
with ryp and m being the projectile dynamic yield stress and Poisson’s ratio,
respectively.
Based on the cylinder cavity expansion theory, Rt is determined by

pffiffiffi 
ryt 3Et
Rt ¼ pffiffiffi 1 þ ln ð5:5Þ
3 ð5
4mÞryt

where ryt and Et are the dynamic yield stress and elastic modulus of the target,
respectively. The other governing equations are the same as A-T model.
(iii) WA model
Based on the momentum equation along the centerline of the penetration, by
assuming the velocity profile along the centerline of both the projectile and target
and the shear behavior of the target material drawn from the 2D numerical simu-
lations, Walker and Anderson (1995) proposed an eroding projectile penetration
model.
The momentum equation along the centerline is given by

1 1 7
q ðv
uÞ2 þ ryp ¼ q0 u2 þ lnðaÞryt ð5:6Þ
2 p 2 3

where a represents the dimensionless extent of the plastic zone in the target, which
is determined by



q0 u2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi q0 a2 u2 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi2ffi
1þ K t
q0 a u ¼ 1 þ
2 2 Kt
q0 u ð5:7Þ
ryt 2Gt

where Kt and Gt are the bulk modulus and shear modulus of the target material. The
deceleration of the projectile is

v_ ¼
ryp =qp l ð5:8Þ

(iv) LW model
Lan and Wen (2010) assumed that the deformed target consists of three distinct
zones, i.e., fluid-like region, plastic region, and elastic region. Based on the arti-
ficially assumed relationship between the particle velocity at the plastic/fluid-like
region interface and the penetration velocity, the pressure balance equation is given
by
(
2 q p ðv
1

uÞ2 þ Yp ¼ 12 q0 ðu
f ðuÞÞ2 þ S þ Cq0 ðf ðuÞÞ2 ru [ UF0
ð5:9Þ
2 q p ðv
1

uÞ2 þ Yp ¼ S þ Cq0 ðf ðuÞÞ2 ; 0\u  UF0
5.2 Eroding Projectile Penetration 171

where S is the static resistance of targets and is obtained by Eq. (5.5). C is the
coefficient of the dynamic resistance of targets and C = 1.5 for incompressible
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
material. UF0 is empirically determined by UF0 ¼ HEL=qt with HEL being the
Hugoniot elastic limit of the target material. f ðuÞ is obtained based on the numerical
results as

 !
u
UF0 2
f ðuÞ ¼ UF0 exp
ð5:10Þ
nUF0

where n is a constant and n = 2.45 for the incompressible material. Similarly, the
other governing equations are identical with the A-T model. In general, the solu-
tions of all these models can be obtained numerically.

5.2.2.2 Comparison with Test Data

The above four models are originally proposed for metallic targets. In this section,
based on the limited eroding projectile penetration test data, we check whether these
models can predict the DOP for the concrete targets. It should be noted that the
empirical parameters used for existing models are taken as the suggested ones in
their respective studies. For example, the effective cross-sectional area ratio
between the mushroomed rod and rigid rod in Rosenberg et al. (1990) is taken as
two. Table 5.1 presents the material parameters for the projectiles and targets in the
tests.
Figure 5.3 shows the three sets of available test data and the corresponding
predicted DOP of the above four models. It is found that these models could not
give satisfied predictions of the DOP for the concrete targets. This indicates that the
dynamic response of the concrete targets is very different from that of the metallic
targets. Therefore, one cannot use the existing models for the metallic targets to
predict DOP of the concrete targets directly.
Recent research by Nia et al. (2014) also concluded that the classical Forrestal
formula (Forrestal et al. 1996) for rigid projectile penetration greatly overestimates
the DOP for the eroding projectile penetrations, and the Bernoulli’s equation also
cannot give satisfied predictions for DOP. Therefore, it is essential to establish an
analytical model for eroding projectile penetration into concrete targets.

Table 5.1 Property parameters of projectiles and targets (Nia et al. 2014; Gold et al. 1996)
Material q (kg/m3) E (GPa) G (GPa) K (GPa) ry (GPa)
Projectile Copper 8800/8900 120 46.1 100 0.48/0.3
Concrete 37.4 MPa 2300 29 12.1 16 0.0374
39 MPa 2209 27.9 11.6 15.4 0.039
51.9 MPa 2535 39.5 16.4 21.8 0.0519
172 5 Eroding Projectile and Shaped Charge Jet Penetrations

(a) 2.5 (b) 3.0

2.0 2.5

2.0
1.5
hpen/L

hpen/L
1.5
Cu 37.4MPa Concrete
1.0 A-T
Cu 39MPa Concrete
1.0 A-T
RMM RMM
WA
0.5 LW 0.5
AW
LW
Hydrodynamic Limit Hydrodynamic Limit
0.0 0.0
500 1000 1500 2000 2500 3000 500 750 1000 1250 1500 1750 2000
V0 (m/s) V0 (m/s)

(c) 3.0
2.5

2.0
hpen/L

1.5
Cu 51.9MPa Concrete
1.0 A-T
RMM
0.5 AW
LW
Hydrodynamic Limit
0.0
500 750 1000 1250 1500 1750 2000
V0 (m/s)

Fig. 5.3 Comparisons of predicted normalized DOP with the test data of the target strengths of
a 37.4 MPa (Gold et al. 1996), b 39 MPa (Nia et al. 2014), and c 51.9 MPa (Nia et al. 2014)

Besides, since the available test data is limited, the corresponding experiments for
the establishment and validation of the analytical model are needed.

5.2.3 Projectile Penetration Test

In this section, the projectile eroding penetration test into mortar targets is intro-
duced, and then, an analytical model for the eroding projectile penetration into
concrete targets is established.
Eighteen flat-nosed cylindrical projectiles are launched by a two-stage light-gas
gun to penetrate mortar targets with the broad striking velocities ranged from 510 to
1780 m/s. The experimental data including DOP, deformation and damage of the
projectile, cratering on the impact surface of the target, and the tunneling profile are
carefully recorded. The test provides sufficient data for the establishment of the
eroding projectile penetration model for the concrete targets.
5.2 Eroding Projectile Penetration 173

5.2.3.1 Projectiles, Sabots, and Obturators

In the impact tests, flat-nosed cylindrical projectiles with diameter of 6 mm, length
of 30 mm, and mass of 6.58 g are used. They are machined from Chinese 45# steel
rods and heat-treated to a hardness of HRc 11.5–11.9. Figure 5.4 presents the
quasi-static stress-strain curve for the projectile material obtained from uniaxial
tension test, in which the yield strength is obtained as 380 MPa.
A two-stage light-gas gun is used to launch the projectile to reach the striking
velocities ranged from 510 to 1850 m/s by adjusting the gas pressure and thickness
of rupture disk installed between the pump tube and gun barrel. Since the diameter
of the gun barrel (20 mm) is larger than that of the projectile, the subcaliber
technology is utilized. The projectiles are placed with a set of special designed
polylactide sabot and polycarbonate obturator which fit snugly into the gun barrel.
In order to prevent the projectile from the reverse penetration into the obturator
during launching, a steel disk with the thickness of 2 mm is stuck to the obturator
by glue. Figure 5.5a shows the separated projectile, obturator, and two pieces of
sabots, as well as the package processing. Figure 5.5b presents the dimensions of
the sabot and obturator. In order to ensure that the sabot and obturator can strip off
the projectile aerodynamically prior to the impact on the target, a cut angle at the
front end of the sabot is prefabricated.

5.2.3.2 Mortar Targets

Mortar targets are cast in 3-mm steel tubes with the diameter of 0.4 m and length of
0.3 m, which are large enough to avoid the rear free-surface and boundary influ-
ences. Before casting, a steel plate is welded on one side of the tube to ensure the
flatness of the impact side. The mixing ingredients’ ratios of the mortar targets are
Portland cement (P42.5R) 1018 kg/m3, water 408 kg/m3, and sand 1320 kg/m3,

Fig. 5.4 Stress-strain curve 900

for the 45# steel projectile Test data
800

700

600
Stress(MPa)

500
400

300

200

100

0
0 2 4 6 8 10 12 14 16
Strain (%)
174 5 Eroding Projectile and Shaped Charge Jet Penetrations

(a)
Cut angle

Sabot

Obturator
Projectile Steel disk

(b) Obturator Steel disk Sabot

20mm

14mm
16mm

2mm

10mm 4mm 20mm 6mm

40mm

Fig. 5.5 Projectile with sabot and obturator: a photograph and b dimensions

and the wet density of the mortar is 2100 kg/m3. The size of sand is controlled
strictly during the casting, and the maximum diameter of the sand is less than
2 mm.
In addition, nine control specimens (100  100  100 mm3 cubes) are cast and
the averaged unconfined compressive strength is obtained as 50 MPa. Strength and
penetration tests are conducted between 40 and 50 days after casting.

5.2.3.3 Test Setup

An advanced two-stage light-gas gun is used to launch the projectile. The diameters
of the pump tube and the gun barrel are 50 and 20 mm, respectively. As shown in
Fig. 5.6, the air is firstly pushed into the sealed vessel by a compressor until to a
5.2 Eroding Projectile Penetration 175

Pump tube filled Target

with hydrogen Gun barrel Laser chamber
Mushroom velocimeter
Air vessel

Rupture disk
Separation chamber filled High speed camera
with nitrogen

Fig. 5.6 Sketch of two-stage light-gas gun

designed pressure. Then, the valve of the air vessel is opened and the piston is
accelerated in the pump tube which is filled with high-pressure hydrogen gas. At the
end of the first stage, pump tube is a conical section, leading down to the
second-stage 20-mm gun barrel that fires the projectile. In this conical section, there
is a rupture disk with a “+” pattern scored onto the middle surface. When the
hydrogen develops sufficient pressure to burst the scored section of the rupture disk,
the hydrogen flows through the hole and accelerates the projectile with the sabot
and obturator located in the second-stage gun barrel. Then, after the projectile is
propelled out of the barrel muzzle and entered into the separation chamber, the
sabot and obturator are separated from the projectile aerodynamically in the sep-
aration chamber filled with nitrogen gas. In order to prevent the sabot impacting on
the target, the sabot discarding device, a mushroomed steel, is designed and fixed
on the end of the separation chamber.
The cylindrical mortar targets are placed with their top faces perpendicular to the
gun barrel. The striking velocity V0 is measured from a pair of laser beams that
placed in the straight trajectory of the projectile. The laser beams are interrupted
successively, which generate voltage changes. Therefore, dividing the distance
between two laser beams by the time interval between the voltage changes, the
impact velocity is calculated.
The impact processes of the projectiles are recorded by the high-speed camera
system, and Fig. 5.7 shows the typical photographs of the projectile during pene-
tration, which indicates that the projectile impacts the target perpendicularly. In
some shots, small yaw angles are found, which are determined approximately by
the high-speed camera images and given in Table 5.2.

Projectile Projectile

Obturator debris
Obturator debris

Fig. 5.7 Typical projectile impacting photographs from high-speed camera

176 5 Eroding Projectile and Shaped Charge Jet Penetrations

Table 5.2 Test data

Shot No. V0 (m/s) Yaw angle (°) DOP (mm) Crater size
Dc (cm) Hc (cm) Vc (cm3)
1 1778 1.5 80.3 19.4 3.15 209
2 1016 5.1 49.2 11.8 2.15 94
3 1373 4.5 72.7 13.8 2.80 141
4 1288 1.6 65.0 13.2 2.15 111
5 1559 4.1 81.4 16.8 2.90 197
6 1855 6.9 79.7 23.5 3.40 489
13 1457 3.4 70.5 22.9 2.49 293
14 510 1.6 31.0 6.1 1.25 15
16 997 2.2 55.8 12.7 1.76 90
18 866 5.8 47.0 12.8 1.41 71
20 679 0.7 46.6 7.2 1.34 24
21 648 3.2 45.6 6.3 1.28 13
22 807 1.6 47.5 9.2 1.70 36
23 710 0.8 56.1 11.2 1.46 58
24 1134 0.1 58.6 13.9 2.47 144
25 1332 2.3 64.6 16.6 2.27 200
26 703 3.8 46.8 9.1 1.51 40
29 1084 1.5 63.1 13.6 2.10 128

5.2.3.4 Test Results

The test data is listed in Table 5.2, including the striking velocity V0, initial yaw
angle, DOP, and the front crater size. Hc, Vc, and Dc represent the crater depth,
crater volume, and equivalent diameter of the crater, respectively. Projectile pen-
etration into concrete targets always forms the frustum-of-cone-shaped crater fol-
lowed by a cylindrical tunnel. Referred to Wu et al. (2015a), the equivalent
diameter Dc is obtained by averaging the maximal dimensions of the crater surface
from four directions, i.e., horizontal, vertical, 45°, and 135°, respectively. As
suggested by Wu et al. (2015d), the sand filling approach is adopted to measure the
crater volume Vc.
(1) Damage of targets
Figure 5.8 shows the damages of the mortar targets, and the shot numbers are
written on the label. The localized damages occurred on the impact surface, which
validates that the dimensions of the targets are large enough to neglect the boundary
influences.
It is well known that the diameter of the tunnel is equal to that of the projectile
when the projectile velocity is relatively low and the projectile can be treated as a
rigid body. However, when the projectile velocity is large enough, it is found that
5.2 Eroding Projectile Penetration 177

Fig. 5.8 Frontal damage of the targets

the tunnel diameter is always larger than the projectile diameter. This observation is
also found in the eroding penetration tests into metallic targets (Forrestal and
Piekutowski 2000). In order to obtain the profile of the tunnel, a series cylindrical
gauge probes with different diameters are used to insert into the tunnel and measure
the depth from the level of the original surface. Figure 5.9 presents the typical
tunnel profiles for Shots 1 and 6, respectively.
In order to capture the shape and dimension of the crater more precisely, 3D
scanning of the crater is conducted. Figure 5.10 presents typical scanning results,
where the damage of the target and dimension of the crater can be clearly observed.
The measured profile and 3D scanning of the crater can be further used to evaluate
the concrete material models for numerically predicting the catering phenomenon.
178 5 Eroding Projectile and Shaped Charge Jet Penetrations

(a) 0 (b) 0
10 10
Crater depth
20 20
30 30

Y(cm)
Y(cm)

40 40
50 Tunnel depth 50
60 60
70 70
80 80
-120 -80 -40 0 40 80 120 -120 -80 -40 0 40 80 120
X(cm) X(cm)

Fig. 5.9 Profiles of the crater and tunnel: a Shot 1 and b Shot 6

Fig. 5.10 3D scanning of the crater for Shot 13 from a oblique view and b cross-sectional view
5.2 Eroding Projectile Penetration 179

(2) Damage of projectiles

After the penetration tests, the targets are carefully cut to recover the residual
projectiles, as shown in Fig. 5.11. Figure 5.12 presents the photographs of the
unfired and recovered projectiles. The length and mass of the unfired projectile
(L0, M0), as well as the residual length, mass, and diameter of the recovered pro-
jectile (Lr, Mr, dr), are given in Table 5.3. It should be pointed out that the residual
diameter of the recovered projectile is measured at the mushroomed position.
Usually, the mushroom is not a regular hemisphere; thus, the diameter is deter-
mined by averaging the maximum and minimum diameters of the mushroom.
It can be observed that when the striking velocity of projectile is lower than
710 m/s, the maximum relative mass loss and blunted length of the projectile are
less than 0.30 and 2.8%, respectively. Correspondingly, as shown in Fig. 5.12, the
deformation of the projectile is minor and can be neglected. Therefore, the

Projectile

Projectile

Fig. 5.11 Photograph of a cutting target

Deforming penetration
Rigid penetration without eroding Eroding penetration Velocity

Fig. 5.12 Photograph of unfired and recovered projectiles

180 5 Eroding Projectile and Shaped Charge Jet Penetrations

Table 5.3 Masses and dimensions of unfired and recovered projectiles

Shot No. V0 (m/s) DOP (mm) Projectile
M0 (g) Mr (g) L0 (mm) Lr (mm) dr (mm)
1 1778 80.3 6.58 2.83 30 8.82 9.84
2 1016 49.2 6.59 6.16 30 20.32 9.74
3 1373 72.7 6.59 3.97 30 16.77 9.46
4 1288 65.0 6.58 4.62 30 18.92 9.81
5 1559 81.4 6.59 Fractured 30 Fractured
6 1855 79.7 6.56 3.03 30 7.24 9.75
13 1457 70.5 6.58 3.51 30 14.53 8.84
14 510 31.0 6.59 6.59 30 29.86 6.35
16 997 55.8 6.59 6.40 30 22.10 9.09
18 866 47.0 6.59 6.45 30 24.42 8.13
20 679 46.6 6.54 6.51 30 29.23 6.91
21 648 45.6 6.58 6.53 30 29.31 6.92
22 807 47.5 6.57 6.54 30 26.60 8.15
23 710 56.1 6.59 6.57 30 29.16 6.73
24 1134 58.6 6.59 6.22 30 19.16 9.17
25 1332 64.6 6.60 4.60 30 16.37 9.98
26 703 46.8 6.57 6.55 30 28.21 7.01
29 1084 63.1 6.60 6.31 30 21.49 8.43

projectile with the initial velocity less than 710 m/s in the present test can be treated
as a rigid body approximately.
When the projectile initial velocity is larger than 710 m/s and less than 997 m/s,
the deformation of the projectile becomes evident, as shown in Fig. 5.12. The
relative blunted length of the projectile reaches to 26.3%, while the relative mass
loss is only up to 2.9%. This indicates that the projectile is deformed but not eroded
during the penetration.
However, if the projectile initial velocity is larger than 997 m/s, the relative mass
loss is exacerbated. As observed in Fig. 5.12 and Table 5.3, the mass loss and
residual length of the projectile increase and decrease with the rising of impact
velocity, respectively. This stage is termed as eroding penetration.
The above three penetration regimes, i.e., rigid projectile penetration, deforming
projectile penetration without eroding, and eroding projectile penetration, are also
observed in the hemispherical nosed 4340 steel projectile penetrating into the
aluminum alloy targets (Forrestal and Piekutowski 2000), which are further dis-
cussed by Chen and Li (2004).
(3) Crater dimensions
Figure 5.13 presents the influences of impact velocity on the crater diameter and
volume, respectively. It is found that the non-dimensional equivalent crater diam-
eter (Dc/d) increases almost linearly with the increase of the impact velocity.
5.2 Eroding Projectile Penetration 181

(a) 50 (b) 500

Test data Test data
45 450
Linear fit (0.975+0.019V0 ) Linear fit (110V02 -21)
40 400
35 350
30 300

Vc (cm3)
Dc/d

25 250
20 200
15 150
10 100
5 50
0 0
400 600 800 1000 1200 1400 1600 1800 2000 0.4 0.8 1.2 1.6 2.0 2.4 2.8 3.2 3.6
V0 (m/s) V02(km2/s2 )

Fig. 5.13 Influences of impact velocity on a non-dimensional equivalent crater diameter and
b crater volume

Similar conclusion is drawn in the flat-nosed projectile penetration test into the
UHPCC target (Peng et al. 2016b). While the crater volume increases almost lin-
early with the increase of the square of impact velocity, it indicates that the crater
volume is proportional to the initial impact kinetic energy of the projectile.
For the ogive-nosed projectile penetrating into the concrete targets, Forrestal
et al. (1994) and Li and Chen (2003) proposed that the crater depth is proportional
to the shank diameter, i.e., Hc = kd. However, for the high-speed flat-nosed pro-
jectile in the present test, it is observed that the non-dimensional crater depth
increases almost linearly with the increase of the initial projectile velocity, as shown
in Fig. 5.14. It can be drawn that for the high-speed penetrations, the non-
dimensional crater depth is related to the impact velocity and projectile nose shape.

Fig. 5.14 Influence of 6

impact velocity on the Test data
non-dimensional crater depth Linear fit (0.0027V0 +0.5 )
5

4
Hc/d

1
400 600 800 1000 1200 1400 1600 1800 2000
V0 (m/s)
182 5 Eroding Projectile and Shaped Charge Jet Penetrations

5.2.4 Penetration Models

5.2.4.1 Rigid Penetration Model

An extended cavity expansion model is established in Sect. 3.3 (Kong et al. 2017),
in which a hyperbolic yield criterion and the nonlinear Murnaghan EOS are
introduced to describe the plastic behavior of concrete material under projectile
penetration. Moreover, a one-dimensional resistance of concrete targets with con-
sidering the projectile nose shape influences is presented. Furthermore, by com-
bining with the one-dimensional momentum theorem, the rigid projectile
penetration model is proposed. In this section, the rigid penetration data in the
present test is analyzed by the above rigid penetration model.
As discussed previously, for the present test, when the initial impact velocity of
the projectiles is less than 710 m/s, the projectile can be treated as a rigid body
approximately. For flat-nosed projectile, it is proved that cavity expansion theory
can be applied to whole penetration process including the cratering and tunneling
stages (Teland and Sjøl 2004). Consequently, the rigid penetration model estab-
lished in Sect. 3.3 can be utilized directly.
Since the parameters of the hyperbolic yield criterion and Murnaghan EOS are
absent in the present DOP experiments, a0, a1, and a2 suggested by Malvar et al.
(1997), as well as C0 and s obtained by data fitting illustrated in Fig. 3.2b, are used,
i.e.,

a0 ¼ 0:2956fc ; a1 ¼ 0:4463; a2 ¼ 0:0808=fc ð5:11aÞ

C0 ¼ 1945 m=s; s ¼ 2:76 ð5:11bÞ

where fc (unit in Pa) is the uniaxial compressive strength of the mortar targets.
Figure 5.15 presents the prediction of non-dimensional DOP, and it is observed
that the prediction agrees well with the experimental data.

Fig. 5.15 Test data and 2.0

predicted non-dimensional Test data
DOP Model in Sec. 3.3
1.6

1.2
hpen /L0

0.8

0.4

0.0
0 100 200 300 400 500 600 700 800
V0 (m/s)
5.2 Eroding Projectile Penetration 183

5.2.4.2 Eroding Penetration Model

As discussed in Sect. 5.2.3, if the projectile initial velocity is larger than 997 m/s,
the projectile starts to erode. Section 5.2.2 demonstrates that the existing projectile
penetration models for the metallic targets cannot be used to predict DOP of the
concrete targets, which is mainly due to the different responses of the metallic and
concrete materials subjected to projectile impact. In this section, a new eroding
projectile penetration model for the concrete targets is established, which is then
used to predict the above-mentioned eroding penetration test data.
The eroding projectile penetration process is usually described by the 1D
modified hydrodynamic Alekseevskii–Tate (A-T) equation given in Eq. (5.1). It is
generally accepted that Yp equals the HEL, i.e., Yp ¼ ryp ð1
mÞ=ð1
2mÞ with ryp
and m being the dynamic yield strength and Poisson’s ratio of the projectile,
respectively. While Rt has not explicit expression for the concrete targets, it must be
obtained by fitting experimental data, which reduces the predictive capability of the
model. For example, based on the spherical-nosed copper and tantalum projectile
penetration tests on 37.4 MPa concrete targets, Gold et al. (1996) computed the
DOP by changing the value of Rt and adopted Rt = 0.426 GPa which seems to give
best agreements with the test data.
To make the A-T equation applicable for the eroding projectile penetration into
the concrete targets, the unified one-dimensional target resistance given in
Eq. (3.36) is combined with the A-T equation from Eq. (5.1) to describe the
pressure balance at the projectile/target interface
pffiffiffiffiffiffiffiffi
qp ðv
uÞ2 =2 þ Yp ¼ N 0 Afc þ N 1 B q0 fc u þ N 2 Cq0 u2 ð5:12Þ

As observed in Fig. 5.12, the nose shape of the eroded projectile is almost
hemispherical. Consequently, the nose shape of the instantaneous eroding projectile
is assumed to be hemispherical; thus, N 0 ¼ 1, N 1 ¼ 2=3, and N 2 ¼ 1=2 from
Eq. (3.32). The deceleration and erosion rate of the projectile presented by
Eqs. (5.2 and 5.3) are still used. Generally, Eqs. (5.2, 5.3 and 5.12) can be solved
numerically to obtain the instantaneous and terminal DOPs, as well as the instan-
taneous projectile lengths.
Based on the relative values of Yp and Afc, there are two different scenarios
discussed as follows:
(i) If Afc  Yp ; the projectile always behaves as fluid until penetration ceases (i.e.,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
u = 0, v ¼ 2 Afc
Yp =qp ). Then, the target will become a rigid body and
the projectile will deform plastically as what happens in a Taylor test, which
will have no contribution to penetration.
(ii) If Afc \Yp , the projectile behaves as fluid until v drops to
 
pffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2  
v ¼ VR ¼
BN 1 q0 fc þ BN 1 q0 fc þ 4Cq0 Yp
AN 0 fc =2N 2 Cq0 , from
which v = u and the projectile penetrates continuously as a rigid body with the
184 5 Eroding Projectile and Shaped Charge Jet Penetrations

initial striking velocity of v = VR. Thus, the rigid penetration model estab-
lished in Sect. 3.3 is used; therefore,

pffiffiffiffiffiffiffiffi

qp Lreff dv=dt ¼
N 0 Afc þ N 1 B q0 fc v þ N 2 Cq0 v2 ð5:13Þ

 
where Lreff ¼ 4Mr = qp pdr2 is the residual effective length of projectile when the
erosion penetration ceases. Mr and dr are the residual projectile mass and diameter
of mushroomed projectile nose. Using the equation of dv=dt ¼ vdv=dh, the
non-dimensional DOP for the rigid stage hpen;r can be obtained by integrating
Eq. (5.13) as follows:
pffiffiffiffiffiffiffiffi ! pffiffiffiffiffiffiffiffi
hpen;r qp Afc N 0 þ N 1 B q0 fc Vr þ N 2 Cq0 Vr2 qp N 1 B q0 fc
¼ ln þ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

Lreff 2N 2 Cq0 Afc N 0 2

N 2 Cq0 q0 fc 4ACN 0 N 2
N 1 B2
2 0 1 0 13
pffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi
6 B N 1 B q0 fc C B C7
6arctanBrffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi C B N 1 B q0 fc þ 2N 2 Cq0 Vr C7
4 @
A
arctan@rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi

A5
2 2
q0 fc 4ACN 0 N 2
N 1 B2 q0 fc 4ACN 0 N 2
N 1 B2

ð5:14Þ

The total DOP is determined by adding the eroding DOP and rigid DOP.

5.2.4.3 Comparison with Test Data

Previous eroding projectile penetration tests on concrete targets are very limited.
Nia et al. (2014) and Gold et al. (1996) conducted the copper projectile penetration
tests on concrete targets with the compressive strengths ranged from 37.4 to
65.6 MPa, respectively. The instantaneous nose shape of the projectile is assumed
to be hemispherical, as observed in Nia et al. (2014) and our experiments. The
dynamic yield stress and Poisson’s ratio of the projectile material are taken as
380 MPa and 0.3, respectively (Johnson and Cook 1983).
Since Afc \Yp is satisfied, there exists both eroding penetration stage and
residual rigid penetration stage, as discussed previously. Figure 5.16 shows the
experimental DOP, as well as the prediction results from Eqs. (5.11a, b–5.14).
Obviously, a good agreement is observed.
The test introduced in Sect. 5.2.3 provides sufficient data for the validation of the
proposed eroding penetration model. Since the parameters of the yield criterion and
EOS are absent in the present DOP experiments, Eqs. (5.11a, b) are used. According
to the dynamic stress-strain curve at the strain rate of 4000 s−1 (Chen et al. 2005),
which is obtained from the SHPB test, the dynamic yield stress of the 45# steel of
projectile is taken as ryp ¼ 1 GPa, and the Poisson’s ratio is taken as 0.3.
5.2 Eroding Projectile Penetration 185

(a) 3.0 (b) 3.0

Nia et al. (2014) Nia et al. (2014)
2.5 Total DOP Total DOP
2.5
Eroding DOP Eroding DOP
Rigid DOP Rigid DOP
2.0 2.0

hpen/Leff
hpen/Leff

1.5 1.5

1.0 1.0

0.5 0.5

0.0 0.0
400 600 800 1000 1200 1400 1600 1800 2000 400 800 1200 1600 2000
V0 (m/s) V0 (m/s)
(c) 3.0 (d) 3.0
Nia et al. (2014) Nia et al. (2014)
2.5 Total DOP Total DOP
Eroding DOP 2.5
Eroding DOP
Rigid DOP Rigid DOP
2.0 2.0
hpen/Leff

hpen/Leff

1.5 1.5

1.0 1.0

0.5 0.5
0.0
0.0
400 600 800 1000 1200 1400 1600 1800 2000
400 600 800 1000 1200 1400 1600 1800 2000
V (m/s) V0 (m/s)
0

(e) 3.0
Gold et al. (1996)
2.5 Total DOP
Eroding DOP
Rigid DOP
2.0
hpen/Leff

1.5

1.0

0.5

0.0
400 800 1200 1600 2000
V0 (m/s)

Fig. 5.16 Experimental and predicted non-dimensional DOP with target strengths of a 39 MPa,
b 44.1 MPa, c 51.9 MPa, d 65.6 MPa, and e 37.4 MPa

The inequation Afc \Yp is also satisfied; thus, there are the successive eroding
and residual rigid penetration stages. The DOP of eroding projectile is determined
using the parameters given above. While for the rigid stage, the diameter of
mushroomed projectile nose dr is needed. Checking the experimental dr in
Table 5.3, it is assumed that dr = 1.5d0 (similar with which used in Fig. 5.16).
Therefore, the DOP of rigid stage can be calculated.
186 5 Eroding Projectile and Shaped Charge Jet Penetrations

Fig. 5.17 Predicted 3.0

non-dimensional DOP and
test data 2.5

2.0

hpen/L0
1.5
Test data
Total DOP
1.0 Eroding DOP
Rigid DOP
0.5

0.0
900 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900
V0 (m/s)

Fig. 5.18 Predicted 1.0

non-dimensional residual Test data
length of projectile and test Present model
data 0.8

0.6
Lreff /L0

0.4

0.2

0.0
1000 1200 1400 1600 1800 2000
V0 (m/s)

Figure 5.17 shows the predicted DOP and the corresponding test data, in which
good agreements are observed. Comparisons of predicted non-dimensional residual
length of projectile with the corresponding test data are shown in Fig. 5.18, where
reasonable agreements are also observed. The overestimations of projectile residual
lengths for relatively low velocities may lie in that the proposed model is
one-dimensional and the residual projectile lengths are measured from the
straight-line distance between the projectile tip and tail, while these projectiles are
bended slightly, as shown in Fig. 5.12.
5.2 Eroding Projectile Penetration 187

5.2.4.4 Further Discussion on Deforming Penetration

As shown in Fig. 5.12, there exists a transition penetration phase between the rigid
penetration and eroding penetration stages, i.e., the deforming penetration without
eroding. In this transition phase, the DOP decreases with the increase of the pro-
jectile velocity. It is interesting to determine the velocity range of the transition
region, which may give a reference for the designs of weapons and protective
structures. Therefore, the upper limit velocity of the rigid penetration (VR) and the
lower limit velocity of the eroding penetration (VH) are discussed in this section.
Since the projectile is confined by the surrounding target material, the projectile
starts to deform when the pressure at the projectile/target interface reaches the
elastic limit of the projectile material under one-dimensional strain condition (Yp).
The pressure at the projectile/target interface is determined by the one-dimensional
resistance of concrete targets; thus,
pffiffiffiffiffiffiffiffi
Yp ¼ N 0 Afc þ N 1 B q0 fc VR þ N 2 Cq0 VR2 ð5:15Þ

Solving Eq. (5.15) gives VR as

 
pffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2  
VR ¼
BN 1 q0 fc þ BN 1 q0 fc þ 4Cq0 Yp
AN 0 fc =2N 2 Cq0 ð5:16Þ

It is assumed that when the eroding rate is larger than the deforming rate of the
projectile (plastic wave velocity), the projectile starts to erode (Tate 1977).
Therefore,
qffiffiffiffiffiffiffiffiffiffiffiffi
VH
u ¼ Eh =qp ð5:17Þ
pffiffiffiffiffiffiffiffi
qp ðVH
uÞ2 =2 þ Yp ¼ N 0 Afc þ N 1 B q0 fc u þ N 2 Cq0 u2 ð5:18Þ

where Eh is the dynamic hardening module of the projectile material. Referring to

Chen et al. (2005), Eh is about 700 MPa for 45# steel material. Then, the lower
limit velocity of eroding penetration VH is obtained. Using Eqs. (5.16 and 5.18), VR
and VH are obtained as 840 and 971 m/s, respectively.
Figure 5.19 presents the predicted DOP for the whole range velocity, as well as
the predicted VR and VH. It is observed that the predicted DOP agrees excellently
with the experimental data, and the predicted VR and VH show reasonable agree-
ment with the test data. Therefore, the proposed eroding penetration model is
validated. Due to the limited test data located in the transition region, no attempt is
made to establish a proper model for the transition region at present. However, the
penetration mechanism of deforming projectile without eroding should be studied
experimentally and theoretically in the future.
188 5 Eroding Projectile and Shaped Charge Jet Penetrations

Fig. 5.19 Comparison of 4.0

predicted DOP and critical Test data
velocities with test data 3.5 Rigid penetration
Eroding penetration
3.0
VR VH
2.5

hpen/L0
2.0

1.5
Deforming penetration
1.0

0.5 Rigid penetration Eroding penetration

0.0
400 600 800 1000 1200 1400 1600 1800
V0 (m/s)

5.3 Shaped Charge (SC) Jet Penetration

5.3.1 A Review of the Existing Studies

SC is often equipped with the armor-piercing projectile to attack the metallic armor,
as well as the ceramic/metal composite targets, e.g., aircrafts, ships, and tanks. Most
efforts have been paid into the optimization of SC to create a larger and deeper
penetration borehole, from the perspectives of the explosive types (e.g., Octol,
PBX, and C-4), liner materials (e.g., aluminum, copper, molybdenum, tungsten, and
tantalum), charge structures (e.g., conical, hemispherical, and K liner), as well as
the stability (not fragmentation) of the SC jet (Samudre et al. 2009; Saran et al.
2013; Fu et al. 2013; Ayisit 2008).
As for concrete targets, Wang et al. (2008) experimentally studied the influences
of cone angles and liner materials on the penetration performance of the SC jet into
the concrete target and derived that the penetration depth and borehole diameter
increased and reduced quadratically with the rising of the jet tip velocity, respec-
tively. However, it should be noted that the above conclusion is derived by
adjusting liner angles without changing the diameters and heights of SC, while the
standoff distance of SC is fixed; thus, the corresponding weights of explosive and
liner change consequently. Murphy (1983) conducted the SC jet penetration tests
into concrete targets (20 MPa) with the standoff distances, liner angles and thick-
nesses varied within 0.75–3 times of charge diameter (CD), 75°–120° and 2–8% of
CD, respectively. The linear relationship between the jet energy and target borehole
volume is obtained experimentally, which is further validated by the numerical
simulations. By comparing the rear crater dimensions of the perforation borehole in
the concrete slabs impacted by the SC jet, Resnyansky and Weckert (2009) con-
cluded that the ultra-high-performance concrete (200 MPa) exhibited higher jet
impact resistance than the plain concrete (40 MPa). With regard to the oil well
5.3 Shaped Charge (SC) Jet Penetration 189

penetrator, by experimentally evaluating the influence of the confining pressures

(40 and 68 MPa) on the performance of the SC jet penetrating into concrete targets
(26–55 MPa), Elshenawy and Li (2013) further introduced the target strength and
confining pressure into Allison–Vitalli equation and obtained better predictions for
the jet penetration depth. Besides, Grove et al. (2008) conducted the oil well charge
perforating test in sandstone cores with various confining pressures (nearly 7–
70 MPa) and pore pressures (nearly 0–63 MPa) and proposed the parabolic rela-
tionship between the penetration depth and effective stress consisted of confining
pressure and pore pressure. Huerta and Vigil (2006) used the SCAP (shaped charge
analysis program) code developed by Sandia National Laboratory (SNL) to design a
nearly world’s largest conical SC with the diameter of 0.7 m, and the field pene-
tration test on tuff rock is further performed. The numerical simulations are con-
ducted with the 2D Eulerian code, and a little larger penetration depth is obtained
compared with the experimental data.
Chanteret and Reck (2013) conducted a systematic experiment of SC jet pene-
trating into masonry materials (e.g., light concrete, adobe, clinker, and face cladding
brick wall) with the compressive strengths ranged from nearly 5 to 90 MPa,
respectively. The penetration borehole profiles including the depth and diameter are
presented. It is derived that the ratios of the experimental DOP of the above
masonry materials to that of the rolled homogeneous armor (RHA) target are larger
than the theoretical predictions by the classical Bernoulli’s equation. While for the
initial cratering, the hydrodynamic theory could well predict the jet consumption
and cratering progression as long as the jet is not severely fragmented. Chang et al.
(2009) experimentally studied the penetration efficiency on the masonry wall by
using three typical SC (e.g., hemispherical, large cone angle, and nonlinear liners).
The linear relationship between the jet energy and target borehole volume is
obtained, and the linear proportional coefficients of the above three SC are obtained
experimentally.
Wang et al. (2015) experimentally and numerically discussed the influences of
liner material, cone angle, liner thickness, and standoff distance on the perfor-
mances of SC jet penetrating into concrete targets (48 MPa), respectively.
Furthermore, the depth and diameter of the SC jet penetration borehole are pre-
dicted based on the modified Bernoulli’s equation and two-stage cavity expansion
theory, respectively. Murphy et al. (2000, 2003) conducted the SC jet penetration
tests into the granite and high-strength concrete targets, respectively. By estab-
lishing the porous equation of state (EOS) for concrete materials, the corresponding
numerical simulations are performed by using CALE code, a 2D arbitrary
Lagrange–Euler hydrocode developed by Lawrence Livermore National
Laboratory. The validity of the porous EOS is verified, and the diameter of pene-
tration borehole is confirmed very sensitive to the compressive strength of the
concrete targets.
Generally, due to the complexity and high costs of field tests, limited experi-
mental investigations are conducted to study the penetration performance of SC jet
into concrete targets systematically. Even for the available tests, the valuable
information, including the shape of SC jet from flash X-ray (FXR) observations and
190 5 Eroding Projectile and Shaped Charge Jet Penetrations

profile of the penetration borehole, is not completely enough. Therefore, the

numerical simulations have becoming the reliable and economic approach for
designing the structure of SC, as well as evaluating the ballistic performance of SC
jet penetrating into concrete targets. However, the related works are rarely
published.

5.3.2 Numerical Simulation

5.3.2.1 Numerical Model

The typical setup of SC jet penetrating into the concrete target is schematically
shown in Fig. 5.20. As shown in Fig. 5.20a, the cylindrical charge with the height
h (cm) and diameter d (cm) is located above the target with the standoff distance of
H (cm). The explosive is enveloped by the metallic casing and cone liner with the
thickness of s (cm) and d (cm), respectively. The inner angle of the cone liner is
denoted as h (°). The concrete target with the dimension of Dt (cm, diameter of the
cylindrical target or width of the cubic target) and thickness of T (cm) is surrounded
by a steel culvert, and the diameter Dt is always 30 times larger than the shaped
charge diameter d, and thus, the circumferential boundary effects can be neglected.
The general process of the SC jet formation and penetration is as follows: The
inner explosive is detonated by a booster, and then, a shock wave propagates at the
Chapman–Jouguet (C-J) pressure which is larger than 30 GPa. When the shock
wave spreads to the conical cavity, the cone collapses and the metallic liner is
accelerated inward normally. A large portion of liner materials converges along the
central line and forms a jet with an instantaneous length l (cm) ejecting outward at a
velocity as high as 8 km/s. As shown in Fig. 5.20b, when the SC jet with the high
temperature, high pressure, and high speed hits the target, the target material is
forced to spall outward and flow radially, and a penetration borehole with the crater
diameter Dc (cm) and average diameter of Dh (cm) are formed. DOP (cm) repre-
sents the final penetration depth of a SC jet. For some situations, the wave shaper,
which is an axial symmetrical inert obstacle, is placed between the booster and the
metallic liner of the SC to alter the angle of detonation wave on the liner and
provide cumulative convergence of the detonation wave along the symmetric axis
and therefore to further improve the SC jet penetration performance significantly.

5.3.2.2 Material Models

Since the setup of SC penetration scenario given in Fig. 5.20a is an axial sym-
metrical structure, to shorten the calculating time and guarantee the accuracy, 1/4
model is adopted in the finite element modeling, as shown in Fig. 5.21. The model
is discretized by with 8-node SOLID164 element, and the total number of the
elements is approximately up to 1,000,000. After examining the mesh density
5.3 Shaped Charge (SC) Jet Penetration 191

(a)
Shaped charge d

Target Wave shaper

H Booster

Explosive
Casing

h τ
θ
T
δ

Steel culvert Liner

Dt

SC jet
(b)
l
Target

DOP

Dh

Dc

Fig. 5.20 Typical setup and dimensions of SC jet penetration test: a pretest and b post-test

influences, the meshes of 0.2 and 0.4 cm are selected for ALE materials (e.g.,
explosive, air, liner, and casing) and the Lagrange target, respectively. The basic
unit of the model is cm-g-ls-K.
(1) Explosive
The explosive is described by the material Type 8 MAT_HIGH_EXPLOSIVE_
BURN in LS-DYNA. The EOS for the high explosive is JWL equation (Hallquist
2007) given in Eq. (5.19), which can describe the relationship of pressure, volume,
and energy features of gaseous product created by explosive explosion.
192 5 Eroding Projectile and Shaped Charge Jet Penetrations

Shaped charge
Air
Target
Explosive

Casing

Liner

Air

Fig. 5.21 Finite element model and meshes



x x xE
P¼A 1
e
R1 V þ B 1
e
R2 V þ ð5:19Þ
R1 V R2 V V

where P is the pressure, A, B, R1, R2, and x are materials constants, E is the
detonation energy per unit volume, and V represents the relative volume.
(2) Liner and casing
In order to describe the strength behavior of metallic liner and casing materials
subjected to large strains, high temperatures, and high strain rates, the Johnson–
Cook (JC) (Johnson and Cook 1983) model (material Type 15
MAT_JOHNSON_COOK in LS-DYNA) given in Eq. (5.20) is adopted.

ry ¼ A þ Benp ð1 þ c ln e_  Þð1
T m Þ ð5:20Þ

where A, B, n, c, and m are materials constants, ry is the flow stress, and ep

represents the plastic strain. e_  ¼ e_ p =_e0 is the effective plastic strain rate in which e_ 0
is the reference plastic strain rate and e_ 0 ¼ 1 s
1 is usually assumed. T  ¼
ðT
Troom Þ=ðTmelt
Troom Þ is the homologous temperature, where Tmelt and Troom
denote the melting and room temperatures, respectively.
The Gruneisen EOS is a fluid elastic-plastic model which can be used to simulate
the material of high strain rate (>105), where the pressure P for the compressed
materials is expressed as (Hallquist 2007)
   
q0 C2 l 1 þ 1
c20 l
a2 l2
P¼ þ ðc0 þ alÞE ð5:21Þ
1
ðs1
1Þl
s2 llþ 1
s3 ðl þl 1Þ2
2 3
5.3 Shaped Charge (SC) Jet Penetration 193

For the expanded materials, the pressure P is defined as

P ¼ q0 C 2 l þ ðc0 þ al ÞE ð5:22Þ

where C is the intercept of the vs–vp curve, in which the vs and vp denote cubic
shock velocity and particle velocity, respectively. s1, s2, and s3 are the dimen-
sionless coefficients of the vs–vp curve slop, respectively. c0 is the dimensionless
parameter, and a is the first-order volume correction to c0. In addition, l ¼
q=q0
1 represents the volumetric strain, where q and q0 represent the current and
initial densities, respectively. E denotes the internal energy.
(3) Air
The air is described by NULL material model and linear polynomial EOS, in which
the NULL material model (*MAT_NULL (009#) in LS-DYNA) can allow a vis-
cosity to be defined and the ignorance of deviatoric stresses. The behavior of the
fluid is defined by the function of the energy per unit volume E given in Eq. (5.23).
 
P ¼ c0 þ c1 l þ c2 l2 þ c3 l3 þ c4 þ c5 l þ c6 l2 E ð5:23Þ

where c0–c6 are coefficients.

(4) Concrete targets
Holmquist–Johnson–Cook (HJC) material model implemented in LS-DYNA as the
*MAT_JOHNSON_ HOLMQUIST_CONCRETE (111#) model is used to describe
the behavior of concrete material under large strains, high strain rates, and high
pressures. HJC model mainly includes three parts: yield surface equation, equation
of state (EOS), and damage model. Shown in Fig. 5.22a, the yield surface equation
of HJC model is defined as
 
r ¼ Að1
DÞ þ BPN ð1 þ C lnðe_  ÞÞ  Smax ð5:24Þ

where r ¼ r=fc represents the normalized equivalent stress, in which r and fc are
the actual equivalent stress and the quasi-static uniaxial compressive strength,
respectively. P ¼ P=fc means the normalized pressure, and P denotes the actual
pressures. e_  ¼ e_ =_e0 represents the dimensionless strain rate, where e_ is the actual
strain rate and e_ 0 ¼ 1 s
1 is the reference strain rate. T  ¼ T=fc denotes the nor-
malized maximum tensile hydrostatic pressure, in which T is the maximum tensile
hydrostatic pressure. Parameters of A, B, and C denote the normalized cohesive
strength, the normalized pressure hardening coefficient, and the strain rate coeffi-
cient, respectively. Besides, D, N, and Smax are the damage parameter, the pressure
hardening exponent, and the normalized maximum strength, respectively.
Shown in Fig. 5.22b, the EOS of HJC model is separated into three response
regions as
194 5 Eroding Projectile and Shaped Charge Jet Penetrations

(a) (b)
P
C
D=0 (Undamaged) P = K 1μ + K 2 μ 2 + K 3 μ 3
Smax
σ * =σ / f c

μ − μ lock
μ=
1+μ lock
D=1 (Fractured)
*
> 1.0

*
=1.0 B
( μplock , Plock )
*
σ * = [ A(1 − D ) + BP * N ][1 + C ln( * )] Pcrush A
T (1 − D )
O
P* = P / fc T (1 − D ) μcrush μlock μ

(c)
( Δε p + Δμ p )
D=∑
(ε pf μ pf )
ε pf + μ pf

EFMIN

T* ε pf + μ pf = D1 ( P * + T * )D 2

P*

Fig. 5.22 HJC constitutive model: a yield surface equation, b EOS, and c damage model

8
< P ¼ Kl P\Pcrush
P ¼ Pcrush þ Klock ðl
lcrush Þ Pcrush \P\Plock ð5:25Þ
:
P ¼ K1 l þ K2 l2 þ K3 l3 P [ Plock

where K ¼ Pcrush =lcrush is the elastic bulk modulus and Pcrush and lcrush represent
the pressure and volumetric strain at the beginning of materials undergoing plastic
deformation, respectively. Klock ¼ ðPlock
Pcrush Þ=ðlplock
lcrush Þ and lplock rep-
resents the volumetric strain under the compacting hydrostatic pressure Plock . l ¼
ðl
llock Þ=ð1 þ llock Þ is the modified volumetric strain. In addition, K1, K2, and K3
are the materials constants.
The damage model shown in Fig. 5.22c takes the accumulated damage from
both the equivalent plastic strain and plastic volumetric strain into account, which is
defined as

X Dep þ Dlp
D¼ ð5:26Þ
efp þ lfp

efp þ lfp ¼ D1 ðP þ T  ÞD2  EFMIN ð5:27Þ

5.3 Shaped Charge (SC) Jet Penetration 195

where Dep and Dlp represent the equivalent plastic strain and plastic volumetric
strain during an integration cycle, respectively. efp and lfp are the equivalent plastic
strain and plastic volumetric strain under a constant pressure until fracture,
respectively. D1 and D2 are the damage constants, and EFMIN is the minimum
plastic strain that is used to suppress fracture from the weak tensile waves.
(5) Wave shaper
The wave shapers can be made of foam, concrete, air, plastic, metal, etc. For
instance, the material of wave shaper in the test conducted by Murphy (1983) is the
polyurethane foam. The NULL material model and Gruneisen EOS are used to
describe the wave shaper in the simulations.

5.3.2.3 Finite Element Analysis Method

The process of the SC jet penetration contains jet’s formation, elongation, and
penetration into targets, which includes the high strain rate, high pressure, and large
distortions. There are two classical algorithms to simulate the continuum problem,
the Lagrange and Euler algorithms. The Lagrange algorithm is widely used in
structural mechanics, in which the grid describing the materials changes as the
shape of materials deforms. The free surfaces and interfaces between different
materials can be easily tracked, while it is weak in describing the large distortion of
computational domain. The Euler algorithm is mainly applied to fluid dynamics, in
which the grid is fixed in the space and the deformed materials move through the
fixed grids. The large distortions in continuum motion can be handled, but it has a
low accuracy in defining the materials interfaces.
The ALE algorithm embedded in LS-DYNA was developed to combine the
advantages of both the Lagrange and the Euler algorithms. The ALE algorithm can
reconstruct the suitable material grid according to the boundary conditions of
deformed materials like the Lagrange algorithm, and the materials move in the fixed
space grids like the Euler algorithm. The ALE algorithm can describe the shape of
the continuum by avoiding the serious distortions; thus, the ALE algorithm can
accurately simulate the high strain rate, high pressure, and nonlinear deformation
characteristics. In this section, the air, explosive, liner, and casing are discretized by
multimaterial ALE elements and the target is discretized by Lagrange elements.
The interaction between the SC jet and target is realized by fluid-structure
interaction (FSI) method, which was developed to solve problems of the interaction
between the fluid and structures, especially in the fields of aeroelasticity. By cou-
pling the compressible fluid (e.g., water, air, explosive, and jet) to the stiff structures
(e.g., concrete and armor), FSI method could handle the flow and structural
equations with separate solvers.
196 5 Eroding Projectile and Shaped Charge Jet Penetrations

5.3.2.4 Penetration Process

The whole process of a SC jet penetration into the target, including the formation,
elongation, and penetration, is reproduced and shown in Fig. 5.23. Illustrated in
Fig. 5.23a, the metallic liner is driven by the gaseous product and the shock wave is
created by the explosive; thus, a high-speed jet is formed along the symmetric axis
of the liner. The velocity gradient exists in the axial direction; thus, the jet elongates
while flying, shown in Fig. 5.23b. Figure 5.23c illustrates the instant of the jet tip
impacting on the target. With the penetration process continuing, the DOP increases
and the velocity of the consumed jet reduces gradually. As shown in Fig. 5.23d,
during the penetration process, the SC jet may be broken due to the existence of
velocity gradient. Finally, in Fig. 5.23e, when the jet is totally consumed, the
penetration terminates and the DOP reaches its maximum value.

5.3.3 Comparison with Test Data

In this section, four sets of the existing SC jet penetration tests on concrete targets
are numerically simulated. Table 5.4 lists the related basic parameters of the SC and
targets, in which only the SC in the tests by Murphy (1983) contained the poly-
urethane foam wave shaper. The velocity, image, and DOP of the SC jet are
obtained from the measurements of flash X-rays (FXR) and the observations of the
post-test targets. The validations of constitutive model and the corresponding

(a) (b) (c)

Jet elongation Jet impact

(d) Jet fragmentation (e) Jet residual

Fig. 5.23 Typical process of the SC jet penetration into the concrete target: a formation,
b elongation, c impact, d penetration, and e termination
5.3 Shaped Charge (SC) Jet Penetration 197

Table 5.4 Parameters of the SC jet penetration tests on concrete targets

Test Shaped charge Targets
h (°) d (cm) d (cm) h (cm) s (cm) H (cm) fc (MPa) T (cm) D (cm)
Resnyansky 90 0.4 7.6 4.6 0.3 8 40 30 70
et al. (2004)
Huang et al. 120 0.13–0.14 5.6 5.6 – 16 44 50 57
(2008)
Huerta and 80 1.91 70 68.5 1.91 150 170 Semi-infinite
Vigil (2006)
Murphy 110 0.68 11.4 11.9 0.2 8.55–34.2 20 91.2 304.8
(1983)

material parameters, as well as the finite element algorithms, are assessed numer-
ically. The parameters of the constitutive model and the EOS of the explosive, liner
and casing, air, concrete target, and the wave shaper are listed in Tables 5.5, 5.6,
5.7, 5.8, and 5.9, respectively.

5.3.3.1 Resnyansky et al.’s Test

Resnyansky et al. (2004) conducted experiments of the SC jet penetration into

concrete targets, with the consideration of the influences of liner angle, liner
thickness, charge height, and standoff distance, respectively. The corresponding
parameters for the explosive B, aluminum liner, and casing, as well as the 40-MPa
concrete, are listed in Tables 5.5, 5.6, 5.7, and 5.8. They only presented the
elongation photograph of the jet at time 39.3 ls after the detonation initiation,
shown in Fig. 5.24a. Figure 5.24b shows the numerically simulated elongation of

Table 5.5 Parameters of high explosive burn model and JWL EOS of explosive
Parameters 8701 B Octol LX-14
Reference density q0 (kg/m3) 1710 1717 1810 1840
Detonation velocity D (m/s) 7980 7980 8480 8800
Chapman–Jouguet pressure PCJ (MPa) 2.86  104 2.95  104 3.24  104 3.7  104
Constant A (MPa) 5.2423  105 5.2423  105 7.486  105 8.524  105
Constant B (MPa) 7.768  10 3
7.68  10 3
1.338  10 4
1.802  104
Constant R1 4.2 4.2 4.5 4.6
Constant R2 1.1 1.1 1.2 1.3
Constant x 0.34 0.34 0.38 0.38
Detonation energy per unit volume E (J/m3) 8.5  109 8.5  109 9.6  109 1.02  1010
Relative volume V 1 1 1 1
Huang et al. (2008), Li et al. (2010), Resnyansky et al. (2004), ANSYS/Autodyn-2D and 3D (2007), Huerta and
Vigil (2006), Murphy (1983)
198 5 Eroding Projectile and Shaped Charge Jet Penetrations

Table 5.6 Parameters of JC model and Gruneisen EOS of metallic liner and casing
Parameters Copper Steel 1100-H aluminum 6061-T6 aluminum
Reference density q0 (kg/m3) 8960 7830 2710 2785
Yield strength A (MPa) 90 792 276 264
Hardening modulus B (MPa) 292 510 92 265
Hardening coefficient n 0.31 0.26 0.26 0.34
Strain rate sensitivity coefficient c 0.02 0.014 0.01 0.015
Thermal softening coefficient M 1.09 1.03 1.34 1
Melting temperature TM (K) 1360 1793 877 775
Room temperature TR (K) 293 293 293 293
Intercept C (m/s) 3940 4569 5350 5328
Coefficient s1 1.49 1.49 1.34 1.338
Coefficient s2 0.6 0 0 0
Coefficient s3 0 0 0 0
Gruneisen parameter c0 1.99 2.17 2 2
Internal energy E (J) 0 0 0 0
Constant a 0.47 0.46 1.5 1.5
Huang et al. (2008), Li et al. (2010), Elshenawy and Li (2013), Johnson and Cook (1983), Murphy (1983),
Steinberg (1980), ANSYS/Autodyn-2D and 3D (2007), Resnyansky and Weckert (2009)

Table 5.7 Parameters of NULL material model and linear polynomial EOS of air
Density q0 Coefficients Energy per unit volume
(kg/m3) c0 c1 c2 c3 c4 c5 c6 E (J/m3)
1.293 0 0 0 0 0.4 0.4 0 2.5  105
Alia and Souli (2006)

the jet at time 40 ls. Figure 5.24c further illustrates the overlay of the experimental
and simulated results (white dashed line) given in Fig. 5.24a, b, and a good
agreement is observed.

5.3.3.2 Huang et al.’s Test

Huang et al. (2008) studied the penetration performance of two kinds of SC (cone
and truncated liners) with large cone angle (120°) against the concrete target
experimentally, while the experiment data of the SC with truncated liner is selected
in this simulation. Correspondingly, the parameters of the explosive of 8701,
copper liner, and 44-MPa concrete target are also given in Tables 5.5, 5.6, 5.7, and
5.8. The elongation shape of the jet at 38.9 ls captured by FXR is shown in
Fig. 5.25a. Similarly, Fig. 5.25b, c illustrates the simulated results at 39 ls and the
overlays of experimental and simulated results (white dashed line), respectively.
Besides, the experimental data and simulated results are listed in Table 5.10,
including the instantaneous jet length l, the jet tip velocity Vh, and the tail velocity
Table 5.8 Parameters of HJC concrete material model
Parameters Concrete Concrete Tuff rock Concrete Concrete
Uniaxial compressive strength fc (MPa) 44 40 170 20 200
Density q0 (kg/m3) 2400 2400 1780 2400 2500
Shear modulus G (GPa) 14.86 15 6.543 12.5 24.3
Normalized cohesive strength A 0.79 0.79 0.55 0.79 0.30
Normalized pressure hardening B 1.6 1.6 1.77 1.6 1.73
Strain rate coefficient C 0.007 0.007 0.007 0.007 0.007
Pressure hardening exponent N 0.61 0.61 0.77 0.61 0.79
5.3 Shaped Charge (SC) Jet Penetration

Maximum tensile hydrostatic pressure T (MPa) 4 3.5 3.3 2.2 20

Normalized maximum strength Smax 7 7 17 7 7
Crushing pressure Pcrush (MPa) 14.7 13.3 56.7 6.9 66.7
Crushing volumetric strain Ucrush 7.4  10−4 6.7  10−4 6.23  10−4 4.1  10−4 2.1  10−3
Locking pressure Plock (MPa) 8  102 8  102 2.5  103 8  102 8  102
Locking volumetric strain Ulock 0.1 0.1 0.38 0.1 0.1
Damage constant D1 0.04 0.04 0.04 0.04 0.04
Damage constant D2 1 1 1 1 1
Pressure constant K1(MPa) 1.7  104 1.7  104 3.1  103 1.7  104 8.5  104
Pressure constant K2 (MPa) 3.8  104 3.8  104 6  103 3.8  104 −1.71  105
4 4 3 4
Pressure constant K3 (MPa) 2.98  10 2.98  10 8.4  10 2.98  10 2.08  105
Huang et al. (2008), Johnson et al. (1998), Resnyansky and Weckert (2009), Resnyansky et al. (2004), Murphy (1983), Béton (1993), Huerta and Vigil (2006),
Fang et al. (2014b), Ren et al. (2016)
199
200 5 Eroding Projectile and Shaped Charge Jet Penetrations

Table 5.9 Parameters of NULL material model and Gruneisen EOS of polyurethane foam
Density q0 Intercept Coefficients Gruneisen Internal energy
(kg/m3) C (m/s) s1 s2 s3 parameter c0 E (J)
320 2540 1.57 0 0 1.07 0
Schwalbe et al. (1996)

Fig. 5.24 Jet elongation photograph: a FXR photograph at time 39.3 ls, b simulation instant at
40 ls, and c simulated result overlays on FXR photograph

Fig. 5.25 Elongation photograph of a jet: a FXR photograph at 38.9 ls, b simulation instant at
39 ls, and c simulated result overlays on FXR photograph
5.3 Shaped Charge (SC) Jet Penetration 201

Table 5.10 Comparison of experiment data and simulation results

Results Time (ls) l (cm) Vh (m/s) Vt (m/s) Dc (cm) Dh (cm) DOP (cm)
Experiment 38.9 6.414 3827 1792 11.2 3.5 25.1
43.9 7.423
Simulation 39 5.520 3477 1970 8 3.37 25.0
44 6.214 3435 1977
Deviations – 13.9–16.3% 9.1–10.2 9.9–10.3% 28.6% 3.7% 0.4%

Vt at the time of 38.9 and 39 ls, as well as the final crater diameter Dc, average
penetration borehole diameter Dh, and DOP. Table 5.10 further gives the relative
deviations between the experimental data and numerical results, and good agree-
ments are observed for the tip and tail velocities of the jet, average borehole
diameter, and DOP. Larger deviation exists for the crater diameter, which may lie in
the inaccuracy of the HJC model to describe the tensile damage of the concrete
targets, and it will be discussed in Sect. 7.4.

5.3.3.3 Huerta and Vigil’s Test

By designing a SC with the diameter of 0.7 m, Huerta and Vigil (2006) conducted
the full-scale SC jet penetration test on the tuff rock target. The parameters of the
copper liner, 6061-T6 aluminum casing, Octol explosive, and 170-MPa tuff rock
are given in Tables 5.5, 5.6, 5.7, and 5.8. The meshes of 2 and 4 cm in the
numerical simulation are selected for the ALE and the Lagrange materials,
respectively. Figure 5.26a illustrates the test setup and post-test penetration bore-
hole of the target. Figure 5.26b further illustrates the comparison of experimental
and simulated penetration borehole profiles. It can be found that the experimental
and predicted penetration boreholes are in good agreement and the deviation of
which is about 0.6%.

5.3.3.4 Murphy’s Test

Murphy (1983) experimentally studied the formation process of a SC jet penetration

process and borehole profile, respectively. The influences of the SC structure and
standoff distance on the penetration performance of the SC jet into concrete targets
are discussed. The parameters for the 1100-H aluminum liner, LX-14 explosive,
steel casing, 20 MPa concrete target, and wave shaper are listed in Tables 5.5, 5.6,
5.7, 5.8, and 5.9. The standoff distance is expressed by a multiple of CD.
Figure 5.27 illustrates the predicted terminal penetration borehole profiles with the
different standoff distances, and it can be seen that the jet residuals adhere to the
wall of borehole as the penetration proceeds. Furthermore, the comparison between
the experimental and simulated SC jet penetration borehole profiles with the
202 5 Eroding Projectile and Shaped Charge Jet Penetrations

(a)

(b) -200 -150 -100 -50 0 50 100 150 200

50 50
0 0
-50 -50
-100 -100
-150 -150
-200 -200
DOP (cm)

-250 -250
-300 -300
-350 -350
-400 -400
-450 Experiment -450
-500 -500
Simulation
-550 -550
-600 -600
-650 -650
-200 -150 -100 -50 0 50 100 150 200
Radius (cm)

Fig. 5.26 Test data and simulation result: a test setup and post-test penetration borehole of the
target and b comparison of experimental and predicted penetration borehole profiles

different standoff distances is illustrated in Fig. 5.28. It is shown that the predicted
DOP of the SC jet is slightly smaller than the experimental data, and the corre-
sponding deviations are 19.4, 11.7, 7.0, and 13.5% for four standoff distances,
respectively. It can be also found that the average borehole diameters of the sim-
ulated results with the different standoff distances are smaller than the experimental
data, and the corresponding deviations are 2.7, 9.7, 14.5, and 34.3%, respectively.
In this section, the available SC jet penetration tests on concrete targets are
numerically simulated. By comparison with the elongation photograph of the SC jet
in the air captured by the FXR, jet tip and tail velocities before reaching the targets
as well as the penetration borehole profiles, rather good agreements between the
numerical results and the test data are observed. However, due to the inadequacy of
HJC model in tensile damage and strain softening, the crater dimensions of concrete
targets are not well reproduced.
5.3 Shaped Charge (SC) Jet Penetration 203

(a) Jet residuals (b) (c) (d)

Fig. 5.27 Predicted terminal penetration borehole profiles with standoff distances of a 0.75CD,
b 1CD, c 2CD, and d 3CD

(a) 10-20 -15 -10 -5 0 5 10 15 20

10
(b) 10-20 -15 -10 -5 0 5 10 15 20
10
0 0
0 0
-10 -10
-10 -10
-20 -20
-20 -20
-30
DOP (cm)

-30
DOP (cm)

-30 -30
-40 -40 -40
-40
-50 -50 -50 Experiment -50
Experiment
-60 Simulation -60 -60 Simulation -60

-70 -70 -70 -70

-80 -80 -80 -80

-20 -15 -10 -5 0 5 10 15 20 -20 -15 -10 -5 0 5 10 15 20
Radius (cm) Radius (cm)

(c) 10-20 -15 -10 -5 0 5 10 15 20

10
(d) 10-20 -15 -10 -5 0 5 10 15 20
10
0 0
0 0
-10 -10
-10 -10
-20 -20 -20 -20
DOP (cm)

-30 -30
DOP (cm)

-30 -30
-40 -40 -40 -40
-50 -50
-50 Experiment -50
Simulation -60 -60 Experiment -60
-60
Simulation
-70 -70 -70 -70
-80 -80 -80 -80
-20 -15 -10 -5 0 5 10 15 20 -20 -15 -10 -5 0 5 10 15 20
Radius (cm) Radius (cm)

Fig. 5.28 Comparisons of experimental and predicted penetration borehole profiles with standoff
distances of a 0.75CD, b 1CD, c 2CD, and d 3CD
204 5 Eroding Projectile and Shaped Charge Jet Penetrations

5.3.4 Parametric Analyses

In this section, based on the experimental parameters of Murphy (1983) given in

Sect. 5.3.3.4, the influences of the standoff distance, liner material, explosive type,
as well as the compressive strength of concrete targets on the ballistic performance
of the SC jet penetration into concrete targets, are discussed.

5.3.4.1 Influence of Standoff Distance

The penetrations of the SC jet with the standoff distances of 0.75CD to 3CD are
discussed in Sect. 5.3.3.4, as shown in Fig. 5.28. In this section, another three sets
of penetrations with the standoff distances of 0.5CD, 4CD, and 5CD are further
numerically simulated and complemented. Figure 5.29 presents the numerical
results of jet tip velocity at the beginning of penetrations with the different standoff
distances. Due to the drag force in the air acted on the high-speed jet, the jet tip
velocity when arriving at the target decreases about 10.4% (from 7.34 to 6.58 km/s)
with the standoff distance increasing from 0.5CD to 5CD.
Figure 5.30 illustrates the profiles of the terminal penetration boreholes at the
different standoff distances. Figures 5.31 and 5.32 show the relationships of the
DOP and average borehole diameter with the standoff distances, respectively. It is
found that the DOP and average diameter of the SC jet penetration borehole
increase and decrease with the rising of the standoff distance, respectively. In
addition, a large drop of the average borehole appears as the standoff distance
increases from 0.5CD to 1CD, and it slightly reduces with the increases in standoff
distance from 1CD to 5CD. The reason lies in that when the standoff distance is
small, the SC jet has relatively larger diameter since the metallic liner material may
not have enough time to converge completely. The fins of the SC jet impact on the
target, which leads to the larger crater and borehole diameters. Correspondingly,

Fig. 5.29 Jet tip velocity at 7.4

the beginning of penetrations
7.3
versus the standoff distances
7.2
7.1
Vh (km/s)

7.0
6.9
6.8
6.7
6.6
6.5
0 1 2 3 4 5 6
Standoff distance (CD)
5.3 Shaped Charge (SC) Jet Penetration 205

Fig. 5.30 Penetration -25 -20 -15 -10 -5 0 5 10 15 20

25
10 10
borehole profiles with
0 0
different standoff distances
-10 -10
-20 -20
-30 -30

DOP (cm)
-40 -40
-50 0.5CD -50
-60 0.75CD -60
1CD
-70 -70
2CD
-80 -80
3CD
-90 4CD -90
-100 5CD -100
-110 -110
-25 -20 -15 -10 -5 0 5 10 15 20 25
Radius (cm)

Fig. 5.31 DOP of a SC jet 100

with different standoff
distances 90

80
DOP (cm)

70

60

50 Experiment
Simulation
40 Deviation of 10%

30
0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0 5.5
Standoff distance (CD)

Fig. 5.32 Average borehole 10

diameter of a SC jet with
different standoff distances 9

8
Dh (cm)

3
0 1 2 3 4 5 6
Standoff distance (CD)
206 5 Eroding Projectile and Shaped Charge Jet Penetrations

when the standoff distance increases, more liner material is converged to the
direction of the jet movement, the jet stretches thinner, and more kinetic energy of
the jet is consumed to increase the depth of the borehole. Thus, the DOP increases
and the average borehole diameter reduces with the increase of standoff distance.
Also, it should be pointed out that when the standoff distance increases continu-
ously and reaches a certain value, the stretched jet may be fragmented, and the DOP
would get decreased reversely.

5.3.4.2 Influence of Target Strength

Figure 5.33 illustrates the borehole profiles of the SC jet penetrating into concrete
targets with different compressive strengths. The corresponding parameters of the
concrete targets are given in Table 5.8, and the standoff distance is 1CD. The jet tip
and tail velocities corresponding to Fig. 5.33 are 7.3 and 1.46 km/s, respectively. It
can be found that within the above velocity range, the DOP of the SC jet declines
slightly with the increase of compressive strength of the target, since only 12.5%
decrease of DOP occurs as the compressive strength of the target increases to ten
times (from 20 to 200 MPa), which coincides with the conclusion drawn by
Murphy et al. (2000).
Under the high-speed impact of a SC jet, the pressure on the interface between
the jet tip and concrete target may reach an order of GPa. The influence of the
concrete target strength on the penetration can be neglected, which is characterized
as a fluid–fluid interaction and governed by the steady-state fluid dynamics (e.g.,
Bernoulli’s equation). As the jet velocity reduces continuously with the penetra-
tions, the role of concrete target strength will play a more and more important role
and the semi-hydrodynamics (e.g., Alekseevskii–Tate model) gradually dominates
the penetration process; thus, the slight differences of DOPs for the various targets
occur. However, comparing with the armored target, the related theoretical

Fig. 5.33 Borehole profiles -20 -15 -10 -5 0 5 10 15 20

10 10
of the SC jet penetration into
concrete targets with different 0 0
compressive strengths
-10 -10
DOP (cm)

-20 -20

-30 -30
20MPa
-40 40MPa -40
44MPa
-50 200MPa -50

-60 -60
-20 -15 -10 -5 0 5 10 15 20
Radius (cm)
5.3 Shaped Charge (SC) Jet Penetration 207

Fig. 5.34 Penetration -20 -15 -10 -5 0 5 10 15 20

10 10
borehole profiles with
different liner materials 0 0
-10 -10
-20 -20

DOP (cm)
-30 -30
-40 -40
-50 -50
-60 -60
1100-H aluminum
-70 Copper
-70
-80 -80
-20 -15 -10 -5 0 5 10 15 20
Radius (cm)

discussions on a SC jet penetrating into concrete targets are limited, which urgently
needs further studies in the future.

5.3.4.3 Influence of Liner Material

For the standoff distance of 2CD, Fig. 5.34 illustrates the borehole profiles for the
SC jet penetrating into concrete targets with the liner materials of 1100-H aluminum
and copper, respectively. The corresponding parameters of the liner materials are
listed in Table 5.6. It indicates that the DOP of the SC with the copper liner is larger
than that of the SC with the 1100-H aluminum liner, while the borehole diameter of
the SC with the copper liner is smaller on the contrary, which coincides with the
experimental conclusions drawn by Wang et al. (2015). The reason is that the
copper is characterized by high density and good plasticity under high temperature,
and hence, the copper jet has larger kinetic energy to increase the DOP, while the
1100-H aluminum possesses the features of low density, high deformation rate, and
good ductility, and thus, a larger contact area forms between the aluminum jet and
target during the penetration process, and more kinetic energy of the aluminum jet
is consumed to enlarge the borehole diameter.

5.3.4.4 Influence of Explosive Type

For the four types of explosives (8701, B, LX-14, and Octol) and the two different
standoff distances (1CD and 4CD), Figs. 5.35 and 5.36 illustrate the numerically
simulated SC jet penetration borehole profiles, jet velocity gradients, and mass
distributions along the jet symmetry axis at the beginning of penetrations, respec-
tively. The corresponding parameters of the explosives are given in Table 5.5.
208 5 Eroding Projectile and Shaped Charge Jet Penetrations

(a) 10-20 -15 -10 -5 0 5 10 15 20

10
(b) 8
8701
7
0 0 B
6 LX-14

Velocity (km/s)
-10 -10 Octol
5
DOP (cm)

-20 -20
4
-30 -30
8701 3
-40 B -40
LX-14 2
-50 Octol -50
1
-60 -60
-20 -15 -10 -5 0 5 10 15 20 -2 0 2 4 6 8 10 12 14
Radius (cm) l (cm)

(c) 16

14 8701 B LX-14 Octol

12 Jet tail
10
l (cm)

2
Jet tip
0
0 2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44

Fig. 5.35 Influence of explosive type: a SC jet penetration borehole profiles, b axial jet velocity
gradients, and c axial mass distributions with a standoff distance of 1CD

Figures 5.35a and 5.36a indicate that the SC jet penetration borehole profiles
with the above four types of explosives nearly overlap with each other under the
standoff distance of 1CD. The borehole profiles produced by the SC jet penetration
differ gradually for various explosives as the standoff distance increases. When the
standoff distance equals 4CD, the largest DOP is formed by the SC jet with the
explosive LX-14 and followed by the explosive Octol, while the explosives 8701
and B possess the minimum value of DOP.
The above differences of the penetration performance of the SC jet with various
explosives are mainly attributed to the axial velocity and mass distributions of the
jet. Given in Table 5.5, the detonation velocity and energy, as well as Chapman–
Jouguet pressure of explosive LX-14, are largest among the four explosives, which
is followed by the explosive Octol, while the explosives B and 8701 have the
almost identical and the smallest values for the above parameters. Thus, the velocity
and velocity gradient of explosive LX-14 are larger than those of explosives Octol,
B, and 8701, shown in Figs. 5.35b and 5.36b. Besides, when the standoff distance
of SC is small, e.g., 1CD in Fig. 5.35c, the jet has not accelerated completely, and
the majority of the jet masses are concentrated at the later portion. Correspondingly,
for the larger standoff distance of SC, e.g., 4CD in Fig. 5.36c, the jet forms almost
5.3 Shaped Charge (SC) Jet Penetration 209

(a) 10 -20 -15 -10 -5 0 5 10 15 20

10
(b) 8
0 0 7 8701
-10 -10 B
6 LX-14

Velocity (km/s)
-20 -20 Octol
DOP (cm)

-30 -30 5
-40 -40
4
-50 8701 -50
B
-60 -60 3
LX-14
-70 Octol -70
2
-80 -80
-90 -90 1
-20 -15 -10 -5 0 5 10 15 20 -5 0 5 10 15 20 25 30 35 40 45
Radius (cm) l (cm)

(c) 26
24 Jet tail Jet tip

22
20 Octol
18
16
14 LX-14
12
10
B
8
6
4 8701
2
0
0 5 10 15 20 25 30 35 40 45
l (cm)

Fig. 5.36 Influence of explosive type: a SC jet penetration borehole profiles, b axial jet velocity
gradients, and c axial mass distributions with a standoff distance of 4CD

completely and the masses of the jet distribute more uniformly. Under this cir-
cumstance, the SC jet can be treated as the cylindrical long-rod projectile.

5.4 Summary

In this chapter, the eroding projectile penetration and the SC jet penetration into
concrete targets are studied, and the main conclusions are given as follows:
(1) In Sect. 5.2, the eroding projectile penetration into the concrete targets is dis-
cussed. Firstly, the existing eroding projectile penetration models for the
metallic targets are evaluated, and it is found that the existing models for the
metallic targets cannot be applied for predicting DOP of the concrete targets
directly. Then, in order to provide sufficient experimental data for the estab-
lishment of the eroding projectile penetration model for the concrete targets,
eighteen shots of the flat-nosed cylindrical steel projectile penetration test into
210 5 Eroding Projectile and Shaped Charge Jet Penetrations

the mortar targets are conducted, in which the striking velocities range from
510 to 1850 m/s. The experimental data including the DOP, deformation and
damage of the projectile, frontal cratering, and the tunneling profile of the target
are carefully recorded. Three penetration regimes, i.e., rigid projectile pene-
tration, deforming projectile penetration without eroding, and eroding projectile
penetration, are firstly observed for projectile penetrating into concrete targets.
Finally, a simple eroding projectile penetration model is established based on
the A-T model and the unified one-dimensional target resistance. The model is
validated by the existing and our test data including the DOP and the residual
projectile length. Furthermore, the upper limit velocity of the rigid penetration
and the lower limit velocity of the eroding penetration are determined, and
rather, good agreements between the model predictions and test data are
obtained.
(2) In Sect. 5.3, regarding the widely applications of the SC jet in the military and
energy extraction fields, as well as the limited investigations on the perfor-
mance of the SC jet penetrations into the concrete targets, the numerical sim-
ulations of the SC jet penetrating into concrete targets are conducted. The
whole process including the formation in the SC, elongation in the air, and
penetration into concrete targets is reproduced. The constitutive models, as well
as the corresponding parameters and finite element algorithm, are validated by
comparing the experimental data and numerical results of the jet striking
velocity as well as the target damages. Within the present discussion ranges, the
parametric analyses indicate that (i) for the unfragmented SC jet, the DOP and
average diameters of penetration boreholes increase and decrease with the
rising of the standoff distance, respectively; (ii) the compressive strength of the
concrete target has slightly influences on the DOP of SC jet; (iii) the jet pen-
etration borehole of the SC with the aluminum liner is larger in diameter, while
the deeper DOP is obtained by the SC with copper liner; and (iv) the influence
of the explosive types on the penetration performance is not obvious with small
standoff distance, while the explosive LX-14 possesses a better penetration
performance than the explosives Octol, B, and 8701 in the case of larger
standoff distances.
Chapter 6
Efficient Decoupled Analytical/Numerical
Approach of Terminal Ballistic Trajectory

6.1 Introduction

The terminal ballistic trajectory of an earth penetrator penetrating into concrete

targets is another issue concerned by both weapon designers and protective engi-
neers. In this chapter, the terminal ballistic trajectory of a projectile is studied by
combining analytical and numerical approaches.
The numerical approaches for predicting projectile trajectory can be classified as
coupled or decoupled one. The coupled approach usually uses the so-called
hydrocode that includes contact and erosion algorithms that are allowed to simulate
the interaction between the target and penetrator. It can be further classified into
Lagrangian, Eulerian, Arbitrary Lagrangian–Eulerian approaches, etc. based on
different representations of the conservation laws for a continuum. Since the cou-
pled numerical approach requires careful modelling of both the target and projec-
tile, the arising problems generally involve a very huge amount of elements and can
be very expensive and time-consuming. In addition, Lagrangian approach often
suffers from mesh entanglement in the target that may terminate numerical com-
putation. In the decoupled approach, the projectile is usually modeled by an explicit
transient dynamic finite element code and the target is represented by the forcing
function as the pressure boundary condition, which is proved to be efficient and
robust for the predictions of structural response and trajectory of the projectile
(Warren and Poormon 2001; Fang et al. 2014a; Kong et al. 2014; Li and
Flores-Johnson 2011). This methodology eliminates the needs for discretizing the
target as well as the contact and erosion algorithms, which reduces the computer
time and memory requirement tremendously.
The accuracy of the decoupled approach depends on how well the forcing
function approximates the actual situation. The forcing function is usually deter-
mined by the dynamic cavity expansion model or semi-empirical model, which is
based on the infinite homogeneous target assumption. The following three issues
should be carefully addressed in the decoupled approach.

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 211
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_6
212 6 Efficient Decoupled Analytical/Numerical Approach …

(i) Due to the free-surface effect of targets, the forcing function cannot be
directly applied on the projectile surface when penetrating into a finite depth
target. Macek and Duffey (2000) developed a finite spherical cavity expan-
sion model to account for the free-surface effect, where they assume the
cavity uniformly expand a layered sphere with a finite radius in a layered
geologic target which is considered to be incompressible and damaged
Mohr–Coulomb material. However, the assumption of fully incompressible
assumption will overestimate the forcing function (Warren and Forrestal
1998; Forrestal and Tzou 1997). Longcope et al. (1999) used the free surface
influence distance (from a surface node of the projectile to free surface along
the normal line to projectile surface), which is obtain empirically or from the
analysis of a spherical cavity expanding in a finite target, to account for the
free-surface effect. The forcing function on the projectile surface is taken as
zero if the distance is shorter than a critical value. This method, however,
neglects the fact that the free-surface effect becomes more obviously as the
cavity expansion velocity increases (Warren and Poormon 2001). Warren
and Poormon (2001) and Warren et al. (2004b) further constructed a decay
function and then multiplied the empirical forcing function by the decay
function to account for the free-surface effect. The decay function is based on
the solution of a dynamically expanding spherical cavity in a finite target that
is assumed incompressible and perfectly plastic material. Good agreements
are obtained for projectiles penetrating into the metallic targets with oblique
angles, while the simulation results are not quite well for projectiles pene-
trating into the geological targets with oblique angles, especially when the
initial velocities are relatively low and the oblique angles are relatively large
(Warren et al. 2004b).
(ii) Similarly, the forcing function will be affected by other layers during pro-
jectiles penetrating into layered targets, which may be called the layering
effect. Macek and Duffey (2000) developed a finite spherical cavity expan-
sion model to account for the layering effect. They assume that the cavity
uniformly expands a layered sphere with a finite radius to obtain the forcing
function with the layering effect. The targets are considered incompressible
and damaged Mohr–Coulomb materials. They also point out that it is
essential to assume the incompressibility, since the time-dependent wave
propagation and reflection need to be tracked along with the moving pro-
jectile. The absence of this assumption would greatly complicate the solution
procedure. However, the fully incompressible forcing function will overes-
timate the target resistance, as stated before.
(iii) Another important issue is the “J-shape” curve and reverse trajectory of the
projectile during deep penetration. For this case, in order to change the
direction of the trajectory, the diameter of the cavity must be larger than that
of the projectile. It means that somewhere on the projectile may detach from
the target. Abrasion patterns on recovered projectiles also indicate that the
reattachment of the target material to the shank of the projectile may occur
(Byers et al. 1978). This phenomenon is termed as the wake separation and
6.1 Introduction 213

reattachment. Consequently, the forcing function will be affected by this

effect, e.g., the forcing function should be set as zero for the detached nodes.
In this chapter, in order to account for the influence of the free-surface effect,
layering effect, and wake separation and reattachment effect on the forcing function,
the decay function for the free-surface effect, the layering function for the layering
effect and the contact function for the wake separation and reattachment effect will
be firstly constructed, respectively. Then, the forcing functions can be obtained by
multiplying the semi-empirical forcing function for the homogenous infinite depth
target by the above-mentioned decay, layering, and the contact functions.
Furthermore, the implementation of the proposed decoupled approach into
ABAQUS and the corresponding predicting terminal projectile trajectory software
for rigid projectile are presented. Finally, some typical scenarios of terminal pro-
jectile trajectories are analyzed and discussed in detail.

6.2 Forcing Function

The semi-empirical forcing function for the infinite homogeneous target is given by

rr ¼ R þ Bq0 Vc2 ð6:1Þ

where rr is the semi-empirical forcing function for the infinite homogeneous target,
q0 is the density of the undeformed target material. B is an empirical constant,
depending mostly on the compressibility of the target material. Vc is the cavity
expansion velocity. R is an empirical constant, which is obtained by curve fitting of
the normal penetration experiments under the condition that all parameters are held
constant except the striking velocity. The above equation implies that the consti-
tutive behaviour of the target and any frictional resistance are all lumped into the
empirical constant R (Warren et al. 2004b).
However, the post-test observations for both concrete and rock targets (Forrestal
et al. 1994; Frew et al. 2000) indicate that there are two regions associated with the
penetration of brittle material targets. The first region is a conical crater that is
generally about two projectile diameters in depth. The second region is the so-called
tunneling region, which starts from the end of the cratering region to the final depth
of penetration. In the tunneling region, Forrestal et al. (1994) gave B = 1 and
R = Sfc for concrete targets, where S is empirical parameter expressed by S ¼
82:6fc0:544 (fc unit in MPa) (Frew et al. 1998), which is employed in this chapter.
For limestone targets, however, Frew et al. (2000) observed a noticeable decreasing
in R as projectile diameter increases and proposed R ¼ a þ bðd0 =d Þ (a ¼ 607 MPa,
b ¼ 86 MPa, and d0 ¼ 25:4 mm). In the cratering region, Forrestal et al. (1994) and
Frew et al. (2000) allowed the target resistance to increase linearly. For the con-
venience of computation, the forcing function acting at the projectile surface in the
cratering region rrc is expressed approximately as
214 6 Efficient Decoupled Analytical/Numerical Approach …

hpen
rrc ¼ rr ; hpen  2d ð6:2Þ
2d

where rr is obtained by Eq. (6.1), d is the projectile diameter, and hpen is the current
penetration depth.
The empirical constants B and R for various targets are summarized in Table 6.1.
Note that B and R for soil targets are determined by the spherical cavity expansion
model. E is Young’s modulus, k and s are the pressure-dependent shear strength
parameters, respectively.

6.2.1 Free-Surface Effect

Equation (6.1) is acquired based on the assumption of the infinite depth homoge-
neous target. As discussed in Sect. 6.1, due to the free-surface effect, there must be
some correction for Eq. (6.1) when penetrating into a finite thickness slab. Treating
the target as the incompressible Mohr–Coulomb material and assuming that the
cavity expansion produces plastic-cracked-elastic response region, the decay
function for the free-surface effect will be constructed for geological targets in this
section.
Considering the general circumstance, i.e., a projectile non-normally impacts a
finite thickness target, as shown in Fig. 6.1.
The equations of momentum and mass conservation in Eulerian coordinates with
spherical symmetry (as shown in Fig. 6.2, where rb, rc, and rd are the locations of
plastic-cracked interface, cracked-elastic interface, and free-surface, respectively)
are


@rr 2 ð rr  r h Þ @v @v
þ ¼ q þv ð6:3aÞ
@r r @t @r

@
q0 ½ðr  uÞ3  ¼ 3qr 2 ð6:3bÞ
@r

Table 6.1 Parameters for various targets

Material B R (MPa) k s (MPa) E (GPa) q0 (kg/m3)
Soil (Forrestal and Luk 1992a) 1.1 20.3 0.75 10 0.16 1860
Limestone (Frew et al. 2000) 1 a+b (d0/d) 1.023 40 25 2300
Concrete (Frew et al. 1998) 1 82.6 fc0:456 – – – –
6.2 Forcing Function 215

Fig. 6.1 Schematic diagram

of projectile non-normally
impacting a finite thickness
slab

rd
free surface
B Y

A rd
X
free surface Z

Fig. 6.2 Response regions

for the cavity expansion
problem

Elastic
Cracked
Plastic

ra rb rc rd r

where rr and rh are the radial and hoop components of the stresses, q0 and q are the
densities of target materials in the undeformed and deformed states, respectively.
q ¼ q0 when the material is assumed to be incompressible. Particle displacement
u and particle velocity v of target material in the radial direction are related by


@u @u
¼v 1 ð6:4Þ
@t @r

Integrating Eq. (6.3b) gives the particle displacement as

"
1=3 #
r3
u ¼ r 1  1  a3 ð6:5Þ
r

Substituting Eq. (6.5) into Eq. (6.4) gives the particle velocity as

ra2 r_ a
v¼ ð6:6Þ
r2

where ra is the cavity radius and the dot represents the derivation of time. In the
elastic region, the small displacement is assumed and the particle displacement in
the radial direction can be written as (Hopkins 1960).
216 6 Efficient Decoupled Analytical/Numerical Approach …

ra3
u¼ ð6:7Þ
3r 3

The elastic strains are

@u 2ra3 u r3
er ¼  ¼ ; eh ¼ ¼ a2 ð6:8Þ
@r 3r 2 r 3r

where er and eh are the radial and hoop strain components, respectively. Hook’s law
for an incompressible material provides the relationship as

2Eðer  eh Þ 2Era3
rr  rh ¼ ¼ ð6:9Þ
3 3r 3

It should be pointed out that Eq. (6.9) is applicable for any incompressible
targets in the elastic region.
Now, we consider a finite thickness slab, as shown in Fig. 6.1, any point A on
the lower projectile surface (the Y-direction component of normal vector to the
projectile surface is negative) is assumed to be uniformly expanding a sphere with
finite radius. The boundary condition on the exterior free surface is zero, while the
particle radial velocity at the projectile surface is r_ a . We firstly assume that the
cavity expansion produces plastic-cracked-elastic response region, the response
solutions for the cavity expansion problem are then derived.

6.2.1.1 Solution for Elastic Region

Putting Eq. (6.9) into Eq. (6.3a), resulting in

@rr 4Era 1  2ra4 r_ a2

þ ¼ q 2r a _
r 2
þ r 2
€
r a þ q ð6:10Þ
@r 3r 4 r2 a a
r5

Neglecting €ra (considering the cavity expanding at a constant velocity) and

integrating Eq. (6.10) throughout the elastic region gives


4Era3 1 21 ra4 r_ a2 1
rr ¼ þ q 2ra r_ a  þ K1 ð6:11Þ
9 r3 r 2 r4

where K1 refers to an integration constant, which can be determined by rr ðrd Þ ¼ 0.

6.2.1.2 Solution for Cracked Region

At the elastic-cracked interface rc, the material in the elastic region reaches its
circumferential tensile strength ft , thus
6.2 Forcing Function 217

2Era3
rr  rh ¼ rr þ ft ¼ ð6:12Þ
3rc3

Hugoniot interface conditions require that the radial stress be continuous for the
incompressible material, therefore


4Era3 1 2 1 ra4 r_ a2 1 2Era3
þ q 2ra _
r a  þ K 1 þ f t  ¼0 ð6:13Þ
9 rc3 rc 2 rc4 3rc3

where rc can be obtained by solving Eq. (6.13) with Newton iteration method in the
range of ra < rc < rd.
In the cracked region, rh ¼ 0; thus, the momentum equation for the cracked
region is

@rr 2rr 1  2r 4 r_ 2
þ ¼ q 2 2ra r_ a2 þ ra2€ra þ q a5 a ð6:14Þ
@r r r r

The radial stress distribution for the cracked region can be found by integrating
Eq. (6.14) (neglecting €ra ) as

ra4 r_ a2 2ra r_ a2 K2
rr ¼ q  q þ 2 ð6:15Þ
r4 r r

where K2 refers to an integration constant, which can be determined by letting

Eq. (6.11) equal to Eq. (6.15) with r = rc.

6.2.1.3 Solution for Plastic Region

The shear behaviour of the geological material is described by the Mohr–Coulomb

yield criterion, thus

P ¼ ðrr þ rh þ ru Þ=3; rh ¼ ru ; rr  rh ¼ kP þ s ð6:16Þ

where P is the hydrostatic pressure.

At the plastic-cracked interface rb, we obtain

ra4 r_ a2 2ra r_ 2 K2
q 4
 q 2 a þ 2 ¼ rr ¼ Y ð6:17Þ
rb rb rb

where Y is the unconfined compressive strength of the target. rb can be obtained by

solving Eq. (6.17) with Newton iteration method in the range of ra < rb < rd.
In the plastic region, taking Eq. (6.16) into Eq. (6.3a), resulting in
218 6 Efficient Decoupled Analytical/Numerical Approach …

@rr akrr 1  2r 4 r_ 2 as
þ ¼ q 2 2ra r_ a2 þ ra2€ra þ q a5 a  ð6:18Þ
@r r r r r

where a ¼ 6=ð3 þ 2kÞ. Neglecting €ra and integrating Eq. (6.18) yields the radial
stress distribution in the plastic region, i.e.

2qra4 r_ a2 2qra r_ a2 s
rr ¼   þ K3 r ak ð6:19Þ
ðak  4Þr 4 ðak  1Þr k

where K3 refers to an integration constant, which can be determined by letting

Eq. (6.15) equal to Eq. (6.19) with r = rb.
With the cavity expansion velocity increasing, the cracked region is eliminated
eventually (rb > rc), and the response is plastic-elastic for the high cavity expansion
velocity. Thus, the plastic-elastic interface is determined by Warren et al. (2004b)

1=3
ra 3s
¼ ð6:20Þ
rb 2E

Under this circumstance, the radial stress solution for the elastic region is the
same as Eq. (6.11). The radial stress solution for the plastic region is also the same
as Eq. (6.19), while the integration constant K3 should be determined by letting
Eq. (6.11) equal to Eq. (6.19) with r ¼ rb .

6.2.1.4 Radial Stress at Cavity Surface

According to the above discussions, the radial stress at the cavity surface can be
obtained by the following steps.
(i) Starting from the free surface where the radial stress is zero, the locations of
the elastic-cracked and cracked-plastic interfaces (rc and rb) are determined
based on Eqs. (6.13) and (6.17), respectively.
(ii) If rc  rb, the response is plastic-cracked-elastic, rr ðra Þ can be obtained by
Eqs. (6.11), (6.15), and (6.19). However, if rc [ rd , tensile failure occurs at
the free surface, so rr ðra Þ ¼ 0.
(iii) If rc < rb, the response is plastic-elastic, rr ðra Þ can be determined by
Eqs. (6.20), (6.11), and (6.19). However, if rb [ rd , compression failure
occurs at the free surface, thus rr ðra Þ ¼ 0.
The radial stress at the cavity surface rr ðra Þ is obtained under the incompressible
condition. If it is directly applied on the projectile surface, the forcing function will
be overestimated, thus a decay function to account for the free-surface effect should
be constructed.
6.2 Forcing Function 219

6.2.1.5 Decay Function for Free-Surface Effect

If the cavity expands in an infinite depth geological target, it will produce the
plastic-cracked-elastic response regions, the radial stress at the cavity surface r0r ðra Þ
can be determined by Eqs. (6.11), (6.15), and (6.19) with rd ! 1.
The decay function is constructed in the way as follows

rr ðra Þ
f ðra ; rd ; r_ a Þ ¼ ð6:21Þ
r0r ðra Þ

where r0r ðra Þ is the radial stress at the cavity surface for the infinite depth target.
Only the frontal free-surface effect should be addressed if the projectile
non-normally penetrates into a semi-infinite target. However, it must be noted that
both the free-surface effect at the front and rear surfaces should be accounted if the
projectile non-normally impacts on a finite thickness slab, as shown in Fig. 6.1.
Plots of the decay function as a function of rd/ra obtained from Eq. (6.21) and
decay function from Warren et al. (2004b) for several expansion velocities are
shown in Fig. 6.3, respectively. The parameters for the limestone target are listed in
Table 6.1. It is observed that, if the expansion velocities are relatively small, the
decay function obtained by the plastic-cracked-elastic response regions (Eq. 6.21)
is lower than that by the plastic-elastic response regions (Warren et al. 2004b).
While the decay functions are identical for high expansion velocities since the
cracked region is eliminated eventually. Additionally, the free-surface effect
extends to larger distances as the cavity expansion velocity increases gradually.

Fig. 6.3 Decay function 1.2

versus the non-dimensional
distance to the free surface 1.0

0.8
Decay function

0.6

100m/s plastic-cracked-elastic
0.4 200m/s plastic-cracked-elastic
400m/s plastic-cracked-elastic
100m/s plastic-elastic Warren et al. (2004b)
0.2 200m/s plastic-elastic Warren et al. (2004b)
400m/s plastic-elastic Warren et al. (2004b)
0.0
0 20 40 60 80 100
rd /ra
220 6 Efficient Decoupled Analytical/Numerical Approach …

6.2.2 Layering Effect

As discussed in Sect. 6.1, due to the layering effect, some corrections should also be
made to Eq. (6.1) when applying on the projectile surface in case of penetration into
the layered targets. In this section, we propose a method to account for the layering
effect, in which a layering effect function is constructed to obtain the semi-empirical
forcing function for the layered targets. This function is derived from a dynamically
expanding layered sphere cavity with finite radius in the layered incompressible
Mohr–Coulomb targets.
Now, we consider the general circumstance, i.e., the projectile impacts the
n layers targets with oblique angle, as shown in Fig. 6.4a. Based on the finite cavity
expansion theory, any point A on the surface of the projectile is assumed to be
expanding a layered sphere with finite radius r = r1 = ra, r = ri, r = ri+1, …,
r = rn+1, respectively, as shown in Fig. 6.4b. The boundary condition on the
exterior free surface is zero, while the particle radial velocity at the projectile
surface is Vc.
The expansion of the cavity usually produces plastic-cracked-elastic response

regions in each layer, as shown in Fig. 6.5a, where rbi , rci , and riþ 1 are the loca-
tions of plastic-cracked interface, cracked-elastic interface, and the interface of ith
and (i + 1)th layers, respectively. As the cavity expansion velocity increases, the
cracked region diminishes eventually (when rci \rbi 
). Thus, the response is
plastic-elastic for high cavity expansion velocity, as shown in Fig. 6.5b. Next, we
derive the radial stress solution of the ith layer for the two modes, respectively.

6.2.2.1 Governing Equations

The equations governing the solution of the radial stress for the ith layer consist of
the incompressible condition of the target, the radial dynamic equilibrium, the yield
criterion of the target in the plastic region, and the continuum of the radial stress.

(a) (b)

2
ra ra ri
A
ri i
Y
ri+1 Z
i+1

rn+1
n+1 X

Fig. 6.4 Schematic diagram of projectile impacting the n layers targets

6.2 Forcing Function 221

(a) (b)

Elastic
Cracked Elastic
Plastic Plastic

ri * rbi* r ci* ri* 1 r* ri * rbi* ri*1 r*

Fig. 6.5 Two response modes in the ith layer a plastic-cracked-elastic b plastic-elastic

The incompressible conditions of the ith layer are presented by Macek and Duffey
(2000) and summarized as follow.

r 3  ri3 ¼ r 3  ri3 ð6:22Þ

 1=3
u ¼ r   r ¼ ri3 þ r 3  ri3 r ð6:23Þ

ri2
vi v¼ ð6:24Þ
r 2



r 1 r 3
ar ¼ i ai þ 2ri v2i 2  i5 ð6:25Þ
r r r

u r
eh ¼  ¼ 1  1=3
ð6:26Þ
r ðr 3  r 3 þ r 3 Þ i i

 3 2=3
@u r  ri3 þ ri3
er ¼  ¼1 ð6:27Þ
@r r

where the radial Eulerian coordinate r with and without the superscript * denote the
deformed and the undeformed configurations, respectively. The subscript i refers to
quantities measured at the inner surface of the ith layer, as shown in Figs. 6.4 and
6.5. u, v, ar, eh , er are the radial particle displacement, radial particle velocity, radial
particle acceleration, hoop strain, and radial strain, respectively. According to
Eq. (6.25), the particle radial acceleration should not be neglected, even though in
the case of the cavity expanding at a constant velocity.
The equation of momentum conservation in Eulerian coordinates with spherical
symmetry under the deformed configuration is

drr 2
þ  ðrr  rh Þ ¼ qi ar ð6:28Þ
dr  r
222 6 Efficient Decoupled Analytical/Numerical Approach …

6.2.2.2 Plastic-Cracked-Elastic Response

The solutions of the stress in each region are given in detail as follows:
(1) Stress in elastic region
The Hook’s law for an incompressible material provides the relationship as

2Ei 2Ei ri3  ri3

rr  r h ¼ ðer  eh Þ  ð6:29Þ
3 3 r 3

where Ei denotes the Young’s modulus of the ith layer.

Putting Eq. (6.29) into Eq. (6.28) results in

 2

drr ri  2 1 ri3 4Ei  

¼ qi 
a i  2q r v
i i i 2
 5
 4 ri3  ri3 ð6:30Þ
dr r r r 3r

Integrating Eq. (6.30) throughout the elastic region yields the radial stress dis-
tribution for rci \r  \riþ 1 ,



1 1 1 1 r 3 ri3
rr ¼ rri þ 1 þ qi ri2 ai     2qi ri v2i    þ i 4  4
r r ri þ 1 r 4r 4ri þ 1


i þ 1 3 
4Ei ri3 ri3 ri
  1  3 ð6:31Þ
9 ri3þ 1 r 3 ri

(2) Stress in cracked region

At the elastic-cracked interface rci , the material in the elastic region reach its
circumferential tensile strength fti , thus


1 1
rr  rh ¼ rr þ fti ¼ rri þ 1 þ qi ri2 ai   
rci ri þ 1


1 1 r 3 ri3
 2qi ri v2i
  
þ i 4  4
ri þ 1 rci 4rci 4ri þ 1

3 3

3
 ð6:32Þ
4Ei ri r ri
  i 1  3 þ fti
9 ri3þ 1 rci3 ri
2Ei ri3  ri3
¼
3 rci3

Letting r c ¼ ri =rci and r i ¼ ri =riþ 1 , and simplifying Eq. (6.32) yields the fol-
lowing equation
6.2 Forcing Function 223

C1 r 4c þ C2 r 3c þ C3 r c þ C4 ¼ 0 ð6:33aÞ


q v2 2Ei ri3
C1 ¼  i i ; C2 ¼  1  3 ; C3 ¼ 2qi v2i þ qi ri ai ;
2 9 ri

 ð6:33bÞ
 qi v2i 4 4 3 ri3
C4 ¼ rri þ 1  qi ri ai r i  2qi vi r i þ
2
r  Ei r 1  3 þ fti
2 i 9 i ri

where r c can be obtained by solving Eq. (6.33a, b) with Newton iteration method in
the range of 0\r c \1.
In the cracked region, rh ¼ 0; thus, the momentum equation for the cracked
region is

 2

drr 2rr ri  2 1 ri3
þ ¼ q a i  2q r v  ð6:34Þ
dr  r i
r i i i
r 2 r 5

The radial stress distribution for the cracked region can be found by integrating
Eq. (6.34), i.e.

ri4 v2i 2r  v2 r 2 ai K1i

rr ¼ qi 4
 qi i i  qi i  þ 2 ð6:35Þ
r r r r

where K1i refers to an integration constant, which can be determined by letting

Eq. (6.31) equal to Eq. (6.35) with r  ¼ rci .
(3) Stress in plastic region
The ith layer material is described by an incompressible Mohr–Coulomb yield
criterion, thus
 
P ¼ rr þ rh þ ru =3; rh ¼ ru ; rr  rh ¼ ki P þ si ð6:36Þ

where ki and si refer to the pressure-dependent shear strength parameters of the ith
layer, respectively.

At the plastic-cracked interface rbi , we obtain

ki rr
rr  rh ¼ rr ¼ þ si ¼ Yi ð6:37Þ
3

where Yi is the unconfined compressive strength of the ith layer. Taking Eq. (6.36)
into Eq. (6.28), resulting in

ri4 v2i 2r  v2 r 2 ai K1i

qi 4
 qi i i  qi i  þ 2 ¼ Yi ð6:38Þ
rbi rbi rbi rbi

Letting r b ¼ ri =rbi


and simplifying Eq. (6.38), the equation for r b is found as
224 6 Efficient Decoupled Analytical/Numerical Approach …

K1i 2  
qi v2i r 4b þ r  2qi Vi2 þ qi ri ai  Yi ¼ 0 ð6:39Þ
ri2 b

where r b can be obtained by solving Eq. (6.39) with Newton iteration method in the
range of 0\r b \1.
Rearranging Eq. (6.36) gives

ð3  ki Þrr 3si
rh ¼  ð6:40Þ
3 þ 2ki 3 þ 2ki

In the plastic region, substituting Eq. (6.40) into Eq. (6.28), resulting in

 2

drr ai ki rr ri  2 1 ri3 ai s i
þ ¼ q a i  2q r v    ð6:41Þ
dr  r i
r i i i
r 2 r 5 r

where ai ¼ 6=ð3 þ 2ki Þ. Integrating Eq. (6.41) yields the radial stress distribution
for the plastic region as




qi ri2 ai 1 2qi ri4 v2i 1 4 2qi ri v2i 1 si
rr ¼  þ   þ K2i ðr  Þai ki
ai ki  1 r  ai ki  4 r  ai k i  1 r  ki
ð6:42Þ

where K2i refers to an integration constant, which can be determined by letting

Eq. (6.35) equal to Eq. (6.42) with r  ¼ rbi

.

6.2.2.3 Plastic-Elastic Response Region

As the cavity expansion velocity increases, the cracked region is eliminated

eventually (rci \rbi

). Therefore, the response of the ith layer is plastic-elastic for
high cavity expansion velocity, as shown in Fig. 6.5b. The solutions of the stress in
each region are given as follows.
(1) Stress in elastic region
The radial stress solution for the elastic region is the same as Eq. (6.31).
(2) Stress in plastic region

At the plastic-elastic interface rbi , ðrr  rh Þ in the elastic region reaches the
current yield stress ðki P þ si Þ, thus
6.2 Forcing Function 225


2Ei ri3  ri3 2ðrr  rh Þ
3
¼ ki p þ s i ¼ k i rr  þ si
3 rbi 3
2

3
2 1 1  2 1 1 ri3 ri3
6 rri þ 1 þ qi ri ai r   r 
  2qi ri vi    þ 4  4
ri þ 1 rbi 4rbi 4ri þ 1 7
6 iþ1 7
¼ ki 6 7 þ si
bi
6
3 3

 3 7
4 4Ei ri ri ri 3
4Ei ri  ri 3
5
 3
 3
1  3
 3
9 ri þ 1 rbi ri 9 rbi
ð6:43Þ

Letting r b ¼ ri =rbi


and simplifying Eq. (6.43) yields the following equation for
rb ,

C5 r 4b þ C6 r 3b þ C7 r b þ C8 ¼ 0 ð6:44aÞ


ki qi v2i 2Ei ri3
C5 ¼  ; C6 ¼  1  3 ; C7 ¼ 2ki qi v2i þ ki qi ri ai ;
2 3 ri


ki qi v2i 4 4 ri3
C8 ¼ si þ ki rri þ 1  ki qi ri ai r i  2ki qi v2i r i þ r i  ki Ei r 3i 1  3
2 9 ri
ð6:44bÞ

where r b can be obtained by solving Eq. (6.44a, b) with Newton iteration method in
the range of 0\r b \1.
The radial stress distribution in the plastic region under this case is the same as
Eq. (6.42), while the integration constant Ki is determined by letting Eq. (6.31)
equal to Eq. (6.42) with r  ¼ rbi

.

6.2.2.4 Radial Stress at the Cavity Surface

According to the above discussions, the radial stress at the cavity surface can be
obtained by the following steps.
(i) Starting from the projectile surface where the radial deformation and the
radial velocity are known, the radial deformation, radial velocity, radial
acceleration, and the strain in each response region can be obtained by using
Eqs. (6.22–6.27) sequentially.
(ii) Starting from the free surface where the radial stress is zero, the location of

the elastic-cracked and cracked-plastic interfaces of the outmost layer (rcn

and rbn ) are then determined based on Eqs. (6.33a, b), and (6.38).
 
(iii) If rcn [ rbn , the response is plastic-cracked-elastic, rrn is obtained by
applying Eqs. (6.31), (6.35), and (6.42).
 
(iv) If rcn  rbn , the response is plastic-elastic, rrn is determined by Eqs. (6.31),
(6.42), and (6.44a, b).
226 6 Efficient Decoupled Analytical/Numerical Approach …

(v) Applying above steps (ii–iv) sequentially, the radial stress at the cavity
surface rr ðra Þ is finally obtained.
Since rr ðra Þ is obtained under the incompressible condition, it will overestimate
the target resistance if applied on the projectile surface directly. Similar to Warren
et al. (2004b), a layering effect function is constructed to account for the layering
effect.

6.2.2.5 Decay Function for the Layering Effect

If the ith layer is infinite, the cavity expansion will produce elastic, cracked, and
plastic regions. The radial stress at the cavity surface r0r ðra Þ under this condition
can be obtained by Eqs. (6.31), (6.33a, b), (6.35), (6.38), and (6.42) with ri ¼ ra ,
vi ¼ Vc , riþ 1 ! 1, and ai ¼ 0 (considering the cavity expanding at a constant
velocity).
We now construct a function to account for the layering effect, which is done by
normalizing rr ðra Þ with r0r ðra Þ

rr ðra Þ
hi ðra ; ri ; ri þ 1 ; . . .; rn þ 1 ; Vc Þ ¼ ð6:45Þ
r0r ðra Þ

Considering the layered targets made up of limestone and semi-infinite depth

soil, the cavity expands in the first limestone layer with finite thickness. Plots of the
non-dimensional radial stress at the cavity surface (rr ðra Þ=s1 ) as a function of η
(the ratio of limestone thickness r2 to the cavity radius ra) for several expansion
velocities are shown in Fig. 6.6. The empirical parameters for the limestone and soil
are listed in Table 6.1. It indicates that, (i) when r2 ! 0, the radial stress at the
cavity surface is just the stress in the case of the cavity expanding in an infinite soil
target; (ii) the influence of the soil target on the radial stress at the cavity surface

Fig. 6.6 Effect of limestone 150

layer thickness on the radial
stress at the cavity surface
120 Quasi-Static
200m/s
400m/s
90 600m/s

60

30

0
0 20 40 60 80 100 120
6.2 Forcing Function 227

decreases with the increase in the thickness of the limestone layer, and there is
almost no influence when η  50; (iii) the layering effect extends to larger distance
as the increase of the cavity expansion velocity.

6.2.3 Separation and Reattachment Effect

In this section, we will firstly establish a model to study the wake separation and
reattachment, and then obtain the contact function.
Separation and reattachment criteria will be obtained based on the following four
assumptions: (i) the oblique angle plane overlaps the attack angle plane; (ii) the
target is incompressible; (iii) ~ ~ (angular velocity
V (velocity vector of the centroid), x
vector), b (attack angle) are constant under a small time step (e.g., 10−6 s); and
(iv) the projectile radius at the separation point is approximately the radius at the
base of the nose, i.e., the projectile radius d/2.
Without the assumption (i), the location of the cavity centreline will be an
irregular curve that makes it difficult to be described and determined. In addition,
without the assumption (ii), the time-dependent wave propagation and reflection
need to be tracked relative to the moving projectile (Macek and Duffey 2000).
Consequently, the analytical solution is infeasible. Obviously, ~ ~, and b are
V, x
changing constantly during penetration process. However, these quantities are
almost unchanged during a small time step (e.g., 10−6 s), thus the quasi-static
approximation may be adequate. The assumption (iv) is reasonable, which will be
demonstrated in Sect. 6.4.1.3.

6.2.3.1 Separation Criterion

Similar to Bernard and Creighton (1979), we assume that a minimum wake sepa-
ration angle umin is required in order to maintain contact. The wake separation
 
angle u is defined as sin u ¼ Vc =~ V  for both normal penetration and oblique
penetration, as shown in Fig. 6.7, where Vc is the component of the velocity normal
to the projectile surface.
Bernard and Creighton (1979) do not consider the influence of the rotational
velocity in the definition of the wake separation angle, since they neglect the
rotational velocity in formulating Vc . Although the wake separation angle defined in
this section is the same as Bernard and Creighton (1979), the influence of the
rotational velocity is taken into account because Vc is based on the 3D rigid pro-
jectile dynamics presented in Sect. 6.3.2. Actually, it is irrational to neglect the
rotational velocity, which will be discussed further in Sect. 6.4.1.3.
228 6 Efficient Decoupled Analytical/Numerical Approach …

(a)

(b)
ξ

ϕ min

r0
Vc
Vc
V
ϕ min V

Va

Fig. 6.7 Criterion for the separation point a normal penetration b oblique penetration

6.2.3.2 Reattachment Criterion

Reattachment may or may not occur on the projectile shank after the surface nodes
of a projectile detaching from the target, depending on the relative motion of the
cavity and the projectile. Similar to Bernard and Creighton (1979), we suppose that
the cavity is a gently curved cylindrical tube with an increasing radius, and the
projectile spins in the tube as it moves forward, as shown in Fig. 6.8. Reattachment
occurs if the projectile rotates into the cavity wall. The reattachment criteria pro-
posed by Bernard and Creighton (1979) can be easily extended to a 3D dynamics
model based on the assumption (i).
The conservation of mass and incompressibility in the target require that

ra r_ a ¼ Constant ð6:46Þ

Evaluating Eq. (6.46) at the separation point, it has

 
ra r_ a ¼ r0 ~
V  sin umin ð6:47Þ

Equation (2.19) in Bernard and Creighton (1979) (i.e., ra r_ a ¼ r0 V1 tan umin , V1

is defined in Sect. 6.3.2) seems questionable because it is only valid for the normal
penetration. However, Eq. (6.47) can be applied to the oblique penetration. The
cavity radius ra is obtained as
6.2 Forcing Function 229

Fig. 6.8 Criterion for the

reattachment point Wake cavity

P2 P3
Reattachment
point
P1

Separation point

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  ffi
ra ¼ r02 þ 2r0 ðn  n0 Þ~ V  sin umin =V1 ðn  n0 Þ ð6:48Þ

where r0 is the projectile radius at the separation point, n is the axial distance
between the shank and the nose tip, and n0 is the value of n at the separation point.
Supposing lateral distance of the point P2 on the projectile axis relative to the
cavity centreline as d, as shown in Fig. 6.8, it is found that

1 x3
d ¼ ðn  n0 Þ2 þ d0 ð6:49Þ
2 V1

where the initial displacement d0 ¼ n sin b is due to the initial attack angle b, and b
 
can be determined by cos b ¼ ~ V ~e1 =~
V .
Rear axial view of the projectile cross section when d\\n is shown in Fig. 6.9.
Supposing the distance between the point P1 and the point P3 as rp, as shown in
Fig. 6.9, one may get
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
< ðd  r0 cos cÞ2 þ ðr0 sin cÞ2 p=2  c\3p=2
rp ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ð6:50Þ
: ðr cos c  dÞ2 þ ðr sin cÞ2 0  c\p=2 or 3p=2  c\2p
0 0

!
where c is determined by cos c ¼ ~ e2  P2 P1 . Reattachment occurs if rp [ ra . There
is no contact between the projectile and the target when both rp  ra and u\umin
are satisfied simultaneously.
The contact function g is determined as follows based on the above discussion
230 6 Efficient Decoupled Analytical/Numerical Approach …

(a) e2 (b) e2

Cavity surface
Cavity surface Projectile surf ace

Projectile surf ace r0sin P1

ra rp
P3 P3
ra r0
δ δ
P2 rp P2
r0
r0cos
r0sin P1

Fig. 6.9 Rear axial view of the projectile’s cross section a p/2  c  3p/2 b 0 < c < p/2 or 3p/
2 < c <2p

1 rp [ ra or u  umin
g¼ ð6:51Þ
0 rp  ra and u\umin

Generally, based on the previous discussions, the forcing function that accounts
for the free surface and layering as well as wake separation and reattachment effects
could be written as
 
rr ¼ f g h R þ Bq0 Vc2 ð6:52Þ

It should be noted that the forcing function given in Eq. (6.52) not only accounts
for the free-surface, layering as well as wake separation and reattachment effects,
but also introduce all of the constitutive behaviours of a target along with any
frictional resistance.

6.3 Efficient Decoupled Analytical/Numerical Approach

In this section, two decoupled approaches are proposed to predict the terminal
projectile trajectory. The first one is combining analytical method with finite ele-
ment (FE) method, in which the target response during penetration is represented by
the forcing function as the pressure boundary condition with the considerations of
the free-surface, layering, and wake separation and reattachment effects, and the
deformed projectile could be easily modelled by FE method. The second one is an
analytical approach for the rigid projectile, in which the undeformed projectile is
represented by a rigid body and the target response is also represented by the
forcing function as the pressure boundary condition.
6.3 Efficient Decoupled Analytical/Numerical Approach 231

6.3.1 Analytical/Numerical Approach for Deformed

Projectiles

Based on the previous sections, the response of the target can be represented by a
forcing function during the penetration process. This methodology eliminates the
need for discretizing the target as well as the need for a contact algorithm, which
greatly reduces the computer time and memory requirements. Figure 6.10 shows an
FE model of the projectile, which defines the pressure boundary condition that acts
on an element. Four nodal pressures are used to calculate for each element (Warren
et al. 2004b), and these nodal pressures acting on the projectile surface (Eq. 6.52)
can be rewritten as


pI ¼ rri ¼ f g h Ri þ Bi qi ðvI  nÞ2 ðI ¼ 1; 4Þ ð6:53Þ

where vI is the nodal velocity vector, n is the outward unit vector normal to the
element surface. The values of pI are updated during each time increment using the
current values of vI and n.
The method introduced above is implemented in ABAQUS explicit solver via
the user subroutine VDLOAD, which is used to define a distributed pressure on the
outer surface of both the projectile nose and shank. The VDLOAD is called in each
time step and calculates the pressure applied on the point I where the distributed
load is defined. The reader is referred to the ABAQUS Analysis User’s Manual
(2004) for further information of VDLOAD.

6.3.2 Analytical Approach for Rigid Projectiles

In this section, the trajectory software for a rigid projectile will be presented based
on the rigid projectile dynamics and pressure boundary condition discussed above.
If the projectile deformation can be neglected during the penetration, then it can
be treated as a rigid body. Consequently, the projectile motion is governed by the

Fig. 6.10 Definition of the p3

3
pressure boundary conditions

4
p4

2
p2

1
p1
232 6 Efficient Decoupled Analytical/Numerical Approach …

Fig. 6.11 Instantaneous

orientation and position of the d
projectile

n2
C e2
n3
P1
n1
dA
V
e3
Y

X
Z
e1

rigid body dynamics, which are presented by Li and Flores-Johnson (2011) and are
summarized as follows.
The orientation and position of a rigid projectile at an arbitrary time t are shown
in Fig. 6.11. [X, Y, Z] represents the fixed global coordinate, while ½~ e1 ;~
e2 ;~
e3 
represents the rigid body coordinate located at its centroid. C is the centroid of the
projectile. ~
V ¼ V1~ e1 þ V2~e2 þ V3~
e3 is the vector velocity of the centroid. x ~¼
x1~e1 þ x2~e2 þ x3~e3 is the angular velocity of the projectile. ~
n3 , ~
n1 , and ~
n2 instead
of ~ ~, and ~
n, m s in Li and Flores-Johnson (2011) represent the unit vector of the
normal direction and tangential direction of the projectile surface at point P1,
respectively. Vc ¼ ~V ~n3 is the normal component of the particle velocity of point
P1 on the projectile surface.
The motion of the projectile is controlled by the well-known rigid body
dynamics

d~
V ~
M ¼F ð6:54aÞ
dt

d~
L
~ ~
þx ~
I¼M ð6:54bÞ
dt

where M is the projectile mass, ~ I ¼ I1 x1~e1 þ I2 x2~

e2 þ I3 x3~
e3 is the angular
momentum about the centroid of the projectile. I1, I2 and I3 are the moment of
inertia about the axes of ½~ e1 ;~ e3 , respectively. ~
e2 ;~ ~¼
F ¼ ðFX ; FY ; FZ Þ and M
ðMX ; MY ; MZ Þ are the resultant force and moment about the centroid of the pro-
jectile, which can be determined by
6.3 Efficient Decoupled Analytical/Numerical Approach 233

ZZ
~
F ¼
~ rr dA ð6:55aÞ

ZZ
~ ¼
~
M rcp1 ~
rr dA ð6:55bÞ

where ~rcp1 is the vector from C to current point P1, ~

rr is rewritten as follows
according to Eq. (6.52)

~
rr ¼ f g h rr ðra Þ~
n3 ð6:56Þ

In order to solve Eq. (6.54a, b), the surface of a projectile is discretized by

elements to obtain the area and coordinate of each element, as discussed in Kong
et al. (2014). The differential form of Eq. (6.54a, b) is
8
>
> ~ vt1 þ ~
vt ¼ ~ Ft1 =M
 
>
< xxt ¼ xxt1 þ Mxt1  Iz  Iy xyt1 xzt1 =Ix
  ð6:57Þ
>
> xyt ¼ xyt1 þ Myt1  ðIx  Iz Þxzt1 xxt1 =Iy
>
:    
xzt ¼ xzt1 þ Mzt1  Ix  Iy xxt1 xyt1 =Iz

where the subscripts t − 1 and t represent the time step. By using Eq. (6.57) and the
generated mesh of the projectile surface, the projectile motion can be calculated
according to the given initial condition. Consequently, the software can be com-
piled. This software has advantages of high computational accuracy, less resources
occupation, wide application range, definite physical meaning, and fast computation
speed.

6.4 Validation

In this section, in order to verify the established models of free-surface, layering as

well as separation and reattachment effects, the simulated results are compared with
the available experimental results. Furthermore, some typical projectile impact
scenarios from the trajectory software are presented and discussed in Sect. 6.5.

6.4.1 Projectile Penetrations

6.4.1.1 Normal Penetration

Figure 6.12 shows the comparisons of DOP between the experimental data of the
various targets under the normal impact with pitch and yaw angles, and the
234 6 Efficient Decoupled Analytical/Numerical Approach …

Fig. 6.12 Comparisons 4.0

51MPa concrete target
between the experimental and Simulation Forrestal et al. (1996)
3.5
predicted DOP 62.8MPa concrete target
Simulation Forrestal et al. (1996)
3.0 Limestone target
Simulation Frew et al. (2000)
2.5 Limestone target

hpen (m)
Simulation Frew et al. (2000)
2.0

1.5

1.0

0.5

0.0
400 600 800 1000 1200 1400 1600
V0 (m/s)

corresponding results predicted by the analytical model. The targets parameters are
summarized in Table 6.1. A good agreement is observed. However, our approach
overestimates the DOP when the projectile velocity is higher than 1000 m/s for
concrete targets. The overestimation may be due to the projectile erosion and the
inaccuracy of the empirical parameter R.

6.4.1.2 Oblique Penetration

Warren et al. (2004b) performed the non-normal projectile penetration test into
limestone targets with 30° oblique angle and various velocities. Figure 6.13 pre-
sents the projectile deformation obtained from experiments (left), the above pro-
posed approach with considering the free-surface effect (middle) and the model
proposed by Warren et al. (2004b) (right), respectively. Figure 6.14 presents the
corresponding final position (horizontal coordinate X and vertical coordinate Y) of
the projectile tip. It can be observed that the projectile deformation and the final
position of the projectile tip obtained from the proposed approach show more close
to the experiments, especially when the initial velocity is relatively small.

6.4.1.3 Deep Penetration

Sandia National Laboratory conducted the penetration experiments (Bernard and

Creighton 1979) on soil targets with the projectile mass of 145 kg and CRH of
9.25. The initial velocity and oblique angle are 711 m/s and 30°, respectively.
These experiments are performed at TTP Antelope dry lake site, which is same as
the place of the test conducted by Forrestal and Luk (1992a). Thus, the empirical
parameters B and R in Table 6.1, which account for the uncertainty of soil prop-
erties varied with depth are reasonable. Rigid projectile penetration is assumed
since little deformation is observed after the experiments. The projectile trajectories
6.4 Validation 235

Fig. 6.13 Comparisons of the experimental projectile deformation and predicted results by
proposed approach and the model by Warren et al. (2004b) a V0 = 413 m/s b V0 = 510 m/s
c V0 = 610 m/s d V0 = 804 m/s

at various values of umin (minimum wake separation angle) are shown in Fig. 6.15.
The solid lines represent the predicted trajectories with various values of umin based
on above proposed approach in which the separation and reattachment effect is
considered. The dashed lines represent the predicted trajectories based on the model
236 6 Efficient Decoupled Analytical/Numerical Approach …

Fig. 6.14 Comparisons of (a) 240

the experimental projectile tip Experiment (Warren et al. (2004b))
position and the predicted Plastic-cracked-elastic
results by proposed approach 200 Plastic-elastic (Warren et al. (2004b))
and the model by Warren
et al. (2004) a horizontal 160

X (mm)
coordinate X b vertical
coordinate Y
120

80

40
200 400 600 800 1000 1200 1400
V0 (m/s)

(b) 350
Experiment (Warren et al. (2004b))
300 Plastic-cracked-elastic
Plastic-elastic (Warren et al. (2004b))
250
Y (mm)

200

150

100

50

0
300 600 900 1200 1500
V0 (m/s)

that rotational velocity is not considered in formulating the criterion for the wake
separation, i.e., the model proposed by Bernard and Creighton (1979). In addition,
“Free-surface effect” represents the calculated trajectory when only the free-surface
effect is considered.
It is observed that, firstly, the projectile trajectory differs greatly from the
experimental data if only the free-surface effect is considered, which indicates that
the wake separation and reattachment effect cannot be ignored in predicting the
projectile trajectory, especially in deep penetration. Secondly, the minimum wake
separation angle umin plays a strong role in predicting the projectile trajectory. The
predicted final position and direction of the projectile reproduce the experiment best
when the value of umin is set as 3°. The overestimation of the DOP may be due to
the uncertainty of soil properties and the incompressible assumption. Thirdly, the
rotational velocity should be considered in formulating the criterion for the wake
separation, especially when the orientation of the trajectory changes greatly.
The section of the projectile that coincides with the plane of the projectile
trajectory is shown in Fig. 6.16. The pressure on the nose in this section at different
6.4 Validation 237

Fig. 6.15 Projectile 0

trajectory with different umin
-5
projectile
-10 direction

-15

-20

Y (m)
-25 φmin=1o
φmin=2o φmin=1o, without ω 3
-30
φmin=3o φmin=2o, without ω 3
-35 φmin=3o, without ω 3
φmin=4o
-40 Experiment φmin=4o, without ω 3
Free-surface effect Target surface
-45
0 5 10 15 20 25 30 35 40 45
X (m)

Fig. 6.16 Section of the

projectile

Fig. 6.17 Pressure on 12

projectile nose at different 0.03s
time 10 0.06s
0.09s
0.12s
8
0.15s
σn (×10MPa)

2
nose tip
0 base of the nose

0.2 0.3 0.4 0.5 0.6 0.7

x (m)

time is shown in Fig. 6.17, where x represents the coordinate in ~ e1 direction. It is

found that the separation points locate at about 4 l0/5 (where l0 represents the nose
length) and the corresponding nose radius of the projectile is about 0.45d. These
findings prove the rationality of the assumption (iv) in Sect. 6.2.3.1.
Figure 6.18 shows the pressure on each node of the projectile’s section at 0.09 s.
It is found that all nodes keep contact with the target if the wake separation and
reattachment effect is not considered, while some nodes are detached from the target
238 6 Efficient Decoupled Analytical/Numerical Approach …

Fig. 6.18 Sectional pressure With the wake separation and reattachment effect
3.5
on projectile at 0.09 s Without the wake separation and reattachment effect
3.0

2.5

σn (×10MPa)
2.0 Reattachment occur

1.5

1.0

0.5 Shank tail Nose tip

0.0
-0.6 -0.4 -0.2 0.0 0.2 0.4 0.6 0.8
x (m)

if the wake separation and reattachment effect is accounted for. The separation
occurs in the middle of the projectile, which demonstrates the reasonableness of the
proposed model.
Figure 6.19 shows the effect of the wake separation and reattachment on the
resultant moments and forces, respectively. The oblique angle plane overlaps the
initial velocity plane in this experiment, thus the MX, MY, FZ equal zero no matter
whether the wake separation and reattachment effect is considered.
The effect of the wake separation and reattachment on MZ is shown in
Fig. 6.19a. It can be found that the value of MZ oscillates around the value of zero if
the wake separation and reattachment effect is not accounted for. The reason can be
explained as follows: at a given time step, the value of MZ may be positive. And the
positive value of MZ will make the nodal velocity vector of the shank in a reverse
direction in the next time step because the diameter of the cavity equals to that of
the projectile, which leads to MZ negative in the next time step.
Figure 6.19b shows the effect of the wake separation and reattachment on FX.
There exists a peak in the FX-time histories at the initial impact in both cases, which
can be explained by the initial non-axisymmetric interaction between the projectile
nose and the target as well as the free-surface effect. The value of FX is always
higher when considering the wake separation and reattachment effect, which makes
the deflection of the projectile trajectory much easier. However, the wake separation
and reattachment almost have no effect on FY, as shown in Fig. 6.19c. The reason
can be explained as follows: the separation occurs mainly on the projectile shank,
where the nodal pressure is relatively small compared with the projectile nose due
to relatively small nodal velocity of the shank.
The finial position of the cavity on the X–Y plane is shown in Fig. 6.20a, and the
corresponding dimensionless diameter of the cavity (the ratio of the cavity diameter
with the projectile diameter) is shown in Fig. 6.20b. Note that the diameters of the
cavity in Fig. 6.20a are enlarged by ten times in order to make the trajectory of the
6.4 Validation 239

Fig. 6.19 Wake separation (a)

and reattachment effect on the 2 with the wake separation and reatachment effect
resultant moment and force without the wake separation and reatachment effect
a MZ b FX c FY
1

Mz (×104 Nm)
0

-1

-2

0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18

Time (s)

(b) with the wake separation and reatachment effect

2.0 without the wake separation and reatachment effect

1.5
FX (×10 6 N)

1.0

0.5

0.0

0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18

Time (s)

(c) 0.0
with the wake separation and reatachment effect
without the wake separation and reatachment
-0.3
FY (×106 N)

-0.6

-0.9

-1.2

-1.5

-1.8
0.02 0.04 0.06 0.08 0.10 0.12 0.14 0.16 0.18
Time (s)
240 6 Efficient Decoupled Analytical/Numerical Approach …

Fig. 6.20 Predicted cavity (a)

a final position -5
b dimensionless diameter
-10

-15

-20

Y (m)
with the wake separation
and reachement effect
-25
without the wake separation
and reachement effect
-30

-35

-40
0 5 10 15 20 25 30 35 40
X (m)
(b)
2.0
Dimensionless diameter of the cavity

with wake separation and reachement

1.8 without wake separation and reachement

1.6

1.4

1.2

1.0

0.8
4 8 12 16 20 24 28 32 36
X (m)

cavity clearly. It is found that the diameter of the cavity is larger than that of the
projectile when the wake separation and reattachment effect is considered, and the
larger values of the cavity diameter occur in the initial period of penetration. This
reason can be explained as follows: the initial non-axisymmetric impact condition
and the free-surface effect introduce a large lateral velocity V2 that will enlarge the
diameter of the cavity (see Eqs. 6.49 and 6.50). This phenomenon has also been
observed experimentally for the cone-shaped projectile penetrating into soil-type
targets (Osipenko 2009). The cavity diameter decreases quickly at the end of the
penetration after a smooth stage. However, the diameter of the cavity equals to that
of the projectile when the wake separation and reattachment effect is not
considered.
6.4 Validation 241

6.4.2 Projectile Perforations

6.4.2.1 Normal Perforation

Based on the proposed analytical/numerical approach, we analyze the perforation

experiments of concrete slabs with unconfined compressive strengths of 48 and
140 MPa, and further compare the residual velocities with experimental data
(Hanchak et al. 1992). The effect of the bottom surface of concrete slabs is con-
sidered, the effect of the rebar, however, is not considered. Recent report by
Dancygier (1997) indicates that slight reinforcement has little effect on the residual
velocity. Figure 6.21a shows that the predictions agree well with the test data for
the 48 MPa concrete slabs when the free-surface effect is considered. However, the
result without free-surface effect underestimates the residual velocity, which is
induced by the overestimation of the forcing function. In addition, for the concrete
slabs with unconfined compressive strength of 140 MPa, the predictions by the
parameter S recommended by Forrestal et al. (1996) are not satisfied. The reason

Fig. 6.21 Comparisons (a) 1200

between test data and Experimental data (Hanchak et al. (1992))
Simulation with free-surface effect
numerical results with the 1000 Simulation without free-surface effect
concrete slabs strengths of
a 48 MPa b 140 MPa 800
Vr (m/s)

600

400

200

0
300 450 600 750 900 1050 1200
V0 (m/s)

(b) 1200
Experimental data (Hanchak et al. (1992))
Simulation with free-surface effect for
1000 S=82.6fc-0.544 Forrestal et al. (1996)
Simulation with free-surface effect for
800 S=127.7fc-0.675 Wu et al. (2012a)
Vr (m/s)

Simulation without free-surface effect

600

400

200

0
300 400 500 600 700 800 900 1000 1100
V0 (m/s)
242 6 Efficient Decoupled Analytical/Numerical Approach …

Fig. 6.22 S versus fc curves

(Wu et al. 2012a) 24 Caculated data
Fitted curve
20 Forrestal et al. (1996)

16

S
12
S=127.7×fc-0.675
8

0 20 40 60 80 100 120 140 160 180 200

fc (MPa)

lies in that the parameter is limited by available test data, and the application range
of S is 13.5 MPa < fc < 97 MPa. A new equation for S ¼ 127:7fc0:675 (fc unit in
MPa) with the consideration of the high strength concrete is proposed by Wu et al.
(2012a), as shown in Fig. 6.22. Good agreement is observed if the new empirical
strength parameter is adopted.

6.4.2.2 Oblique Perforation

Based on the analytical/numerical approach in which the free-surface effect is

considered at both the front and rear surfaces, we analyse the experimental data by
Fan and Zhou (2000), in which the projectiles oblique impact (30°) the concrete
slabs with the unconfined compressive strength of 42 MPa and various thicknesses
of 250–550 mm. Figure 6.23 shows the comparisons of the residual velocities and
the angles of directional change between the experimental data and predicted
results, respectively. The results are reasonable. However, the test on the concrete
slabs with the thickness of 400 mm is not well predicted. We conjecture that this
may be due to the scatter of the experimental data and the inaccuracy of the
empirical strength parameter S.

6.4.2.3 Ricochet Limit

Ricochet limits for various impact velocities are of interest. Consequently, a series
of calculations are performed to predict the ricochet limits for the projectile pen-
etrating into concrete slabs with unconfined compressive strengths of 48 MPa for a
range of impact velocities from 450 to 1100 m/s. The parameters k and s are
obtained through curve fitting to the test data in Hanchak et al. (1992). Figure 6.24
shows the numerically calculated ricochet limits for various impact velocities. The
6.4 Validation 243

Fig. 6.23 Comparisons (a) 600

between test data (Fan and 250mm thickness simulation
250mm thickness test
Zhou 2000) and predicted 500 300mm thickness simulation
results a residual velocity 300mm thickness test
b angle of directional change 400
400mm thickness simulation
400mm thickness test

Vr (m/s)
550mm thickness simulation
300 550mm thickness test

200

100

250 300 350 400 450 500 550 600 650

V0 (m/s)

(b) 20
250mm thickness simulation
18
Angle of directional change ( )

250mm thickness test

16 300mm thickness simulation
300mm thickness test
14 400mm thickness simulation
12 400mm thickness test
550mm thickness simulation
10 550mm thickness test
8
6
4
2
0

200 300 400 500 600 700

V0 (m/s)

Fig. 6.24 Comparisons 80

between numerical results and
Tate’s formula 70
)

60
Ricochiet angel (

50

40

30
Tate fomular (Tate (1979))
20 model prediction
10

0
300 450 600 750 900 1050 1200
V0 (m/s)
244 6 Efficient Decoupled Analytical/Numerical Approach …

ricochet limits increase with the rising of impact velocity, which is consistent with
the Tate’s formula shown in Eq. (6.58).
" #"
1=2 #
2 V02 qp ðL þ l0 Þ2 þ d 2 qp
tan f ¼
3
1þ ð6:58Þ
3 Yp ðL þ l0 Þd q0

where f represents the ricochet angle, qp is the projectile densities, and Yp is a

characteristic projectile strength which is closely related to the Hugoniot elastic
limit of the material. A good agreement between the numerical results and Tate’s
formula is obtained for a value of Yp = 7.5 GPa.

6.4.3 Projectile Penetration into Multi-layered Targets

6.4.3.1 DOP of Normal Impact

In order to verify the proposed model of the layering effect, we conduct the DOP
experiments with the 37 mm hard-core projectiles into three sets of layered targets.
The details of the experiments can be referred in Sect. 12.3. The geometries of the
projectiles used in the experiments are shown in Fig. 6.25. The noses of the pro-
jectiles are flat to make the layering effect more obviously. The 35CrMnSi steel
projectiles have an approximate nominal Vickers hardness of 241, average mass of
695 g. The projectiles are launched by a 37 mm power gun.

Fig. 6.25 Geometry of the

projectile a before experiment
b after experiment
6.4 Validation 245

(a) (b)
6@100

12

Fig. 6.26 Concrete target a dimension in mm b steel mesh

The first layer of the target is concrete with different unconfined compressive
strengths and different thicknesses, while the second layer is soil with a fixed
thickness of 1.5 m. Figure 6.26 shows the dimension and configuration of the steel
reinforced concrete layer. The uniaxial compressive strength and thickness of the
concrete layer for three sets are 52, 58, 97 MPa, and 35, 25, 20 cm, respectively.
The targets are cast in the steel tubes with the diameter of 1 m. The
pressure-dependent shear strength parameters k and sfor the concrete and soil are
obtained by curve fitting to the triaxial compression test data. These values along
with other parameters (density, Young’s modulus) are summarized in Table 6.2.
In Fig. 6.27, we present the post-test photographs of the impact surface for the
58 MPa concrete layered targets with the initial velocity of 580 m/s as well as the
97 MPa concrete layered targets with the initial velocity of 528 m/s, respectively.
In the analysis, we simplify the projectile as a truncated-conical nose projectile
due to the complex curve and treat it as a rigid body because the deformation of the
projectile can be neglected, as shown in Fig. 6.25. The real and simplified
geometries of the projectiles are presented in Fig. 6.28. For truncated-conical nose
projectile, we adopt the analytical model proposed by Jan and Henrik (2004) in
which they redefine the transition point between the cratering and tunneling regions
according to a more physical criterion and argued that cavity expansion theory can
be applied to penetration of flat nose projectiles. According to the model proposed
by Jan and Henrik (2004), the cratering region finishes when the nose has com-
pletely entered the target and the target resistance increases linearly in the cratering
region. For the convenience of computation, the forcing function in the cratering
region is determined by the method discussed in Sect. 6.2. Figure 6.29 shows the
comparison of the experimental and predicted DOP. Firstly, it is observed that the
predictions agree well with the experimental results if the layering effect is taken
into account. The underestimation of DOP may be due to the simplification of the
projectile shape and the assumption of incompressible targets. However, the
numerical results without considering the layering effect underestimate the test data
246 6 Efficient Decoupled Analytical/Numerical Approach …

Table 6.2 Parameters of the layered targets

H (cm) fc (MPa) E (GPa) q0 (kg/m3) k s (MPa)
Concrete 35 52 24 2460 0.65 41
25 58 26 2460 0.7 44
20 97 33 2460 0.67 75
Soil 150 – 0.08 1700 0.65 0.8

Fig. 6.27 Post-test photographs of the impact surface a 58 MPa concrete layered targets with
V0 = 580 m/s b 97 MPa concrete layered targets with V0 = 528 m/s

Fig. 6.28 Geometries of the (a) (b)

projectile (mm) a real
18 18
b simplified

22
30
130
63
17

33 33
37 37

about 15%. Secondly, the layering effect becomes more evident as the projectile
approaching to the interface of the concrete and soil. The reason can be explained as
follows: the spherical cavity is getting closer to the interface as the projectile
approaching to the interface. Therefore, the radial stress at the cavity surface
becomes smaller, as indicated in Fig. 6.6.
6.4 Validation 247

Fig. 6.29 Comparison of (a) 36

DOP for the composite targets Experiment
a fc = 52 MPa, H = 35 cm 32 With layering effect
b fc = 58 MPa, H = 25 cm Without layering effect
c fc = 97 MPa, H = 20 cm 28

h pen (cm)
24

20

16

12
480 520 560 600 640 680 720 760 800
V0 (m/s)

(b) 26
Experiment
24
With layering effect
Without layering effect
22
h pen (cm)

20

18

16

14

12
480 510 540 570 600 630
V0 (m/s)

(c) 20
Experiment
18 With layering effect
Without layering effect

16
h pen (cm)

14

12

10
480 500 520 540 560 580 600
V0 (m/s)
248 6 Efficient Decoupled Analytical/Numerical Approach …

6.4.3.2 Trajectory of Oblique Impact

Projectile trajectory is an important issue during earth penetration because it does

not always follow a straight path. Under certain circumstances, the projectile may
have a “J-shape” curve trajectory and even result in a reverse trajectory direction. It
should be noted that the previous studies on the projectile trajectory during pene-
tration are based on the monolithic target assumption (e.g., Li and Flores-Johnson
2011). However, in order to reach the desired position, earth penetrators usually
need to penetrate through multi-layered targets. To the best of our knowledge, no
similar analysis on the projectile trajectory in layered targets is available in the
public literature based on the cavity expansion model.
In order to investigate the influence of the layering effect on the projectile
trajectory at oblique impact, we perform a series of predictions. The projectile and
layered targets are presented in Fig. 6.30. The projectile is modeled as a rigid body
due to the concerned impact velocity is less than 600 m/s. The projectile impacts
the layered targets with various velocities and oblique angles. The empirical
parameters of the layered targets are given in Table 6.1. Figure 6.31 illustrates the
predicted projectile trajectories at various conditions. Firstly, it can be observed that
the influence of the layering effect on the projectile trajectory is obvious, and large
oblique angle can produce ricochet. Thus, the layering effect should be taken into
account if the final position and the trajectory of projectile are the main concern.
Secondly, with the increase in the oblique angle, the layering effect on the projectile
trajectory is more evident. This attributes to that the layering effect and the initial
non-axisymmetric impact condition lead to a large moment about the centroid of the
projectile, which make the orientation of the trajectory change greatly.

6.5 Numerical Examples

Based on the proposed terminal trajectory software, in this section, two scenarios
that small diameter bomb (SDB) perforating three spaced concrete slabs and
impacting a typical underground structure are calculated and discussed. The
geometry of SDB is shown in Fig. 6.32, which has the mass of 113.4 kg.

(a) (b)
Soil: 0.5m
285.6
mm
95.2mm

Concrete: 1m

516mm 158mm
Soil: infinite

Fig. 6.30 Geometries of the a projectile and b layered targets

6.5 Numerical Examples 249

(a) 0.0
300m/s with layering effect
(b) 0.0 300m/s with layering effect
300m/s without layering effect 300m/s without layering effect
-0.3 450m/s with layering effect -0.3 450m/s with layering effect
450m/s without layering effect 450m/s without layering effect
-0.6 600m/s with layering effect -0.6 600m/s with layering effect
600m/s without layering effect 600m/s without layering effect
Y (m)

Y (m)
-0.9 -0.9

-1.2 -1.2

-1.5 -1.5

-1.8 -1.8
0.0 0.1 0.2 0.3 0.4 0.5 0.0 0.2 0.4 0.6 0.8 1.0 1.2
X (m) X (m)

(c) 0.0 300m/s with layering effect

(d) 0.0
300m/s without layering effect
-0.3 450m/s with layering effect -0.2
450m/s without layering effect
600m/s with layering effect -0.4
-0.6 600m/s without layering effect
Y (m)

Y (m)
-0.6
-0.9 300m/s with layering effect
-0.8 300m/s without layering effect
450m/s with layering effect
-1.2 450m/s without layering effect
-1.0
600m/s with layering effect
600m/s without layering effect
-1.5 -1.2
0.0 0.3 0.6 0.9 1.2 1.5 1.8 0 1 2 3 4 5
X (m) X (m)

Fig. 6.31 Projectile trajectories with initial obliquity angle of a 15° b 30° c 45° d 60°

Fig. 6.32 Geometry of SDB

CRH=3 1.5761m
1.8288m

6.5.1 Perforation of Spaced Concrete Slabs

Shown in Fig. 6.33a, a SDB impacting on a typical three spaced C30 concrete slabs
with the striking velocity of 300 m/s and oblique angle of 30° is predicted by the
proposed terminal trajectory software. Since the concrete target is relatively “hard”
and its thickness is small, the minimum wake separation angle should be small,
consequently it is assumed as zero in the calculation. In other words, only the
free-surface effect is considered in this analysis. The predicted projectile trajectories
at different time are shown in Fig. 6.33b–f. Due to the free-surface effect and the
250 6 Efficient Decoupled Analytical/Numerical Approach …

(a) (b) (c)

30°
28.1°
0.3m

3m
C30 concrete
0.3m

3m
0.3m

(d) (e) (f)

25.1° 15.1°

13.5°

Fig. 6.33 a Geometry of three spaced slabs and perforation processes of SDB projectile at b 0 ms
c 9.2 ms d 15.8 ms e 32.5 ms f 47 ms

initial asymmetric impact, the deflection of the projectile trajectory is clearly

observed.
Figure 6.34 presents the schematic diagram of the trajectory deflections. Due to
the free-surface effect (rr2 [ rr1 ) and the asymmetric of the initial impact
(force-bearing area II > I), an anticlockwise moment Mz is generated, which
induces the deflection of the trajectory and a negative attack angle (positive is
defined in anticlockwise direction between the projectile axial and velocity direc-
tion). When the side I of the projectile surface proceeds further from the free surface
(position 2 in Fig. 6.34a), the free-surface effect becomes less important, conse-
quently the projectile trajectory becomes stable. It is worth pointing out that Mz is
dependent on the initial oblique angle, velocity, attack angle, and location of the
mass centre. Ricochet may happen when the initial velocity or the oblique angle is
relatively large (Fang et al. 2014a). When the projectile approaches to the rear
surface of the concrete slab (position 3 in Fig. 6.34b), due to the rear free-surface
effect, the forcing function in side II is less than that in side I. Thus, a clockwise
moment Mz and a positive attack angle are induced, which makes the projectile
trajectory clockwise deflect. A clockwise angular velocity xz is generated when the
projectile perforates the first slab, as shown in Fig. 6.34b.
6.5 Numerical Examples 251

(a) (b)

(c) (d)

Fig. 6.34 Schematic diagram of projectile trajectory deflection a front surface of 1st slab b rear
surface of 1st slab c 2nd slab d 3rd slab

When the projectile is located between the first and second slabs, the positive
attack angle becomes larger due to the contribution of xz (position 5 in Fig. 6.34c).
When the projectile starts to perforate the second concrete slab (position 6 in
Fig. 6.34c), it deflects to the free surface which results from the contributions and
counterbalance of both the free-surface effect and the positive attack angle. Then,
the free-surface effect becomes less important as the projectile proceeding (position
7 in Fig. 6.34c), and consequently the projectile trajectory becomes stable. When
the projectile approaches to the rear surface of the second concrete slab (position 8
in Fig. 6.34c), due to the rear free-surface effect, the forcing function in side II is
less than that in side I. Thus, a clockwise moment Mz and a positive attack angle are
induced, which makes the projectile trajectory clockwise deflect. After which
(position 9 and 10 in Fig. 6.34d) the situation is the same as Fig. 6.34b and no
further discussions are needed. Finally, the SDB stops at the third slab.
252 6 Efficient Decoupled Analytical/Numerical Approach …

6.5.2 Penetration into Underground Structures

A typical underground protective structure is shown in Fig. 6.35. A SDB impacting

the structure with the striking velocity of 400 m/s and oblique angle of 30° is
predicted by the proposed terminal trajectory software, in which the layering effect,
free-surface effect, as well as the wake separation and reattachment effect are
considered. In the analysis, the minimum wake separation angles are assumed as 0°
and 1° for the concrete and soil targets, respectively.
The predicted projectile trajectories at the different time are shown in Fig. 6.36.
Due to the asymmetric of the initial impact, the free-surface effect, the layering
effect, as well as the wake separation and reattachment effect, the oblique angle
becomes larger after perforating the bursting layer, as shown in Fig. 6.36b. Then,
due to the contribution of the free-surface effect, the projectile still has the angular
velocity after perforating the first concrete slab, which makes the attack angle lager
and reduces the projectile penetration capability. The SDB projectile finally stops at
the second concrete slab, illustrated in Fig. 6.36d.
Figure 6.37 presents the schematic diagram of the projectile trajectory. When the
projectile enters into the first soil layer (position 1), due to the free-surface effect
(rr2 [ rr1 ) and the asymmetric of the initial impact (force-bearing area II > I) (the
layering effect is not obvious since the projectile is far from the soil-concrete
interface), an anticlockwise moment Mz is generated, which induces the deflection
of trajectory. The projectile proceeds to move, and due to the wake separation and
reattachment effect, the anticlockwise deflection becomes more obvious. When the
projectile approaches to the soil-concrete interface, the layering effect makes
rr2 [ rr1 and consequently the projectile continues to deflect anticlockwise.
When the projectile enters into the concrete layer (position 2), due to the layering
effect (rr2 [ rr1 ) of the side I of the projectile surface (the layering effect of side II
0.5 0.5

Soil
C30 concrete

Soil
1

0.1

C30 concrete
3

Air
33

0.15 4.8 4.8 4.8

Fig. 6.35 Geometry of a typical underground structure (unit in m)

6.5 Numerical Examples 253

(a) (b)
30°

30.5°

(c) (d)

34.4°
34.9°

Fig. 6.36 Locations of the projectile at the time of a 0 ms b 8.5 ms c 30 ms d 48.4 ms

Fig. 6.37 Schematic diagram of trajectory deflection

is not obvious since it is far from the concrete-soil interface), the projectile con-
tinues to deflect anticlockwise. When the projectile approaches to the concrete-soil
interface (position 3), due to the layering effect of the side I of the projectile surface
(rr2 \rr1 ), a clockwise moment is induced, consequently the projectile starts to
deflect clockwise.
254 6 Efficient Decoupled Analytical/Numerical Approach …

When the projectile moves into the third soil layer, the wake separation and
reattachment effect makes it keep deflecting clockwise. As the projectile entering
into the first concrete slab of the underground structure (position 4), due to the
layering effect (rr2 [ rr1 ), an anticlockwise moment Mz is generated, which
induces the anticlockwise deflection of the trajectory. When the projectile perforates
the first concrete slab of the underground structure (position 5), due to the
free-surface effect, the projectile starts to clockwise deflect. After that, the situation
is the same as the one shown in Fig. 6.34c. Finally, the projectile stops at the
second concrete slab of the underground structure.

6.6 Summary

In this chapter, a decoupled approach is adopted to study the terminal ballistic

trajectory of a projectile, in which the projectile is modelled by an explicit transient
dynamic finite element code and the target is represented by a forcing function as
the pressure boundary condition. This approach eliminates the need to discretize the
target by a large number of finite elements as well as the complex contact algorithm.
Three effects that is the free-surface effect, the layering effect, as well as the sep-
aration and reattachment effect, are firstly addressed and the influential functions are
constructed to consider the above effects on the terminal ballistic trajectory of the
projectile penetrating into diversified targets. Then the forcing functions for the
projectile penetration and perforation are obtained by multiplying the
semi-empirical forcing function for the homogenous infinite depth target by the
above-mentioned decay, layering, and the contact functions. The proposed
methodology is implemented into ABAQUS explicit solver via the user subroutine
VDLOAD. Besides, a prediction approach for terminal ballistic trajectory of rigid
projectiles is proposed based on the 3D rigid dynamics, and the corresponding
software is complied. Selected prediction results are presented and compared with
the available experimental results. Good results are observed when the free-surface
effect, the layering effect, and the separation and reattachment effect are considered,
which demonstrates that the proposed methodology is efficient. Finally, based on
the trajectory software, the perforation of three spaced concrete slabs and pene-
tration of a typical underground structure by a small diameter bomb (SDB) are
calculated and discussed. The proposed methodology can be utilized for the design
of penetrators and prediction of the projectile terminal ballistic trajectory.
Chapter 7
Numerical Simulation of Projectile Impact
on Concrete Targets

7.1 Introduction

The numerical simulation of the so-called hydrocode aims to solve the conservation
equations (mass, momentum, and energy) with the initial and boundary conditions
in each premeshed element. The commercial hydrocodes, such as LS-DYNA and
ATUODYN, are usually employed for the numerical simulation of projectile
penetration. These codes can well deal with the intense dynamic loading where the
extremely high pressure, high strain rate, and large deformation are usually induced.
It can be classified into Lagrangian, Eulerian, arbitrary Lagrangian–Eulerian
approaches, etc., based on different representations of the conservation laws for a
continuum body.
Compared with the analytical approach, numerical simulation can reproduce the
projectile penetration process in a more clear and precise way, in which the con-
cerned quantities, such as stress, strain, and damage, can be obtained for any times
and any positions. Consequently, the reliable numerical simulation results can
provide the support and evidence for the establishment of analytical models, such as
the eroding long-rod penetration model proposed by Walker and Anderson (1995),
in which the assumed particle velocity profile is based on the simulation results.
Besides, since the projectile penetration experiments are always expensive, detailed
parameter studies by experiment are infeasible. However, these studies can be
easily performed economically on the computer, providing insight into the influence
of each parameter to terminal ballistic parameters systematically.
It is generally accepted that numerical method for solving the conservation
equations in these commercial hydrocodes, such as the central difference method used
in LS-DYNA EXPLICIT, has enough precision for the engineering applications.
However, for the material model of concrete materials, as will be discussed in detail in
Sect. 7.2.2, there are still some shortages of the frequently used material models,
which results from the complexity of the concrete material and the high pressure, high
strain rate, as well as the large deformation induced by projectile impact.

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 255
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_7
256 7 Numerical Simulation of Projectile Impact on Concrete Targets

This chapter focuses on the projectile penetration simulation at macroscopical

scale where the material is considered as homogeneous. Firstly, the frequently used
concrete models in the commercial hydrocodes are briefly introduced, in which the
shortcomings of these models are discussed in detail. Then, calibrations of HJC
model parameters for both NSC and HSC are conducted. Computation of the
effective strain rate in the material models is revisited, and it is found that the
effective strain rate under the uniaxial stress state computed by the most hydrocodes
is larger than the axial strain rate, which may overestimate the dynamic strength of
the concrete material when the SHPB test data is directly applied. Based on the
relationship between the effective strain rate and axial strain rate, a correction
method is proposed. In addition, the original HJC and K&C models are modified to
better simulate the cratering and scabbing phenomena induced by the projectile
penetration, which are validated by comparing with the test data and the simulation
results of the original models.

7.2 Material Models in Hydrocodes

7.2.1 General Description

In the solution process of numerical simulation, the strain increment tensor during a
time step in an element is firstly determined by the differentiation of displacement
solved by the hydrocodes. In order to know the stress state after this time step, the
stress increment tensor needs to be determined by the material model. In other
words, the purpose of the material model is to update the stress state when a strain
increment tensor is given. For all material models in hydrocodes, there are two sets
of data which have to be specified. One is the constitutive relation that is used to
describe the deviatoric behavior of the material. The other is the equation of state
used to describe the compaction behavior of the material.
In tensor symbolic notation, the total stress rij is composed of deviatoric part r0ij
and spherical part rii . Correspondingly, the total strain eij is composed of deviatoric
part e0ij and spherical part eii . The relation of the deviatoric stress r0ij and deviatoric
strain e0ij is governed by the constitutive relation, while the relation of the spherical
stress rii and spherical strain eii is governed by the equation of state.
For the elastic material, the constitutive relation is simple, which is given by

r0ij ¼ 2Ge0ij ð7:1Þ

where G is the shear modulus. However, for the elastic-plastic material, especially
for the concrete material, the constitutive relation is rather complex, which is
usually represented by the following yield surface:
7.2 Material Models in Hydrocodes 257

pffiffiffiffiffiffiffi  
3J2  f r0ij ; P; k ¼ 0 ð7:2Þ

P
where J2 is the second deviatoric stress invariant, P ¼ 31 rii is the hydrostatic
pressure, and k is the internal variable related to the plastic strain.
An appropriate flow rule is required to determine the plastic strain, which has the
general form:

@g
e_ pij ¼ u_ ð7:3Þ
@rij

where e_ pij is the plastic strain rate tensor, u_ is the consistency parameter, and g is the
plastic potential function determined by
pffiffiffiffiffiffiffi
g¼ 3J2  xf ð7:4Þ

When x ¼ 1, Eq. (7.4) defines the full associative flow rule; x ¼ 0 defines
non-associative and no dilatancy plastic flow; 0\x\1 defines that fractional
associative plastic flow. According to Eqs. (7.2–7.4), the deviatoric stress incre-
ment dr0ij is found to be

@g
dr0ij ¼ 2Gde0eij ; deeij ¼ deij  depij ¼ deij  du ð7:5Þ
@rij

where the prefix d refers to the increment, the superscript e refers to the elastic part,
and the superscript p refers to the plastic part.
The
P3 
equation of state (EOS) is used to describe the relation of pressure
P3
P 1 rii and the volumetric strain l ¼ 1 eii . Concrete has a complex nonlinear
compression behavior, which results from its large-scale heterogeneity and porosity
of typically 10%. Consequently, the EOS for concrete material is always piecewise,
such as the EOS in HJC model (Holmquist et al. 1993) and the well-known P−a
EOS (Herrmann 1969). When a proper EOS is introduced, the pressure can be
easily obtained by the volumetric strain at the current time step.

7.2.2 Frequently Used Concrete Models

A comprehensive review and evaluation of several widely used concrete material

models in commercial hydrocodes are presented by Tu and Lu (2009). This section
briefly introduces the frequently used HJC model and K&C model (Malvar et al.
1997), in which the shortcomings of these models are discussed in detail. Readers
are referred to the original papers of these models for further information.
258 7 Numerical Simulation of Projectile Impact on Concrete Targets

7.2.2.1 HJC Model

In the HJC model, the deviatoric response is governed by the following yield
surface:
 
req ¼ Að1  DÞ þ BPN ð1 þ C ln e_  Þ  Smax ; P  0 ð7:6Þ
pffiffiffiffiffiffiffi
where req ¼ 3J2 =fc is the normalized equivalent stress. fc is the unconfined
uniaxial compressive strength. Smax is the normalized maximum strength that the
concrete material can withstand. P ¼ P=fc denotes the normalized pressure. e_  ¼
e_ =_e0 represents the dimensionless strain rate, where e_ is the actual strain rate and e_ 0
is the reference strain rate. The normalized cohesive strength A, the normalized
pressure-hardening coefficient B, the pressure-hardening exponent N, and strain rate
coefficient C are material constants. D is the accumulated damage determined by

X Dep þ Dlp
D¼ ð7:7Þ
epf þ lpf

where Dep and Dlp are the effective plastic strain increment and plastic volumetric
strain during a cycle of integration, respectively. epf þ lpf is the total plastic strain
under a constant pressure until fracture, which is determined by

epf þ lpf ¼ D1 ðP þ T  ÞD2  EFMIN ð7:8Þ

where T  ¼ T=fc is the normalized tensile strength and T is the unconfined uniaxial
tensile strength. EFMIN is a material constant used to suppress fracture from weak
tensile waves. D1 and D2 represent the damage constants, respectively.
For the negative pressure ðP \0Þ, it seems that a linear dependence between P
and req is assumed, as shown in Fig. 7.1, which results in the following yield
surface:

AP
req ¼ þA P \0: ð7:9Þ
T  ð1  D Þ

It can be observed from Eqs. (7.7 and 7.9) that the tensile behavior of HJC
model is damage dependent, i.e., it is not perfect elasto-plastic for tension.
However, as will be shown in Sect. 7.4.2, tensile behavior of HJC model in
LS-DYNA is perfect elasto-plastic, indicating that it may induce an error during the
implementation of the yield surface for the negative pressure. In addition, the yield
surface is discontinuous at P ¼ 0 when the damage D starts to accumulate, as can
be seen from Eqs. (7.6 and 7.9).
7.2 Material Models in Hydrocodes 259

Fig. 7.1 Equation of state in

HJC model (Holmquist et al.
P
1993)
P = K1 μ + K 2 μ 2 + K 3 μ 3

Plock

Pcruch Κ Κ1

μ crush μ lock μ plock

μ
-T(1-D)

The pressure-compaction response is governed by the equation of state

(EOS) shown in Fig. 7.1. The first region is linear elastic from the negative pressure
cutoff T ð1  DÞ to the elastic limit Pcrush, beyond which the second transition
region begins. In the transition region, the air voids are gradually compressed out of
concrete and the plastic
 volumetric
 strain produces the compaction damage until
reaching the point lplock ; Plock . In this transition region, loading follows the linear
path shown in Fig. 7.1, while unloading occurs along the modified path that is
interpolated from the nearby region. Reloading follows the unloading path until it
reaches the unloading beginning point and then continues along the loading path. In
the third region, all air voids are removed from concrete that is assumed completely
nonlinear elastic. Loading and unloading follow the same path shown in Fig. 7.1,
where P ¼ K1 l þ K2 l2 þ K3 l3 with K1, K2, and K3 being the material constants,
and a modified volumetric strain l ¼ ðl  llock Þ=ð1 þ llock Þ is introduced.
HJC model does consider most of the important aspects of concrete behavior
under impact loads, such as pressure dependency, strain rate effect, damage
cracking, and compaction (Polanco-Loria et al. 2008). However, from a critical
view, the following important aspects are not taken into account.
(i) The independence of the third stress invariant cannot capture the transition of
the deviatoric section from triangular shape at low pressures to circular shape
at high pressures;
(ii) Strain hardening is not considered;
(iii) Strain softening is only dependent on the cutoff for the fracture strain that is
usually difficult to determine;
(iv) Strain rate enhancement for compression and tension is identical, which is
contrary with the available experiments (Malvar and Ross 1998; Hao et al.
2012);
(v) Tensile damage is not considered, which may play an important role in the
simulation of scabbing (Xu and Lu 2006; Li and Hao 2014).
260 7 Numerical Simulation of Projectile Impact on Concrete Targets

7.2.2.2 K&C Model

The K&C material model is first developed for DYNA3D (Malvar et al. 1997) and
now is available in LS-DYNA as material #72. K&C model uses three independent
strength surfaces of compressive meridians, namely the initial yield strength surface
Dry , the maximum strength surface Drm , and the residual strength surface Drr ,
which are summarized as follows.
8  
< a0y þ P= a1y þ a2y P ; P  0:15fc
Dry ¼ 1:35T þ 3Pð1  3T=fc Þ; 0  P  0:15fc ð7:10Þ
:
1:35ðP þ T Þ; P  0
8
< a0 þ P=ða1 þ a2 PÞ; P  fc =3
Drm ¼ 1:5=/ðP þ T Þ; 0  P  fc =3 ð7:11Þ
:
3ðP=g þ T Þ; P  0

Drr ¼ a0f þ P=ða1f þ a2f PÞ ð7:12Þ

where ai , aiy , and aif (i = 0, 1, 2) are the constants, which need to be determined
from a suitable set of triaxial compression data. u denotes the tensile-to-
compressive meridian ratio. Advantages of these strength surfaces, especially at
p < fc/3, are illustrated by Tu and Lu (2009) and not repeated here. A typical view
of the strength surfaces is shown in Fig. 7.2.
The yield surface is determined as follows:
0   
pffiffiffiffiffiffiffi r g Drm  Dry þ Dry ; strain hardening
0 ¼ 3J2  ð7:13Þ
r 0 ½gðDrm  Drr Þ þ Drr ; strain softening

where r 0 is the ratio of the current meridian to the compressive meridian. η is the
yield scale factor related to the damage function k which is defined as

Fig. 7.2 Three independent strength surfaces

7.2 Material Models in Hydrocodes 261

8 Rt
>
> b1
< dep =ð1 þ P=T Þ ; P[0
k¼ 0
Rt ð7:14Þ
>
>
: dep =ð1 þ P=T Þb2 ; P0
0

qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
where dep ¼ 2=3depij depij is the effective plastic strain increment with depij being
the plastic strain increment tensor. The damage constants b1 and b2 usually have
different values to consider different damage evolution of concrete for compression
and tension. The yield scale factor η follows a general trend, i.e., it varies from zero
to unity when the current failure surface changes from the initial yield strength
surface to the maximum strength surface, corresponding to the strain-hardening
branch. In addition, it varies from unity back to zero when the current failure
surface changes from the maximum strength surface to the residual strength surface,
corresponding to the strain-softening branch. It should be noted that there is no
accumulated plastic strain when the stress path is close to the triaxial tensile path,
where stress deviators approach zero. Consequently, both the damage function k
and yield scale factor η remain zero, which cannot represent the actual situation. To
overcome this problem, a volumetric damage scalar related to pressure is introduced
(Malvar et al. 1997), which is not repeated here.
To account for the strain rate effect, the current failure surface and damage
function k are modified as follows:

Dr ¼ rf DrðP=rf Þ; k ¼ kðP=rf Þ=rf ð7:15Þ

where rf is the dynamic increase factor (DIF), which is defined separately for
compression and tension, since experimental results show that DIF for tension is
much higher than compression.
K&C model uses the EOS 8 in LS-DYNA, which is defined by tabular input.
The pressure is determined by

P ¼ CðlÞ þ c0 TðlÞE0 ð7:16Þ

where µ is the volumetric strain, CðlÞ is the tabulated functions of pressure and
volumetric strain, c0 is a temperature-related constant, TðlÞ is the temperature, and
E0 is the initial internal energy. Up to ten points and as few as two points are used to
define the tabulated functions of CðlÞ; TðlÞ and bulk unloading modulus KðlÞ,
respectively. In the loading phase, pressure is defined by extrapolating of the tab-
ulated function. Unloading occurs along the unloading bulk modulus KðlÞ to the
pressure cutoff. Reloading follows the unloading path to the point where unloading
began and continues on the loading path.
Although K&C model addresses several issues that are not considered in HJC
model, there are still some disadvantages as addressed in the following:
262 7 Numerical Simulation of Projectile Impact on Concrete Targets

(i) In the version of Release III, the automatic generation of parameters is

provided, which is usually adopted by many researchers. However, auto-
matic generated parameters, such as the parameters for strength surfaces, are
suitable for relatively low pressure and may be inappropriate for projectile
penetration simulation with very high pressure.
(ii) If the automatic generation is not adopted, 48 parameters are needed to
calibrate before numerical simulation, which should be based on a large
amount of test data. However, these data, such as shear strength-pressure
data and pressure-volumetric strain data, are always absent in most projectile
penetration experiments.
(iii) It is known that b2 is dependent on the element size to ensure the amount of
fracture energy in an element, but the relationship between b2 and element
size is not provided in the original paper.
(iv) Recent experiments (Weerheijm and Van Doormaal 2007; Schuler et al.
2006) show that the fracture energy is dependent on the strain rate and
suggest that at high strain rate, the fracture strain is a constant and the
fracture energy increases with the increase of strain rate. However, the
fracture energy and fracture strain in K&C model cannot follow this trend.
(v) Since the fractionally associated flow rule is adopted in K&C model, the
effective strain rate is larger than the axial strain rate in uniaxial dynamic test.
This leads to the overestimation of dynamic strength when the DIF-_e relation
obtained from the uniaxial dynamic test data is applied directly.

7.2.3 Calibration of HJC Model Parameters

The parameters for the yield surface and EOS are crucial for the projectile pene-
tration simulation, in which the triaxial compression test data and
pressure-volumetric strain data from hydrostatic compression test or
flyer-plate-impact test are required. However, these experiments are always absent
in the projectile penetration tests. Consequently, the parameters of the material
models usually use the corresponding values in the original papers, which may not
represent the real dynamic properties of various concrete targets.
In this section, we firstly collect the test data for NSC (normal strength concrete,
fc  60 MPa) and HSC (high-strength concrete, fc  60 MPa), respectively.
Similar researches are conducted by Chen et al. (2008a), Xiong et al. (2009a, 2010),
Meyer (2011), and Fang et al. (2014b), but are not systematic, especially for HSC.
These data are then used to determine the yield surface and EOS parameters.
Furthermore, comparisons between the numerical results predicted by the suggested
parameters and the parameters in the original paper show the advantages of the
suggested parameters.
7.2 Material Models in Hydrocodes 263

7.2.3.1 NSC

The strength parameters of HJC model contain A, B, N, and Smax, where B, N, and
Smax can be determined by the triaxial compression test data. But for the normalized
cohesive strength parameter A, due to the lack of test data, Holmquist et al.
(1993) assumed that the cohesive strength is 0.75 fc for quasi-static condition
ðe_ ¼ 0:001Þ. Normalizing to e_  ¼ 1:0 gives A = 0.79. Based on the plastic yield
surface theory, Xiong et al. (2010) assumed that the compression meridians of HJC
model and Mohr–Coulomb models intersect at two points, i.e., uniaxial compres-
sion and pure shear. For the pure shear-stress state, one can obtain

1
A ¼ pffiffiffiffiffiffi ð7:17Þ
2 K0

where K0 is the slope of linear relationship between axial pressure and confining
pressure. Figure 7.3 shows the test data of axial pressure and confining pressure
obtained from the triaxial compression tests of NSC. As can be seen, the axial
pressure and confining pressure basically meet the linear relationship with
K0 = 3.21 and A = 0.28.
Without considering the influences of damage and strain rate effect, the equation
of yield surface Eq. (7.6) may be simplified as

r ¼ 0:28 þ BPN ð7:18Þ

where r ¼ ðr1  r3 Þ=fc ; P ¼ ðr1 þ r2 þ r3 Þ=3fc ; r1 is the axial pressure and

r2 ¼ r3 is the confining pressure. Based on the triaxial compression test data
shown in Fig. 7.3, the relationship of P*–r* for NSC is presented in Fig. 7.4.
Therefore, the parameters B, N, and Smax can be obtained as 1.85, 0.84, and 15 by
the curve fitting, respectively.
Based on the Hugoniot experiments conducted by Grady (1996a, b), the
parameters for EOS can be determined. For the first region, there are Pcrush ¼ fc =3

Fig. 7.3 Triaxial 700

Xie et al. (1996) 38MPa
compression test data for NSC Xiong (2009b) 34MPa
600 Xiong (2009b) 60MPa
Ansari and Li (1998) 47MPa
500 Yan et al. (2007) 9.8MPa
Yanet al. (2007)10.6MPa
σ1 (MPa)

Yanet al. (2007) 11.4MPa

400 Wu et al. (2007) 25MPa
Hou (2006) 27MPa
300 Hou (2006) 25MPa

200
Fiting curve σ 1 =3.21σ 3+40.98
100

0
0 20 40 60 80 100 120 140 160 180 200
σ 3 (MPa)
264 7 Numerical Simulation of Projectile Impact on Concrete Targets

Fig. 7.4 Determinations of 18

strength parameters for NSC Fitting curveB =1.85 N =0.84 Smax=15
15

12
Xie et al. (1996)38MPa

σ*
9 Xiong (2009b) 34MPa
Xiong (2009b) 60MPa
Ansari and Li (1998) 47MPa
6 Yan et al. (2007) 9.8MPa
Yan et al. (2007) 10.6MPa
Yan et al. (2007) 11.4MPa
3 Wu et al. (2007) 25MPa
Hou (2006) 27MPa
Hou (2006) 25MPa
0
0 2 4 6 8 10 12 14
P*

Fig. 7.5 Determinations of 25

EOS parameters Grady (1996a)
Grady (1996b)
20 Fitting curve
Third Region

0.016
15
P (GPa)

First Region
0.012

0.008
10 0.004
( μcrush , Pcrush )
0.000
0.0000 0.0005 0.0010 0.0015

5
Second Region
( μplock , Plock)
0
0.0 0.1 μ lock 0.2 0.3 0.4 0.5
μ

(Holmquist et al. 1993) and lcrush ¼ Pcrush =K. For the third region, K1 = 12 GPa,
K2 = 135 GPa, and K3 = 698 GPa are obtained from curve fitting to the experiment
data presented by Grady (1996a). For the second region, by fitting the experiment
data given by Grady (1996b), lplock = 0.16 and Plock = 1.21 GPa are obtained. The
grain density qgrain is taken as 2500 kg/m3, so that llock ¼ qgrain =q0  1 (Holmquist
et al. 1993) can be obtained, where q0 is the initial density (Fig. 7.5).

7.2.3.2 HSC

The cohesive strength parameter A is difficult to determine since there is no

available experimental data below the pressure of fc/3. Previous studies show that
A is close to 0.30 (Malvar et al. 1997, 2000). Besides, the parameter A for NSC
7.2 Material Models in Hydrocodes 265

Fig. 7.6 Determinations of 8

yield surface parameters for
HSC 7

6
Lu and Hsu (2006) 67MPa
5 Lu and Hsu (2006) 69MPa
Xie et al. (1995) 60.2MPa
Xie et al. (1995) 92.2MPa

σ∗
4 Xie et al. (1995) 119MPa
Sovják et al. (2013b) 123MPa
Sovják et al. (2013b) 148MPa
3 Farnam et al. (2010b) 76MPa
Farnam et al. (2010b) 87MPa
2 Farnam et al. (2010b) 146MPa
Farnam et al. (2010b) 171MPa
Xiong (2009b) 61MPa
1 Ren et al. (2016), 95MPa
Ren et al. (2016), 129MPa
Fitting curve
0
0 1 2 3 4 5 6 7 8
P∗

obtained from data fitting is also close to 0.30; thus, A is also assumed as 0.30 for
HSC. When the damage and strain rate effect are ignored, the yield surface
Eq. (7.6) can be simplified as

r ¼ 0:30 þ BPN ð7:19Þ

By fitting the previous and our triaxial compression test data on HSC specimens,
the strength parameters B = 1.73 and N = 0.79 could be obtained, respectively, as
shown in Fig. 7.6.
Based on the Hugoniot experiments conducted by Yan et al. (2000) and Marsh
(1980), the EOS parameters are determined as follows. As shown in Fig. 7.7, for
the third region, based on the test data presented by Marsh (1980), K1 = 116 GPa,
K2 = −243 GPa, and K3 = 506 GPa can be obtained by the curve fitting. For the

Fig. 7.7 Determinations of 25

EOS parameters for HSC
0.05
20 First Region Third Region
0.04
0.03
0.02
15
P (GPa)

0.01 ( μ crush , Pcrush )

0.00
0.000 0.001 0.002
10

5 Second Region Yan et al. (2000)

Marsh (1980)
( μ plock , Plock) Fitting curve
0
0.0 μ lock 0.1 0.2 0.3 0.4
μ
266 7 Numerical Simulation of Projectile Impact on Concrete Targets

second region, by fitting the experiment data given by Yan et al. (2000) linearly,
lplock = 0.11 and Plock = 3.47 GPa are obtained. qgrain is taken as 2753 kg/m3 (Yan
et al. 2000), so that llock ¼ qgrain =q0  1 can be obtained.

7.2.4 Validation of HJC Model Parameters

In the following, by using the finite element program LS-DYNA, the projectile
penetration and perforation tests on the NSC and HSC targets are numerically
simulated to validate the proposed parameters for HJC model.
Since the projectiles and targets are axisymmetric in the experiments, the 2D
Solid 162 element type and the 2D axisymmetric Lagrange algorithm are used. The
projectiles are modeled by the *MAT_RIGID (20#) model, and the targets are
modeled by the *MAT_JOHNSON_ HOLMQUIST_CONCRETE (111#) model
with the proposed parameters of yield surface and EOS, as well as the original
parameters for damage (D1, D2, and EFMIN) and strain rate effect (C), respectively.
The diameters of the cylindrical targets are over 20 times larger than the projectile
diameters; thus, the influence of circumferential boundary is neglected, which is set
as the free boundary. The contact between the projectile and concrete is modeled
by using the keyword “CONTACT_2D_AUTOMATIC_SINGLE_SURFACE”
implemented in LS-DYNA. In order to simulate the physical fracture or crushing of
the concrete target numerically, the erosion algorithm *MAT_ADD_EROSION is
brought in and the erosion criterion is defined by the maximum principal strain. In
the high-velocity impact analysis, when the instantaneous geometric strain of the
element reaches the above predefined value, the element is removed and would not
offer further resistance. For each test, the value of erosion strain is determined by
trial-and-error method and is assumed as the one that gives the best predictions by
the lowest, middle, and largest shooting velocities. Figures 7.8 and 7.9 show the
finite element model of the projectile penetration and perforation tests as well as the
damage contour of the targets.

Fig. 7.8 Finite element model of penetration test and the damage contour of the target
7.2 Material Models in Hydrocodes 267

Fig. 7.9 Finite element model of perforation test and the damage contour of the target

7.2.4.1 NSC

In this section, a series of projectile penetration and perforation tests on NSC targets
are numerically simulated to validate the proposed parameters for the yield surface
and EOS in HJC model.
Table 7.1 gives the related parameters for all sets of rigid projectile penetration
and perforation tests on the NSC targets, where d, CRH, and V0 are the diameters of

Table 7.1 Parameters of penetration and perforation tests (Frew et al. 1998; Forrestal et al. 1994,
1996, 2003a; Gran and Frew 1997; Hanchak et al. 1992; Wu et al. 2015a; Cargile et al. 1993)
Test No. d (mm) CRH V0 (m/s) fc (MPa) D (mm) H (mm)
Penetration 1 20.3 3 442–1165 58.4 510 940–2280
30.5 3 445–1225 58.4 910 1070–3050
2 26.9 2 277–800 36 1370 760–1830
3 12.9 3 371–1126 13.5 300 310–1040
12.9 4.25 345–1063 13.5 300 310–1030
4 12.9 3 492–1142 21.6 410 760–1270
12.9 4.25 473–1190 21.6 410 760–1270
5 30.5 3 405–1201 51 910 1830–2740
6 76.2 3 139.3–378.6 23 1830 1220, 1830
76.2 6 238.4, 378.6 23 1830 1830
7 76.2 3 238.1–456.4 39 1830 1220, 1830
76.2 6 312.5, 448.5 39 1830 1830
8 50.8 3 315 43 1370 1220
Perforation 9 25.4 3 301–1058 48 610 178
10 25.3 3 292–729 41 675 100
11 25.3 3 536–620 41 675 150
12 25.3 3 540–651 41 675 200
13 25.3 3 540–737 41 675 300
14 50.8 3 306–319 36.5 1520 127
50.8 3 311, 325 39.9 1520 215.9
50.8 3 309, 318 36.5, 39.9 1520 254
50.8 3 309, 317 39.9 1520 284.4
15 50.8 3 309–473 36.5, 39.9 1520 254
268 7 Numerical Simulation of Projectile Impact on Concrete Targets

ogive-nosed projectiles and caliber-radius-head and striking velocities, respectively.

D and H are the diameters and thickness of the targets, respectively.
Figure 7.10 shows the comparisons between the test data and numerical results
of DOP for the penetration tests 1–7 given in Table 7.1. It indicates that for the
projectile penetration tests on the semi-infinite NSC targets, the numerical results of
DOP by using the proposed parameters are in good agreement with the test data.
In Test 6, by conducting the projectile penetration tests on semi-infinite NSC
targets (23 MPa) with the striking velocities of 139.3–378.6 m/s, Forrestal et al.
(2003a) recorded the deceleration-time histories of the projectile for the different
striking velocities. Comparisons of the experimental and numerically predicted
deceleration-time histories are shown in Fig. 7.11, and good agreements are
observed.
In Test 8, Gran and Frew (1997) conducted projectile penetration tests on NSC
targets with the projectile striking velocities of 315 m/s. Three stress gages are
embedded at various depths and radii from the penetration path to measure the
radial stress, and their positions are shown in Fig. 7.12. Figure 7.13 shows the
comparisons of the test data and numerical results of the in-target radial stress-time
histories. It indicates that the numerical results of the rising phase and peak stress by
using the above-determined parameters are in good agreement with the test data,
and quite differences of the residual stress occur in the falling phase.
For the projectile perforations, Fig. 7.14 shows the comparisons of the test data
and numerical results for the residual velocities Vr in the perforation Test 9–Test 15
shown in Table 7.1. It is found that the numerically predicted residual velocities by
the proposed parameters are in good agreement with the test data. Figure 7.14b–e
also shows that the rear steel liner with the thickness of 1 mm has little influence on
the residual velocity, which is consistent with the test observation given by Wu
et al. (2015a).
Wu et al. (2015a) conducted projectile perforation tests on the NSC slabs with
the thickness of 300 mm and the striking velocity was 641.5 m/s. The projectile
perforation deceleration-time histories are recorded by the accelerometers, as shown
in Sect. 12.2. Comparison of experimental and numerically predicted projectile
deceleration-time history during the perforation process is shown in Fig. 7.15. It is
found that the numerical predictions based on the above-determined parameters are
in good agreement with the test data, and the proposed parameters can be used to
predict the projectile deceleration-time history.

7.2.4.2 HSC

In order to validate the proposed yield surface and EOS parameters for HSC, a
series of projectile penetration and perforation tests on HSC targets are numerically
simulated. Correspondingly, the simulation results by the original parameters
(Holmquist et al. 1993) are also presented for comparison purpose. Table 7.2 gives
the related parameters of the above projectile penetration and perforation tests on
the HSC targets.
7.2 Material Models in Hydrocodes 269

(a) 3.0 Frew et al. (1998) (d =20.3mm)

(b) 1.2 Forrestal et al. (1994)
Proposed parameters (d =20.3mm) Proposed parameters
2.5 1.0
Frew et al. (1998) (d =30.5mm)
Proposed parameters (d =30.5mm)
2.0 0.8
DOP (m)

DOP (m)
1.5 0.6

1.0 0.4

0.5 0.2

0.0 0.0
200 400 600 800 1000 1200 1400 200 300 400 500 600 700 800 900
V0 (m/s) V0 (m/s)

(c) 1.0 (d) 1.0

Forrestal et al. (1996) (CRH=3.0) Forrestal et al. (1996) (CRH=3.0)
Proposed parameters (CRH=3.0) Proposed parameters (CRH=3.0)
0.8 Forrestal et al. (1996) (CRH=4.25) 0.8 Forrestal et al. (1996) (CRH=4.25)
Proposed parameters (CRH=4.25) Proposed parameters (CRH=4.25)
DOP (m)

DOP (m)
0.6 0.6

0.4 0.4

0.2 0.2

0.0 0.0
200 400 600 800 1000 1200 400 500 600 700 800 900 1000 1100 1200 1300
V0 (m/s) V0 (m/s)

(e) 2.5 (f) 1.4

Test data Forrestal et al. (2003a) (CRH=3.0)
Proposed parameters 1.2 Proposed parameters (CRH=3.0)
2.0 Forrestal et al. (2003a) (CRH=6.0)
1.0 Proposed parameters (CRH=6.0)
DOP (m)

DOP (m)

1.5 0.8

1.0 0.6

0.4
0.5
0.2
0.0 0.0
200 400 600 800 1000 1200 1400 100 150 200 250 300 350 400
V0 (m/s) V0 (m/s)

(g) 1.2 Forrestal et al. (2003a) (CRH=3.0)

1.0 Proposed parameters (CRH=3.0)
Forrestal et al. (2003a) (CRH=6.0)
Proposed parameters (CRH=6.0)
0.8
DOP (m)

0.6

0.4

0.2

0.0
200 250 300 350 400 450 500
V0 (m/s)

Fig. 7.10 Comparisons between the test data and numerical results of penetration tests: a Test 1,
b Test 2, c Test 3, d Test 4, e Test 5, f Test 6, and g Test 7
270 7 Numerical Simulation of Projectile Impact on Concrete Targets

8
(a) Forrestal et al. (2003) (b) 10 Forrestal et al. (2003)
Proposed parameters Proposed parameters
6 8
Deceleration (10 4g)

Deceleration (10 4g)

6
4
4
2
2
0
0

-2 -2
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Time (ms) Time (ms)

(c) 10 Forrestal et al. (2003) (d) 10 Forrestal et al. (2003)

Proposed parameters Proposed parameters
8 8
Deceleration (10 4g)

6 Deceleration (10 4g) 6

4 4

2 2

0 0

-2 -2
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7 8
Time (ms) Time (ms)

(e) 10
Forrestal et al. (2003)
Proposed parameters
8
Deceleration (10 4g)

-2
0 1 2 3 4 5 6 7 8
Time (ms)

Fig. 7.11 Comparisons of experimental and predicted deceleration-time histories of projectile

during penetration: a 200 m/s, CRH = 3.0, b 250 m/s, CRH = 3.0, c 337 m/s, CRH = 3.0,
d 379 m/s, CRH = 3.0, and e 379 m/s, CRH = 6.0

Figure 7.16 presents the comparisons of the test data and numerical results. It is
observed that for the projectile penetration tests on the semi-infinite HSC targets,
the proposed parameters are in good agreement with the test data, while the original
parameters presented by Holmquist et al. (1993) cannot reproduce the experimental
data well, and the deviation increases with increase of the striking velocity. As for
the perforation test, the predicted residual velocity by the proposed parameters also
shows higher degree of accuracy.
7.3 Computation of the Effective Strain Rate 271

Fig. 7.12 Schematic diagram d=2rp=50.8mm

of the location of in-target
stress gages (Gran and Frew
1997)

356mm

83.8mm

r/rp
2.5 z/rp=4.0
2.0
1.5

Gages

7.3 Computation of the Effective Strain Rate

Apart from the yield surface and EOS, another important issue for the concrete
material models in hydrocodes is the strain rate effect, which is discovered in the
dynamic compressive tests (Abrams 1917) and must be taken into consideration in
material models. Bischoff and Perry (1991) summarized many dynamic test data as
shown in Fig. 7.17 and recommended a relation between the dynamic increase
factor (DIF) and strain rate that is adopted in CEB (CEB-FIP Model Code 1990
1993).
There are various methods used to account for the strain rate enhancement in
material models, which can be classified into three approaches. The first one is the
so-called overstress approach in which the strain rate-independent yield function
used for describing viscoplastic strain is related to the overstress represented by the
difference between dynamic stress and inviscid stress. Perzyna model (Perzyna
1966) and Duvaut–Lions model (Duvaut and Lions 1972) belong to this approach.
The second one is the “consistency approach” (Wang 1997; Winnicki et al. 2001) in
which the yield function is strain rate dependent; thus, the actual stress state must
remain on the yield surface and the standard Kuhn–Tucker condition remains valid.
The third one termed as “simplified approach” is frequently used in concrete
material models, such as K&C model in LS-DYNA. In the “simplified approach,”
272 7 Numerical Simulation of Projectile Impact on Concrete Targets

(a) 350 (b) 200 Shot 3, r/rp =2.1 Gran and Frew (1997)
Shot 3, r/rp =1.4 180 Proposed parameters
300 Shot 2, r/rp =1.5 160

Radial Stress (MPa)

(r/rp=2.0, z/rp=4.0)
Radial Stress (MPa)

Gran and Frew (1997)

250 Shot 3, r/rp =1.7 Proposed parameters 140 Shot 2, r/rp =2.2
(r/rp =1.5, z/rp =4.0) 120
200
100
150 80
100 60
40 Shot 1, r/rp =2.1
50
20
0 0
-50 -20
0.00 0.25 0.50 0.75 1.00 1.25 1.50 0.00 0.25 0.50 0.75 1.00 1.25 1.50
Time (ms) Time (ms)

(c) 140 Shot 1, r/rp =2.6

Shot 2, r/rp =2.6
120 Gran and Frew (1997)
Radial Stress (MPa)

Proposed parameters
100 (r/rp=2.5, z/rp =4.0)
80
60
40
20
0
-20
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Time (ms)

Fig. 7.13 Comparisons of test data and numerical results of in-target radial stress-time histories

the yield function is also strain rate dependent, but it is enlarged prior to plastic flow
according to the total strain rate; thus, the consistency parameter (also known as
viscoplastic multiplier) obtained by Kuhn–Tucker condition is time independent.
Otherwise stated, the strain rate enhancement approach used in this section is the
“simplified approach.”
Generally, strain rate effects are usually determined by dynamic tests with
hydraulic testing machines at low strain rates (10−5–10−1 s−1) and SHPB (split
Hopkinson pressure bar) or SHTB (split Hopkinson tensile bar) devices at high
strain rates (100–103 s−1) (Bischoff and Perry 1991; Field et al. 2004), in which
only the relation between axial stress and axial strain or strain rate can be measured.
Unfortunately, two important problems arise when taking use of the test data
obtained from uniaxial dynamic tests. One is the lateral inertia confinement in
dynamic testing. Due to the Poisson’s ratio effect and plastic flow effect, the
specimens inevitably undergo lateral deformation when subjected to longitudinal
impact loading. Since many materials, such as concrete and geological materials,
are sensitive to the confining pressure, the influences of lateral inertia confinement
on dynamic compressive behavior of these materials have been widely discussed by
many researchers (Li and Meng 2003; Lu and Li 2011a; Hao et al. 2012; Fang et al.
2014c; Magallanes et al. 2010). In a word, because of the inertia effect, the DIF-_e
relation obtained from dynamic tests at high strain rates is attributed not only to the
7.3 Computation of the Effective Strain Rate 273

(a) 1200 Hanchak et al. (1992)

(b) 700 Wu et al. (2015a) (with steel liner)
Proposed parameters Wu et al. (2015a) (without steel liner)
1000 600 Proposed parameters

800 500

Vr (m/s)
Vr (m/s)

600 400

400 300

200 200

0 100
200 400 600 800 1000 1200 250 300 350 400 450 500 550 600 650 700 750
V0 (m/s) V0 (m/s)

(c) 700
Wu et al. (2015a) (with steel liner)
(d) 600 Wu et al. (2015a) (with steel liner)
Wu et al. (2015a) (without steel liner) Wu et al. (2015a) (without steel liner)
600 Proposed parameters 500 Proposed parameters

500
400
Vr (m/s)

Vr (m/s)
400
300
300

200 200

100 100
350 400 450 500 550 600 650 700 750 400 450 500 550 600 650 700 750
V0 (m/s) V0 (m/s)

(e) 500 (f) 300

Wu et al. (2015a) (with steel liner) Cargile et al. (1993)
Proposed parameters Proposed parameters
400 250

200
300
Vr (m/s)

Vr (m/s)

150
200
100
100 50

0 0
500 550 600 650 700 750 0.12 0.15 0.18 0.21 0.24 0.27 0.30
V0 (m/s) H (m)

(g) 300
Cargile et al. (1993)
Proposed parameters
250

200
Vr (m/s)

150

100

50

0
300 350 400 450 500
V0 (m/s)

Fig. 7.14 Comparisons of test data and numerical results of perforation tests: a Test 9, b Test 10,
c Test 11, d Test 12, e Test 13, f Test 14, and g Test 15
274 7 Numerical Simulation of Projectile Impact on Concrete Targets

Fig. 7.15 Comparisons of 20

test data and numerical results Test data
of projectile perforation Proposed parameters
deceleration-time histories 15

Deceleration (10 4g)

10

-5
0 1 2 3 4 5 6
Time (ms)

material properties, but also the structural responses. It would lead to the overes-
timation of dynamic strength when the DIF-_e relation obtained from the dynamic
test data is applied directly.
Another problem is the computation of effective strain rate in numerical analysis,
which is frequently ignored in most studies. In numerical simulations for 3D
problem, strain rate is a second-order tensor. Since no component of strain rate
tensor could stand for the axial strain rate in most cases, an effective strain rate is
usually used. In summary, the following four frequently used effective strain rate
ðe_ Þ expressions shown in Eqs. (7.20a–d) are adopted in various material models.
rffiffiffiffiffiffiffiffiffiffiffiffi
2
e_ ¼ e_ ij e_ ij ð7:20aÞ
3
rffiffiffiffiffiffiffiffiffiffiffiffi
2 p p
e_ ¼ e_ e_ ð7:20bÞ
3 ij ij
rffiffiffiffiffiffiffiffiffiffiffiffi
2 0 0
e_ ¼ e_ e_ ð7:20cÞ
3 ij ij

Table 7.2 Basic parameters of penetration and perforation tests (Wu et al. 2015c; Yan 2001;
O’Neil et al. 1999; Hanchak et al. 1992)
Test No. d (mm) CRH V0 (m/s) fc (MPa)
Penetration 1 25.3 3 510, 850 67.5–125.2
2 14.5 0.5 273–793 71.7
3 26.9 2 229–796 104
4 26.9 2 229–796 157
Perforation 5 25.4 3 376–996 140
7.3 Computation of the Effective Strain Rate 275

(a) 0.6 Wu et al. (2015c) Wu et al. (2015c)

(b) 0.20
Proposed parameters
Yan et al. (2001)
Proposed parameters
0.5 Original parameters Original parameters
Proposed parameters
Original parameters
V0 =510m/s V0 =850m/s 0.15
0.4
DOP (m)

DOP (m)
0.3 0.10

0.2
0.05
0.1

0.0 0.00
60 70 80 90 100 110 120 130 140 200 300 400 500 600 700 800 900
fc (MPa) V0 (m/s)

(c) 1.0 O'Neil et al. (1999)

(d) 0.8
0.9 0.7 O'Neil et al. (1999)
Proposed parameters
Proposed parameters
Original parameters
0.8 0.6 Original parameters

0.7 0.5
DOP (m)

DOP (m)

0.6 0.4
0.5 0.3
0.4 0.2
0.3 0.1
0.2 0.0
500 550 600 650 700 750 800 850 200 300 400 500 600 700 800
V0 (m/s) V0 (m/s)

(e) 1000
900 Hanchak et al. (1992) (140MPa)
Proposed parameters
800 Original parameters
700
600
Vr (m/s)

500 120
100
400
Vr (m/s)

80
300 60
40
200 20
100 0
374 376 378 380 382 384
V0 (m/s)
0
200 400 600 800 1000 1200
V0 (m/s)

Fig. 7.16 Comparisons of test data and predicted results: a Test 1, b Test 2, c Test 3, d Test 4, and
e Test 5

rffiffiffiffiffiffiffiffiffiffiffiffiffi
2 0p 0p
e_ ¼ e_ e_ ð7:20dÞ
3 ij ij
0
where e_ ij is the strain rate tensor, e_ pij is the plastic strain rate tensor, e_ ij is the
0
deviatoric strain rate tensor, and e_ ijp is the deviatoric plastic strain rate tensor.
Therefore, a question whether effective strain rate equals the axial strain rate
276 7 Numerical Simulation of Projectile Impact on Concrete Targets

Fig. 7.17 Test results of strain rate effect Bischoff and Perry (1991)

appears when taking use of the above-mentioned uniaxial test data in numerical
simulations.
This section aims to present the relation between effective strain rate and axial
strain rate theoretically for the above-mentioned four different effective strain rate
computations based on a bilinear elasto-plastic model and analyze the discrepancy
between the numerical predictions based on the different definitions of effective
strain rate and the test data. Furthermore, a corrective computation of effective
strain rate in K&C model is carried out and the inertia effect in SHPB test is
discussed.

7.3.1 Different Definitions of the Effective Strain Rate

The strain rate is a second-order tensor that could be written as

e_ ¼ ½e_ 11 ; e_ 22 ; e_ 33 ; e_ 12 ; e_ 23 ; e_ 31 . The strain rate can be decomposed into the sum of
elastic part and plastic part

e_ ¼ e_ e þ e_ p ð7:21Þ

In order to simplify the theoretical analysis, a bilinear elasto-plastic model

shown in Fig. 7.18 is used, where E is the Young’s modulus and b is the hardening
or softening parameter. The loading rule and unloading rule are given by
7.3 Computation of the Effective Strain Rate 277

Fig. 7.18 Bilinear

σ Linear hardening
elasto-plastic model
βΕ
σ0
Ideal elasto-plastic
βΕ
E Linear softening
E
εp εe
o ε0 ε

r11 ¼ r0 þ bE ðe11  e0 Þ; Loading ð7:22aÞ

r11 ¼ Eee11 ; Unloading ð7:22bÞ

Substituting Eqs. (7.22a, b) into Eq. (7.21), the relation between axial strain rate
and axial plastic strain rate could be obtained:

e_ p11 ¼ e_ 11  e_ e11 ¼ ð1  bÞ_e11 ð7:23Þ

In uniaxial dynamic tests, shear strain rates are negligible, and the lateral strain
rates could be assumed to be equal and as a ratio of axial strain rate, i.e.,

e_ 22 ¼ e_ 33 ¼ a_e11 ; e_ 12 ¼ e_ 23 ¼ e_ 31 ¼ 0 ð7:24aÞ

e_ p22 ¼ e_ p33 ¼ ap e_ p11 ; e_ p12 ¼ e_ p23 ¼ e_ p31 ¼ 0 ð7:24bÞ

where a is the ratio of lateral strain rate to axial strain rate and ap is the ratio of
lateral plastic strain rate to axial plastic strain rate.
The lateral elastic strains are induced by the Poisson’s effect, while the lateral
plastic strains are induced by plastic flow

e_ e22 ¼ e_ e33 ¼ m_ee11 ; e_ p22 ¼ e_ p33 ¼ ap e_ p11 ð7:25Þ

where m is the Poisson’s ratio. Substituting Eqs. (7.24a, b) and (7.25) into
Eq. (7.21),
 
e_ 22 ¼ e_ e22 þ e_ p22 ¼ m_ee11 þ ap e_ p11 ; e_ 22 ¼ a_e11 ¼ a e_ e11 þ e_ p11 ð7:26aÞ

Thus,

m_ee11 þ ap e_ p11
a¼ ð7:26bÞ
e_ e11 þ e_ p11
278 7 Numerical Simulation of Projectile Impact on Concrete Targets

Substituting Eq. (7.23) into Eq. (7.26b), the relation between a and ap could be
obtained:

a ¼ ð1  bÞap  mb ð7:27Þ

The plastic potential function of fractionally associated flow rule can be written
as (Malvar et al. 1997)
pffiffiffiffiffiffiffi
g¼ 3J2  xf ð7:28Þ

According to the definition of plastic flow (Malvar and Simons 1996), we have
0
!
@g 3rij xdij df
depij ¼ du ¼ pffiffiffiffiffiffiffi þ du ð7:29Þ
@rij 2 3J2 3 dP

where du is a proportionality constant. Substituting Eqs. (7.24a, b) into Eq. (7.29)



x df 1 x df
dep11 ¼ 1þ du; dep22 ¼  þ du ð7:30Þ
3 dP 2 3 dP

Hence, the ratio ap could be solved as

dep22 3 þ 2x dP
df
ap ¼ p ¼ ð7:31Þ
de11 6 þ 2x dP
df

The yield surfaces of metal-like materials are basically hydrostatic pressure

independent, which means df =dP ¼ 0, and thus, ap ¼ 1=2. While for the con-
crete materials, the yield surfaces are highly dependent on hydrostatic pressure. The
linear Drucker–Prager model is adopted in the following theoretical analysis, so that
the yield surface f ¼ nP and df =dP ¼ n. Substituting into Eq. (7.31),

3  2xn
ap ¼ ð7:32Þ
6  2xn

Based on the effective strain rate definitions (Eqs. 7.20a–d), some reasonable
assumptions (Eqs. 7.24a, b), and derivations (Eqs. 7.27 and 7.32), the four ratios w
between effective strain rate and axial strain rate are obtained as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
u
u2

 2 !
e_ 3  2xn
w1 ¼ ¼ t 1 þ 2 ð 1  bÞ  mb ð7:33aÞ
e_ 11 3 6  2xn
7.3 Computation of the Effective Strain Rate 279
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u
u2
2 !
e_ 3  2xn
w2 ¼ ¼ t 1þ2 ð 1  bÞ ð7:33bÞ
e_ 11 3 6  2xn



 
e_ 2 3  2xn
w3 ¼ ¼ 1  ð 1  bÞ  mb ð7:33cÞ
e_ 11 3 6  2xn



e_ 2 3  2xn
w4 ¼ ¼ 1 ð 1  bÞ ð7:33dÞ
e_ 11 3 6  2xn

7.3.2 Influence of Material Parameters

It can be found that all of these ratios w are related to the hardening or softening
parameter b, the flow rule parameter x, and the yield surface parameter n.
Moreover, as shown in Eqs. (7.33a–d), the flow rule parameter x and the yield
surface parameter n are always coupled. Therefore, the influences of parameter b
and xn on the ratios w are discussed in this section.
For metal-like materials, the yield surfaces are hydrostatic pressure independent,
that is, n ¼ 0. x ¼ 0 if the non-associative flow rule is used, which means that no
dilatancy would arise. In these two cases, xn ¼ 0. Substituting xn ¼ 0 into
Eqs. (7.33a–d), the influence of the hardening or softening parameter on the ratios
w could be obtained, as shown in Fig. 7.19. It is indicated that the ratios w are not
equal to unit in most cases and with b increasing from negative to positive, the
ratios decrease gradually. For hardening situations, the effective strain rate would be
lower than the axial strain rate, while softening situations are reversed. Only if the
ideal elasto-plastic model is applied, all of these four ratios w = 1. Additionally, the
definition of effective strain rate in Eqs. (7.20b, d) would lead to a very large

Fig. 7.19 Effect of b on w 2.0

when xn = 0 Eq. (7.20a)
Eq. (7.20b) and Eq. (7.20d)
Eq. (7.20c)
1.5

1.0
ψ

0.5

0.0
-1.0 -0.5 0.0 0.5 1.0
β
280 7 Numerical Simulation of Projectile Impact on Concrete Targets

(a) 4.0 (b) 2.4

Eq. 7.20(a)
Eq. (7.20a) and Eq. (7.20b)
3.5 Eq. 7.20(b)
Eq. (7.20c) and Eq. (7.20d) 2.0 Eq. 7.20(c)
Eq. 7.20(d)
3.0
1.6
2.5

ψ
ψ

1.2
2.0 1.0
0.8
1.5

1.0 0.4
0.0 0.5 1.0 1.5 2.0 0.0 0.5 1.0 1.5 2.0
ωξ ωξ

(c) 7
Eq. (7.20a)
6 Eq. (7.20b)
Eq. (7.20c)
5 Eq. (7.20d)

4
ψ

1
0.0 0.5 1.0 1.5 2.0
ωξ

Fig. 7.20 Effect of xn on w: a b = 0, b b = 0.5, and c b = –0.5

deviation, while the definition of effective strain rate in Eqs. (7.20a, c) shows rather
good agreement with axial strain rate. Therefore, without any correction to the
computation of effective strain rate in numerical simulation, it would be more
suitable for pressure-independent materials or non-dilatancy materials to adopt the
definition of effective strain rate as shown in Eqs. (7.20a, c) when hardening or
softening effects considered.
However, for concrete or geological materials, the yield surfaces are highly
dependent on hydrostatic pressure, and dilatancy effect should be taken into con-
sideration. Hence, xn 6¼ 0. Moreover, the hardening and softening effects should
also be included.
The qualitative analyses based on an ideal elasto-plastic model, a linear hard-
ening model with b ¼ 0:5, and a linear softening model with b ¼ 0:5 are carried
out, respectively, and illustrated in Fig. 7.20. Obviously, with xn increasing, all of
these four ratios w increase rapidly. Moreover, it is noticed that the effective strain
rate defined in Eq. (7.20a) is always higher than that in Eq. (7.20c). Similarly, the
effective strain rate defined in Eq. (7.20b) is higher than that in Eq. (7.20d).
7.3 Computation of the Effective Strain Rate 281

However, none of these four effective strain rate representations can be in good
agreement with axial strain rate, and the effective strain rate is larger than the axial
strain rate in most cases. Therefore, the overestimation of the dynamic strength will
occur if the test data such as DIF-axial strain rate relations is embedded into
material models without the consideration of the difference between effective strain
rate and axial strain rate.

7.3.3 Verification and Correction

7.3.3.1 Verification by Single-Element Numerical Analysis

For a sophisticated constitutive model, such as K&C model, many parameters have
to be defined to describe the evolution of the hardening and softening, different
strength surfaces, and complex flow rule. In order to verify the effect of the different
definitions of effective strain rate in numerical computations, a special
rate-dependent material model based on the K&C model in LS-DYNA with the
yield surface (maximum strength surface) that defined as Eq. (7.34) is used.

P
f ¼ a0 þ ð7:34Þ
a1 þ a2 P

The differential of the yield surface can be solved as

df a1
¼ ð7:35Þ
dP ða1 þ a2 PÞ2

Since only the maximum strength surface is used to represent the yield surface,
the uniaxial stress-strain curve is similar to the ideal elastic-plastic model that has
no hardening and softening after yielding.
The remaining parameters, including the Poisson’s ratio, the elastic modulus, the
EOS, and damage parameters, are defined as the same as the original K&C model
with automatic input (Malvar et al. 2000). The four definitions of the effective strain
rate presented in Eqs. (7.20a–d) can be easily embedded into available finite ele-
ment analysis programs, and the CEB recommended the DIF  e_ relation is
adopted. Similar to the case shown in Fig. 7.16a, the results predicted by Eq. 7.20a
are close to those by Eq. 7.20b, and the results predicted by Eq. 7.20c are close to
those by Eq. 7.20d. Additionally, the effective strain rate adopted in K&C model is
solved from the deviatoric strain rate components shown in Eq. (7.36) and similar
to Eq. 7.20c.
282 7 Numerical Simulation of Projectile Impact on Concrete Targets

Fig. 7.21 Single-element Displacement load

model

m
1m
1mm

1mm
Symmetry planes

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi


2 02 02 02 1 2 
e_ ¼ e_ þ e_ 22 þ e_ 33 þ e_ 12 þ e_ 23 þ e_ 31 2 2 ð7:36Þ
3 11 2

Three symmetric planes are used to model the unconfined boundary condition,
as shown in Fig. 7.21. A single cube element with the size as small as 1 mm  1
mm  1 mm is used in the analysis in order to reduce the inertia effect to the
minimum. The strain rate in the numerical simulation is controlled by displacement
load.
Figure 7.22 shows the dynamic stress-strain curves under the different strain
rates, and Fig. 7.23 gives the DIF  e_ relation based on the different definition of
effective strain rate. It is shown that the flow rule parameter x has a significant
influence on the dynamic strength. Only if x closes to zero, the four different
definitions of effective strain rate come to the same results and the obtained DIFs
show good agreement with the input ones recommended by CEB (CEB-FIB Model
Code 1990 1993). Otherwise, the obtained DIFs would be higher than the input
ones and the deviation becomes much more significant with increase of x.

Fig. 7.22 Dynamic -200

stress-strain curves Eq. 7.20(a, b) Eq. 7.20(c, d)
ω = 0.01
-160 ω = 0.5 200 /s
ω = 0.9
Stress (MPa)

-120 100 /s

-80
1 /s 10 /s

-40

0
0.00 -0.01 -0.02 -0.03 -0.04 -0.05 -0.06
Strain
7.3 Computation of the Effective Strain Rate 283

Fig. 7.23 DIF  e_ relation 3.5

Eq. 7.20(a, b) Eq. 7.20(c, d)

3.0
= 0.01
= 0.5
2.5 = 0.9

DIF
CEB recommendation

2.0

1.5

1.0
1 10 100
Strain rate (1/s)

Additionally, there is also some distinction between the different definitions of

effective strain rate. In this case, the definition in Eq. 7.20a, b would have more
significant influence on the simulation results than that in Eq. 7.20c, d.
Figure 7.24 shows the relation between the lateral stress and axial strain in the
case of the effective strain rate definition in Eq. 7.20a and x ¼ 0:5. It is obvious
that the lateral confinement is not in an order of magnitude with axial stress. In fact,
the element size is so small that the inertia effect could be ignored, and the increase
of dynamic strength is only induced by the difference of effective strain rate and
axial strain rate. Therefore, in the previous studies of the inertia effect (Li and Meng
2003; Lu and Li 2011a; Hao et al. 2012; Fang et al. 2014c; Magallanes et al. 2010),
it is unreasonable to wholly attribute the phenomenon that the output DIFs are
higher than the input ones to the inertia effect. In order to obtain the real strain rate

Fig. 7.24 Relation between 0.4

lateral stress and axial strain
(Eq. 7.20a and x = 0.5)
0.3
Lateral stress (MPa)

0.2

0.1

0.0

0.00 -0.01 -0.02 -0.03 -0.04 -0.05 -0.06

Axial Strain
284 7 Numerical Simulation of Projectile Impact on Concrete Targets

effect, a corrective computation of effective strain rate is firstly proposed, and then,
the inertia effect could be evaluated precisely.

7.3.3.2 Corrective Methodology of Computation

of the Effective Strain Rate

It is proved that when the peak stress is reached, the elastic strain increment could
be negligible (Lu and Li 2011a), which means no hardening or softening happens at
this moment and l ¼ 0. Based on the relation between effective strain rate and axial
strain rate, the computation of effective strain rate in K&C model must be corrected
to ensure effective strain rate equal to axial strain rate in uniaxial dynamic tests,
which could be written as


x df
e_ new ¼ 1þ e_ ð7:37Þ
3 dP

where e_ new is the corrected effective strain rate and e_ is the original effective strain
rate obtained from Eq. (7.36). df =dP could be solved from Eq. (7.35), and the ratio
e_ new =_e is shown in Fig. 7.25.
Since the frictional effect between the pressure bars and the specimen in SHPB
system is not taken into consideration, the simplified model of SHPB tests without
the pressure bars is presented in Fig. 7.26 in which the stress response is similar to
the complete SHPB test system (Hao et al. 2012). In this numerical study, a
60-mm-diameter and 40-mm-height specimen applied with a displacement load
shown in Fig. 7.26 is employed.
Both the original and corrective computations of the effective strain rate are
implemented in K&C model and the strain rate effect recommended by CEB is
adopted. A comparison of output DIFs with CEB recommendations is shown in
Fig. 7.27. It is obvious that the output DIFs are always larger than the input ones,

Fig. 7.25 Variation of e_ new =_e

with P 1.0

0.8

0.6
new /

0.4 =0
= 0.3
= 0.5
0.2
= 0.8
=1
0.0
0 20 40 60 80 100
P (MPa)
7.3 Computation of the Effective Strain Rate 285

Fig. 7.26 Simplified model

Displacement load
of specimen in SHPB

30mm

40mm
Specimen

and the corrective computation of effective strain rate would be always lower than
the original computation. It should be noticed that at low strain rates such as
e_ ¼ 30=s, the inertia effect is insignificant and the output DIFs should approach to
the input ones. However, it indicated that the original computation of effective
strain rate still leads to higher output DIFs at the low strain rate, while the corrective
computation of effective strain rate is kept in good agreement with the input DIFs.
The reason lies in that the original computation of effective strain rate would make
the effective strain rate much larger than the axial strain rate and the differences
between the original and corrective computation of effective strain rate enlarge with
the increasing of strain rate.
However, the output DIFs after the correction are still larger than the input DIFs
as shown in Fig. 7.27, which is attributed to the inertial effect. Therefore, an iter-
ative corrective methodology should be proposed to eliminate the inertia effect. The
schematic diagram of the corrective computational methodology is shown in
Fig. 7.28, and the main steps are summarized as follows: Firstly, the computation of
effective strain rate should be corrected as mentioned above. Then, set the input

Fig. 7.27 Comparison of 5

input and output DIFs Input DIFs
- CEB recommendation
Ouput DIFs
4 - original effective strain-rate computation
Ouput DIFs
- corrected effective strain-rate computation
DIF

1
1 10 100
Strain rate (1/s)
286 7 Numerical Simulation of Projectile Impact on Concrete Targets

Correction of effective strain-rate

i 1
DIFinput 1
i i 1 i 1
DIFinput DIF0 DIFN DIF input where i 2,3... is the iteration step

i
DIFinput +numerical SHPB test

DIF0 DIFNi No
is
DIF0

YES
i
DIFinput is the real strain-rate

Fig. 7.28 Corrective computational procedure

DIFs (noted by DIF1input Þ to be unity so that the obtained DIFs (noted by DIF1N Þ in
numerical simulation are only attributed to the inertia effect. Generally speaking,
the DIF1N is lower than the test data DIF0 . Thirdly, the difference between DIF0 and
DIF1N is set as the input DIFs (noted by DIF2input Þ, and the output DIFs (noted by
DIF2N Þ are obtained. The difference between DIF0 and DIF2N is also used as the
deviation of the input DIFs. The ith input DIFs ðDIFiinput Þ may be considered as the
real strain rate effect when this difference is small enough.
Based on the corrective methodology described above, the correction of the
inertia effect is made to both the original and the corrected effective strain rate, and
the derived DIF  e_ relations are illustrated in Fig. 7.29. Since the computation of
effective strain rate has a great contribution to the DIF enhancement, the derived
DIF  e_ relation based on the corrected effective strain rate is much higher than the
original one. In other words, the real inertia effect would not be as large as
described in previous studies (Li and Meng 2003; Lu and Li 2011a; Hao et al. 2012;
Fang et al. 2014c; Magallanes et al. 2010).

7.4 Modified HJC Model

In Sect. 7.2.2.1, we have pointed out five disadvantages on HJC model. The
parameter calibration of the yield surface and EOS, as well as strain rate effect, is
discussed in previous sections. In this section, we focus on the numerical
7.4 Modified HJC Model 287

Fig. 7.29 Derived real DIF-_e 3.5

relations
CEB recommendation
3.0 With correction of effective strain-rate
Without correction of effective strain-rate

2.5

DIF
2.0

1.5

1.0
1 10 100
Strain rate

simulations of cratering and scabbing phenomena induced by projectile penetration

and perforation, and a modified HJC model is presented to reach this goal.
When concrete slabs are subjected to projectile impact, cratering and scabbing
are frequently observed on the impact and distal surfaces, respectively (Hanchak
et al. 1992; Wu et al. 2015a). Cratering on the impact surface is a complex phe-
nomenon involving the compressive failure of material, propagation of the com-
pressive stress wave, and the reflected tensile stress wave on the free surface of the
impact side (Wang et al. 2007b). The mechanism for the scabbing on the distal
surface is straightforward, which is caused by the tensile stress wave induced by the
reflection of the compressive wave on the distal free surface. Numerical analyses of
cratering and scabbing in concrete slabs subjected to projectile impact have been
studied intensively by various concrete material models. Previous studies (Wang
et al. 2007b; Huang et al. 2005; Liu et al. 2009; Leppänen 2006; Tu and Lu 2010)
show that the tensile strength, strain softening, fracture energy, and strain rate effect
on the tensile strength in the material model are crucial for the correct modeling of
cratering and scabbing phenomena in concrete targets.
As discussed in Sect. 7.2.2, HJC model is a simple and effective model for
penetration studies. In fact, HJC model is valid mainly under the stress states
dominated by compression. However, for the stress states dominated by tension, it
uses the ideal elastic-plastic model, which is independent of the facture energy. The
compression and tension refer to the stress states under positive and negative
hydrostatic pressures, respectively. Therefore, HJC model is not suitable for the
numerical simulation of tension-dominated phenomena, e.g., cratering and scab-
bing. Further modifications of HJC model mainly focus on the compressive
288 7 Numerical Simulation of Projectile Impact on Concrete Targets

behavior (Polanco-Loria et al. 2008) and strain rate effect (Polanco-Loria et al.
2008; Islam et al. 2012), but still not suitable for the prediction of the cratering and
scabbing. Liu et al. (2009) presented a modified version of HJC model, where they
employ TCK model (Taylor et al. 1986) to describe the dynamic tensile behaviors,
while the compressive behaviors were still described by the original expression in
HJC model. Based on this model, the authors (Liu et al. 2009) show reasonable
predictions of cratering and scabbing phenomenon in concrete slabs. However, the
yield surface of their concrete material model is discontinuous due to the separate
treatment of the compressive behavior and tensile behavior. Besides, TCK model
has nine differential equations, which must be carefully integrated in a time step
during the implementation and may lead to numerical instability. In addition, some
parameters, such as the volumetric strain rate at fracture, are difficult to determine
experimentally.
Modifications of the dynamic tensile behavior of RHT model (AUTODYN
2003) are carried out by Leppänen (2006) and Tu and Lu (2010), respectively, in
order to obtain improved simulation of the cratering and scabbing phenomena in
concrete targets, where bilinear tensile softening model based on the fracture energy
and new dynamic increase factor for tension are used. Both Leppänen (2006) and
Tu and Lu (2010) assume that the fracture energy is a constant and the fracture
strain decreases with the increase of the strain rate in tension. However, recent
experiments on tensile strength and fracture energy from instrumented spalling tests
(Weerheijm and Van Doormaal 2007; Schuler et al. 2006) suggest that the fracture
energy increases with the increase of the strain rate. In addition, Weerheijm and
Van Doormaal (2007) further suggest that at high strain rate, the fracture strain is a
constant and the fracture energy increases with the increase of the strain rate. In
addition, RHT model is not as simple as HJC model for the practical use.
This section aims to modify HJC model for the numerical simulation of concrete
slabs subjected to impact loading, especially for the improved simulations of the
cratering and scabbing phenomena in concrete targets. Modifications of HJC model,
including the modified yield surface, and introduction of the tensile damage, Lode
angle dependency and strain rate effect, are presented. The improved performances
of the model due to the above modifications are demonstrated by the numerical
simulations of single finite element and projectile impacting concrete slabs. In
addition, the detailed discussions and parametric studies are presented (Kong et al.
2016a).

7.4.1 Yield Surface

In order to avoid the discontinuous description of the yield surface and take the
tensile damage into account at the same time, a simple continuous yield surface is
proposed as follows:
7.4 Modified HJC Model 289

8 0
>
>
> h3½p þ T ð1  Dt Þr ; P\0 i
> 3T ð1  D Þ  9PT ð1Dt Þ þ 3P r 0 ; 0  P  f ð1  D Þ=3
>
>
pffiffiffiffiffiffiffi < h 
t
 fc ð1Dc Þ
c c
i
req ¼ 3J2 ¼ 3P 1  3T 0
>
> fc þ 3ð1  Dc ÞT r ; fc ð1  Dc Þ=3  P  fc =3
>   
>
> N
>
: fc  3TDc þ Bfc fPc  13 r 0  Smax ; P [ fc =3

ð7:38Þ

where Dc is the accumulated compressive damage, which is still determined by the

damage formula of the original HJC model, i.e., Eqs. (7.7 and 7.8). r 0 is the ratio of
the current meridian to the compressive meridian and will be defined in Sect. 7.4.3.
Dt is the cumulative tensile damage, which will be presented in Sect. 7.4.2. Other
constants have the same meaning of the original HJC model introduced in
Sect. 7.2.2.1. A typical view of the yield surface in compressive meridian is shown
in Fig. 7.30.
As observed in Fig. 7.30 and Eq. (7.38), the yield surface is piecewise-defined.
For the first segment (P < 0), a linear dependence between the yield surface and the
pressure is assumed, where a pressure cutoff (− T(1 − Dt)) is introduced. The
cohesive strength is assumed to be 3T (Malvar et al. 1997) and to be degraded with
the cumulative tensile damage. For the second segment ð0  p  fc ð1  Dc Þ=3Þ,
linear dependence between the yield surface and the pressure is also assumed,
where the equivalent stress is equal to fc(1 − Dc) when the pressure is equal to fc(1 −
Dc)/3. The unconfined uniaxial compressive strength is assumed to be degraded
with the cumulative compressive damage. It should be pointed out that
fc ð1  Dc Þ=3 must be no lower than 3T(1 − Dt) to ensure that the equivalent stress
increases with the increase of the pressure. The third segment of the yield surface is
assumed to be linearly parallel with the corresponding maximum strength surface
(undamaged), as shown in Fig. 7.30, where ðfc  3TDt Þ  fc ð1  Dc Þ should be

Dc =Dt =0 (undamaged)

3J 2
fc
0<Dc <1, 0<Dt <1, (partial damaged)

3Τ
parallel

fc-3TDt

fc(1-Dc)
3Τ(1-Dt) Dc =Dt =1(totally damaged)

-T -T(1-Dt) fc(1-Dc)/3 fc/3 P

Fig. 7.30 The yield surface in compressive meridian

290 7 Numerical Simulation of Projectile Impact on Concrete Targets

satisfied. The forth segment of the yield surface ðp [ fc =3Þ is the same as the
original HJC model, as shown in Eq. (7.38). It is observed that the distinctive
feature of the present model is the separate treatment of the compressive damage
and tensile damage, in which the compressive damage and tensile damage are
accumulated under positive and negative pressures, respectively. Similar treatment
is also adopted by Liu et al. (2009). However, unlike the report presented by Liu
et al. (2009), the continuity of the present yield surface is ensured due to the new
definition shown in Eq. (7.38).

7.4.2 Tensile Damage

Extensive experimental studies on the tensile softening behavior have been per-
formed. Weerheijm and Van Doormaal (2007) gave the definition of the softening
curve as a power function (“Hordijk–Reinhard expression”). Consequently, the
tensile damage is proposed as follows:

 !

ep 3 ep ep  
Dt ¼ 1 þ c1 exp c2  1 þ c31 expðc2 Þ ð7:39Þ
efrac efrac efrac
P
where ep ¼ Dep is the effective plastic strain. efrac is the fracture strain, which is
related to the element size. c1 and c2 are constants and can be taken as c1 = 3 and
c2 = 6.93.
A typical stress-strain curve for uniaxial tension is shown in Fig. 7.31. For a
given fracture energy Gf, the following specific strain energy in the entire cracking
process should hold (Tu and Lu 2009):

“Hordijk-Reinhardt’’ function

Gf/hc
ε el
ε p =0 ε p =ε frac ε

Fig. 7.31 Typical stress-strain curve for uniaxial tension

7.4 Modified HJC Model 291

Zefrac
rde ¼ Gf =hc ð7:40Þ
eel

where eel is the peak strain. hc is the characteristic length of the element, which may
be approximated by the cube root of the volume of the element in a 3D analysis (Tu
and Lu 2009). Using Eq. (7.39), Eq. (7.40) can be replaced by

Zefrac Zefrac Zefrac

rde  rdep ¼ T ð1  Dt Þdep ¼ 0:195Tefrac ð7:41Þ
eel 0 0

Using Eqs. (7.39 and 7.40), the fracture strain is obtained by,

5:136Gf
efrac ¼ ð7:42Þ
hc T

Recent study by Xu (2013) adopts the similar tensile damage as Eq. (7.39),
where he uses the maximum principal strain instead of the effective plastic strain ep
to calculate the tensile damage, which may cause some problems. For example, the
maximum principal strain will be greater than zero after an elastic uniaxial com-
pression test. Consequently, the tensile damage is greater than zero when the ele-
ment starts to be subjected to tensile loads, which is not in accord with the practical
situation. In addition, Xu (2013) assumed that the fracture strain is independent of
element size and has a constant value of 0.007, which cannot ensure the amount of
the fracture energy in a finite element. However, the present tensile damage model
not only follows the general trend of strain softening suggested by Weerheijm and
Van Doormaal (2007), but also ensures the amount of the fracture energy in a finite
element.
It should be noted that there is no accumulated plastic strain when the stress path
is close to the triaxial tensile path, where stress deviators approach zero.
Consequently, the tensile damage Dt remains zero, which cannot represent the
actual situation. To overcome this problem, a volumetric damage scalar related to
volumetric strain is introduced as follows:

DDt ¼ b  fd  Dev ð7:43Þ

where b is a constant and Dev is the volumetric strain increment during a cycle of
integration. fd is the parameter that limits the volumetric damage accumulation close
to the triaxial tensile path and given by (Malvar et al. 1997)
pffiffiffiffiffiffiffi  pffiffiffiffiffiffiffi 
1   3J2 =P=0:1 0   3J2 =P\0:1
fd ¼ pffiffiffiffiffiffiffi ð7:44Þ
0  3J2 =P  0:1
292 7 Numerical Simulation of Projectile Impact on Concrete Targets

Table 7.3 Input parameters of the modified and original HJC models for 48-MPa concrete
Yield surface Damage EOS Others
A 0.3 c1 3 pcrush 16 MPa q 2440 kg/m3
B 2.0 c2 6.93 µcrush 8.1e−4 fc 48 MPa
N 0.75 b 0.5 K1 17.4 GPa G 14860 MPa
Smax 11 hc 3 mm K2 38.8 GPa T 4 MPa
D1 0.04 K3 29.8 GPa
D2 1 plock 1.05 GPa
EFMIN 0.01 µlock 0.10
Gh 80 Nm/m2
efrac 0.03

In order to gain some insight into the proposed tensile damage accumulation,
uniaxial, biaxial, and triaxial tensile tests for 48-MPa concrete are numerically
conducted, respectively. Referring to (CEB-FIB Model Code 1990 1993), the
fracture energy Gf is taken as 80 Nm/m2. The side length of the cubic element is
10 mm, so the characteristic length hc is also 10 mm. Using Eq. (7.42), efrac is
obtained as 0.011. The other constants are summarized in Table 7.3. The predicted
stress-strain curves are shown in Fig. 7.32, where the numerical simulations by the
original HJC model are also presented for the comparison purpose.
It is observed that for the present model, the strain softening is well described
and is consistent with the suggestions by Weerheijm and Van Doormaal (2007),
which is obtained from a large amount of static tensile tests. While for original HJC
model, the predicted stress-strain curve is perfect elasto-plastic, which is indepen-
dent of fracture energy. The influence of the volumetric damage parameter b on the
stress-strain curve for the triaxial tension is also of interest and shown in Fig. 7.32c,
where the stain softening increases with the increase of b.

7.4.3 Lode Angle Dependency

In order to capture the transition of the deviatoric section from a triangular shape at
low pressures to a circular shape at high pressures, the dependence of yield surface
on the Lode angle should be introduced.
As proposed by Willam and Warnke (1975), the ratio of the current meridian to
the compressive meridian r 0 is defined as
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
0 2ð1  u2 Þ cos h þ ð2u  1Þ 4ð1  u2 Þ cos2 h þ 5u2  4u
r ðh; uÞ ¼ ð7:45Þ
4ð1  u2 Þ cos2 h þ ð1  2uÞ2

where h denotes the Lode angle, which can be determined by

7.4 Modified HJC Model 293

Fig. 7.32 Numerical (a)

simulation results of the 12
stress-strain curves by the Modified model
present model and HJC HJC model

Tensile stress (MPa)

model: a uniaxial tension, 9
b biaxial tension, and
c triaxial tension
6

0
0.000 0.002 0.004 0.006 0.008 0.010 0.012
Tensile strain

(b)
6

5 Modified model
Tensile stress (MPa)

HJC model
4

0
0.000 0.001 0.002 0.003 0.004 0.005 0.006
Tensile strain

(c)
4
Modified model, b=0.1
Modified model, b=0.2
Tensile stress (MPa)

3 Modified model, b=0.3

HJC model

0
0.000 0.003 0.006 0.009 0.012 0.015
Tensile strain
294 7 Numerical Simulation of Projectile Impact on Concrete Targets

pffiffiffi  
3 3 sij 
cos 3h ¼ ð7:46Þ
2 J 3=2
2

in which sij is the deviatoric stress tensor. u is the ratio of the tensile meridian to the
compressive meridian. As discussed by Tu and Lu (2009), u proposed by Malvar
et al. (1997) covers a number of important factors, which is summarized as follows:
8
>
> 0:5 P0
> 0:5 þ 1:5T=f
> P ¼ f3c
<  c

uðPÞ ¼ 1:15= 1 þ Bð1:3=3ÞN P ¼ 2:3f c
ð7:47Þ
>
>
3
>
> 0:753 P ¼ 3fc
:
1 P  8:45fc

Equation (7.47) is adopted here. For computational purpose, uðPÞ is determined

by a linear interpolation of the discrete points described in Eq. (7.47).
The numerical predicted stress-strain curves of 48 MPa concrete under uniaxial
compression (UC, compression meridian, h = 60°) and biaxial compression (BC,
tensile meridian, h = 0°) by the modified HJC model and the original HJC model
are shown in Fig. 7.33, respectively. It is clearly observed that for uniaxial com-
pression, the original HJC model overestimates the peak stress in the case of
uniaxialcompression. The uniaxial compression test corresponds to the point

p ; req ¼ ð1=3; 1Þ in the yield surface; thus, A þ B=3N ¼ 1 should be satisfied
for the original HJC model. However, this relationship is always dissatisfied since
A, B, and N are usually obtained by fitting the triaxial compression test data. For the
values of A, B, and N used in this book as well as in the original paper (Holmquist
et al. 1993), ðA þ B=3N Þ [ 1. Consequently, uniaxial compressive strength is
overestimated. However, for the modified model, this relationship is satisfied due to
the piecewise-defined yield surface shown in Eq. (7.38); thus, the uniaxial

Fig. 7.33 Stress-strain 320

curves under uniaxial and Proposed model, UC
280 Proposed model, BC
biaxial compression predicted
Compressive stress (MPa)

by the modified model and HJC model, UC

240
original HJC model HJC model, BC
200

160

120

80

40

0
0.000 0.005 0.010 0.015 0.020
Compressive strain
7.4 Modified HJC Model 295

compressive strength is reproduced. For biaxial compression, an overestimation of

about three times of the uniaxial compressive strength is obtained by the original
HJC model where r 0 is equal to unity, while an increase of only 11.5% of the
uniaxial compressive strength is found for the modified model where r 0 is deter-
mined by Eq. (7.47), which is in accordance with the test data by Kupfer et al.
(1969).

7.4.4 Strain Rate Effect

In the original HJC model, strain rate effect is considered by the so-called normal
enhancement that is achieved by multiplying static yield surface by DIF, which
means the yield surface is enhanced normal to the pressure axis. As discussed by
Hartmann et al. (2010), from a continuum mechanics point of view, this is a
simplified approach. Thus, the “radial enhancement” approach instead is more
reasonable, in which DIF is also applied to the corresponding pressure.
According to the “radial enhancement” approach, the yield surface and com-
pressive damage Dc are modified as follows (Malvar et al. 1997):

req ¼ rf req ðp=rf Þ; Dc ¼ Dc ðp=rf Þ=rf ð7:48Þ

where rf is DIF, which is defined separately for compression and tension, since
experimental results show that DIF for tension is much higher than compression.
The relationship between DIF and strain rate used in the previous studies
(Leppänen 2006; Tu and Lu 2010; Islam et al. 2012; Hartmann et al. 2010) is the
CEB-FIB-recommended one (CEB-FIB Model Code 1990 1993) or obtained by
fitting to the experimental data directly, such as the split Hopkinson pressure bar
tests. Three issues should be addressed carefully. The first one is the computation of
effective strain rate discussed in Sect. 7.3, in which the true strain rate is overes-
timated when xn 6¼ 0. For the J2 flow rule adopted here, x ¼ 0; thus, this issue
does not exist here. The second one is that the unavoidable inertia effect induced by
the radial confinement makes an additional contribution to the dynamic compres-
sion enhancement when the strain rate is greater than a critical transition strain rate
which is between 101 and 102 s−1 for concrete (Li and Meng 2003; Zhang et al.
2009; Li et al. 2009). Consequently, the strain rate enhancement will be overesti-
mated if one uses the DIF, which is obtained by fitting to experimental results
directly as the numerical input. The third one is that there is no cutoff for DIF in the
previous studies. For the impact problems, the magnitude of the strain rate may
reach as high as 106 s−1. Therefore, using suggested formulas in the previous
studies would yield DIF magnitude between 10 and 50, which is not realistic.
Recent study by Xu and Wen (2013) suggests semiempirical equations for DIF
of concrete materials used for numerical input, which show good agreement with
available test data for concrete:
296 7 Numerical Simulation of Projectile Impact on Concrete Targets

   
DIFt ¼ ½tanhððlogðe_ =_e0 Þ  Wx ÞSÞ Fm =Wy  1 þ 1 Wy ð7:49aÞ

DIFc ¼ ðDIFt  1ÞðT=fc Þ þ 1 ð7:49bÞ

where DIFt and DIFc are the dynamic increase factors for tension and compression,
respectively. e_ 0 ¼ 1 s1 is the reference strain rate; Fm = 10, Wx = 1.6, S = 0.8, and
Wy = 5.5 are the fitting constant. This semiempirical formulas, which address the
above two issues well, are used in the present study.
Figure 7.34 presents the numerical results of dynamic tensile tests for 48-MPa
concrete under different strain rate, where element size and fracture energy are the
same as those in Sect. 7.4.2. It can be observed that dynamic tensile strength for the
strain rate value of 1 and 10 s−1 is almost same when the original HJC model is
used, which is contrary to the dynamic tensile experiments (CEB-FIB Model Code
1990 1993; Weerheijm and Van Doormaal 2007). Besides, there is no
strain-softening branch in the stress-strain curves. Furthermore, at high strain rate
value of 100 s−1, the predicted stress-strain curve by the original HJC model is
irregular, which may result from the error of the implementation of the yield surface
for the negative pressure, as discussed in Sect. 7.4.1. For the modified HJC model,

(a) 25 Strain-rate 1s-1 (b) 25

Original HJC model Strain-rate 1s-1
Strain-rate 10s-1 Modified HJC model Strain-rate 10s-1
Tensile stress (MPa)
Tensile stress (MPa)

20 Strain-rate 100s-1 20
Strain-rate 100s-1

15 15

10 10

5 5

0 0
0.000 0.003 0.006 0.009 0.012 0.000 0.003 0.006 0.009 0.012
Tensile strain Tensile strain

(c) 900 10
800 Fracture energy
Fracture energy (Nm/m2 )

DIFt 8
700
600
6
500
DIFt

400
4
300
200 2
100
0 0
-2.0 -1.5 -1.0 -0.5 0.0 0.5 1.0 1.5 2.0 2.5
log

Fig. 7.34 Dynamic tensile stress-strain curves: a HJC model, b present model, and c strain
rate-dependent fracture energy
7.4 Modified HJC Model 297

since the fracture strain is assumed to be a constant and the dynamic tensile strength
increases linearly with the increase of DIFt, the fracture energy increases linearly
with DIFt, as shown in Eq. (7.42), Figs. 7.34b, c. The predicted dynamic tensile
stress-strain curves of the modified model are shown in Fig. 7.34b. It is observed
that the general trend of strain softening for various strain rates well reproduces the
experimental observations by Weerheijm and Van Doormaal (2007).

7.4.5 Numerical Simulation

The proposed modified concrete material model is implemented in finite element

code LS-DYNA (Hallquist 2007) through user-defined material model. To ensure
that the modified model can predict the residual velocity of the projectile, and the
cratering and scabbing phenomenon in concrete slabs, the results from several
experiments with varying projectile mass, impact velocity, and slab’s thickness
must be numerically reproduced. The projectile perforation experiments by
Hanchak et al. (1992) and by Wu et al. (2015a) are chosen to simulate, compare,
and discuss in this section. In order to show the improved performance of the
modified model, the corresponding simulation results of the original HJC model are
presented for the comparison purpose.

7.4.5.1 Hanchak et al.’s Test

In the experiments performed by Hanchak et al. (1992), the projectiles are shot into
cuboid 48-MPa reinforced concrete (RC) targets with the size of 610 mm  610
mm  178 mm. The steel projectiles have an ogive-nose of caliber-radius-head
(CRH) 3.0, length of 144 mm, diameter of 25.4 mm, and mass of 0.5 kg. In order
to save the computational cost, only a quarter of the RC target and projectile are
modeled and shown in Fig. 7.35. The 3D Solid 164 element type in LS-DYNA is
employed for modeling the projectile and concrete target, and beam element for the
reinforcement bars. Local mesh refinement is deployed in the mesh division of the

Fig. 7.35 FE model of the

RC target

Symmetric planes
298 7 Numerical Simulation of Projectile Impact on Concrete Targets

concrete targets to guarantee the requirement of computational precision and save

computational cost. The minimum element size for the target is considered as 3 mm
near the impact location, and the element size is enlarged to 6 mm in the region
beyond five times of the projectile diameter. Since the target width is over twenty
times of the projectile diameter, the influence of boundary is neglected and set as
non-reflecting boundary. The eroding surface-to-surface contact is employed to
define the contact behavior between the projectile and RC target without friction.
The steel projectile is modeled as rigid material. The reinforcement bars are
modeled as von Mises material model (Mat_003 in LS-DYNA), with elastic
modulus 210GPa and yield stress 400 MPa. For the concrete material, the constants
of the yield surface (B, N, and Smax) use the suggested values by Xiong et al.
(2010), which are obtained by fitting to a large amount of triaxial compression
experimental data. Tensile damage constants (c1 and c2) are taken from (Weerheijm
and Van Doormaal 2007), and the compressive damage constants (D1, D2, EFMIN)
are the same as the original HJC model. The parameters of EOS are taken from
Johnson et al. (1998). The other damage constants ðGf ; hc ; efrac Þ are obtained by the
method discussed above. Table 7.3 presents all the parameters of the modified and
original HJC model.
Element erosion technique with a criterion based on the effective plastic strain ep
and fracture strain efrac is adopted for the modified HJC model, i.e., the element will
be deleted immediately if ep or efrac is greater than a critical value. The element
erosion technique for the original HJC model is based on the maximum principal
strain, which is achieved through *MAT_ADD_EROSION in LS-DYNA.
The typical numerical simulation results for the impact velocities of 360 and
606 m/s are shown in Fig. 7.36. It can be clearly observed that the modified model
reproduces the cratering and scabbing of the concrete slab well, while the original HJC
model cannot predict the cratering and scabbing phenomenon, since the dynamic
tensile behavior is perfect elasto-plastic, as discussed in Sects. 7.4.2 and 7.4.4.

v=360m/s
Modified HJC Model Original HJC Model
Cratering diameter

Cratering depth

Scabbing depth

Scabbing diameter

v=606m/s
Cratering diameter

Cratering depth

Scabbing depth

Scabbing diameter

Fig. 7.36 Simulation results for the impact velocity of 360 and 606 m/s
7.4 Modified HJC Model 299

Fig. 7.37 Comparisons of Experiment (Hanchak et al. (1992))

the residual velocities 1000
Modified HJC model
between experimental results Original HJC Model

Residual velocity/m/s
and numerical simulations for 800
48-MPa concrete slab
600

400

200

0
0 200 400 600 800 1000 1200
Initial velocity/m/s

Fig. 7.38 Compressive damage a and tensile damage b contours for the impact velocity of
360 m/s

Both the modified HJC model and the original HJC model reproduce the residual
velocities well, as shown in Fig. 7.37. The reason is that the residual velocities mainly
depend on the compressive behavior of the concrete material model. Figure 7.38
shows the compressive damage and tensile damage contour for the impact velocity of
360 m/s, respectively.

7.4.5.2 Wu et al.’s Test

Recently, we conduct twenty-five shots of reduce-scaled projectile perforation test

on five configurations of monolithic and segmented RC slabs with a rear steel line,
and the details can be found in Sect. 12.2 (Wu et al. 2015a). These experiments give
us a good opportunity to further validate the modified HJC model, since the cra-
tering size, scabbing size, residual velocity, and the high-amplitude accelerations of
the projectile were carefully recorded in the test. The finite element model and
element type for the RC target and projectile are the same as that adopted in
previous section expect for the rear steel line, which is modeled by 3D Solid 164
element type. The steel projectile is also modeled as rigid material, as the projectile
deformation and mass loss can be neglected according to the experiment
300 7 Numerical Simulation of Projectile Impact on Concrete Targets

observations. Both the reinforcement bars and steel line are modeled by von Mises
material model with elastic modulus 210 GPa and yield stress 400 MPa. The
determination of the modified HJC model’s constants, element erosion technique,
and contact behavior between the projectiles and RC targets follow the same way
discussed in previous section, which are not repeated here. In order to avoid
repeated simulations and save the computational time, the Shots 1-1, 2-3, 4-3, and
5-3 are chosen to validate the modified model, where the slab thickness and impact
velocity vary. Predicted results of the original HJC model are not presented in this
section, since it has been demonstrated that it cannot reproduce the cratering and
scabbing phenomena in the previous section.
Figure 7.39 presents the simulations of cratering, scabbing, and residual
velocities and the corresponding experimental results. The left of Fig. 7.39 shows
the simulation results, including cratering diameter, cratering depth, scabbing
diameter, scabbing depth, and residual velocities. The right of Fig. 7.39 presents the
corresponding experimental observations. It is indicated that on the whole, the
predicted cratering size and scabbing size agree well with the corresponding
experimental observation. For the thin concrete slabs with the thickness of 100 and
150 mm, the cratering depth and scabbing depth are underestimated, which may
result from the irregular shape of the cratering and scabbing observed in experi-
ments. The predicted residual velocities also show good agreement with the
experimental data.
Comparisons of the acceleration, velocity, and displacement of the projectile for
Shot 1-1 between the experiment and simulation are shown in Fig. 7.40. The
high-amplitude acceleration during the perforation is recorded in the experiment by
the newly developed small-caliber accelerometer, which is proved accurate and
robust in Wu et al. (2015a). It is found that the numerical prediction agrees with the
recorded acceleration well. The time histories of projectile velocity and displace-
ment also coincide with experiment data well, which are obtained by integrating the
recorded time-acceleration curve.

7.4.6 Parametric Analyses

In the modified HJC model, there are six parameters, i.e., T, Gf, efrac , DIFt, hc, and r 0
governing the tensile dynamic behavior of concrete, which are crucial for the
reasonable modeling of cratering and scabbing phenomena. As shown in
Eq. (7.42), T, Gf, and hc are related to the fracture strain efrac , so the sensitivity
analysis of the three independent parameters, i.e., efrac , DIFt, and r 0 is performed by
varying one parameter by ±50 percentages while holding the others constant. It
should be noted that the purpose of the parameter studies is not to compare the
simulation results with the corresponding experiments, but to investigate the
influence of each individual parameter on the cratering and scabbing phenomenon.
In order to save computational time, the simulation in Sect. 7.4.5.1 with projectile
initial velocity of 606 m/s is chosen as a baseline.
7.4 Modified HJC Model 301

(a) Front surface Rear surface

240mm

40mm

300mm 380mm

60mm
Cratering depth: 55mm Sacbbing depth: 55mm
340mm

Residual velocity 257m/s Residual velocity 272m/s

Shot 1-1: Thickness 300mm, Initial velocity 641m/s

(b)
210mm

41mm

270mm
220mm

55mm
Cratering depth: 50mm Sacbbing depth: 75mm
230mm
Residual velocity 310m/s Residual velocity 323m/s
Shot 2-3: Slab thickness 200mm, Initial velocity 544m/s

(c)
260mm

30mm

200mm 230mm

38mm

Cratering depth: 50mm Sacbbing depth: 50mm

210mm

Residual velocity 506m/s Residual velocity 508m/s

Shot 4-3: Slab thickness 150mm, Initial velocity 643m/s

(d)
110mm

18mm
150mm 200mm

46mm

Cratering depth: 50mm Sacbbing depth: 50mm

185mm
Residual velocity 387m/s Residual velocity 354m/s

Shot 5-3: Slab thickness 100mm, Initial velocity 486m/s

Fig. 7.39 Comparisons of predicted results by the modified HJC model with experiments by Wu
et al. (2015a)
302 7 Numerical Simulation of Projectile Impact on Concrete Targets

(a) 20
Experiment (Wu et al. (2015a))
Modified HJC model
15
Acceleration (10 4g)
10

-5
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0
Time (ms)

(b) -250
Data (Wu et al. (2015a))
-300 0.0
Modified HJC model
-350

Displacement (m)
-0.1
Velocity (m/s)

-400 Velocity

-450
-0.2
-500

-550 -0.3
-600 Displacement
-650 -0.4
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Time (ms)

Fig. 7.40 Comparisons of the acceleration, velocity, as well as displacement of the projectile for
shot 1-1 between the experiment and simulation results

7.4.6.1 Influence of Fracture Strain

The influence of efrac on the cratering and scabbing is shown in Fig. 7.41, where the
simulated scabbing size decreases with the increase of efrac . This is a result of the
increase of the fracture energy with the increase of efrac , as can be seen in
Eq. (7.42). The cratering in the concrete slab is not sensitive to efrac , which implies
that the cratering phenomenon is also dependent on the compressive behavior of the
concrete material.
7.4 Modified HJC Model 303

Fig. 7.41 Effect of efrac on

the cratering and scabbing:
a 0.015, b 0.03, and c 0.045

7.4.6.2 Influence of DIFt

The influence of DIFt on the cratering and scabbing phenomenon is also of interest.
Figure 7.42 presents the simulation results under various DIFt, including recom-
mendations of the original HJC model, modified HJC model (Eq. 7.49a, b), fre-
quently used CEB and CEB with a cutoff value of 10. It should be noted that the
DIFt recommended by the original HJC model is applied on the yield surface
directly through the “normal enhancement” approach, while others are applied on
the yield surface through the “radial enhancement” approach discussed in
Sect. 7.4.4. It can be observed that using the DIFt recommended by the original
HJC model cannot reproduce the cratering and scabbing phenomenon, implying
that the “normal enhancement” approach is inappropriate for the simulation of
cratering and scabbing phenomenon. Using the CEB-recommended DIFt cannot
reproduce the cratering and scabbing phenomenon well, and even a cutoff value of
10 is introduced, which indicates that the CEB-recommended DIFt is not accurate.
It also indicates that it is crucial to accurately model DIFt in order to correctly
304 7 Numerical Simulation of Projectile Impact on Concrete Targets

Fig. 7.42 Effect of DIFt on the cratering and scabbing phenomenon: a HJC-recommended DIFt,
b present DIFt (baseline), c CEB-recommended DIFt, and d CEB-recommended DIFt with cutoff

simulate the cratering and scabbing phenomenon, especially the description of DIFt
below the cutoff value around 10.

7.4.6.3 Influence of r′

The effect of the ratio of the current meridian to the compressive meridian r′ on the
cratering and scabbing phenomenon is shown in Fig. 7.43. Using r′ = 1 means that
the deviatoric section is assumed to be circular shape, which is adopted in the
original HJC model. It can be found that using r′ = 1 can well describe the scabbing
phenomenon, while the cratering phenomenon cannot be reproduced, which indi-
cates that r′ is crucial for the simulation of cratering phenomenon. The reason is that
the cratering is a complex phenomenon involving the compressive failure of
material, propagation of the compressive stress wave, and the reflected tensile stress
wave on the free surface of the impact side, while the scabbing is caused by the

Fig. 7.43 Effect of r′ on the cratering and scabbing phenomenon: a r′ = 1 and b proposed r′
7.4 Modified HJC Model 305

tensile stress wave induced by the reflection of the compressive wave on the distal
free surface.

7.5 Modified K&C Model

In the previous section, we present several improvements of HJC model, including

modified yield surface, introduction of the tensile damage, introduction of Lode
angle dependency, and modified strain rate effect, for the numerical simulations of
the cratering and scabbing phenomena. Although the predictions have reasonably
good agreements with corresponding experimental results, the modified HJC model
still has the following shortcomings: (i) Strain hardening is not considered and
(ii) strain softening is only dependent on the cutoff of fracture strain (EFMIN) that is
usually difficult to determine.
The Karagozian and Case concrete model (K&C model, material #72REL3 in
LS-DYNA) is originally developed to analyze structural response subjected to blast
loading, which has become one of the popular concrete models for the response of
concrete responses under intensive dynamic loads. However, this model still has
some disadvantages, as discussed in Sect. 7.2.2. This section proposes a modified
K&C model to improve its capability to predict tensile failure in concrete material
(e.g., cratering and scabbing). After a brief description of the original K&C model,
modifications for the strength surface parameters, the dynamic increment factor for
tension, the relationship between yield scale factor and damage function, and the
tensile damage accumulation are presented. The improved performances of the
modified K&C model are demonstrated by single-element numerical tests and the
simulations of concrete slabs subjected to projectile impact.

7.5.1 Strength Surface Parameters

In the original K&C model, the strength surface parameters are obtained from the
triaxial compression data of 45.4-MPa concrete and expressed as functions of
concrete compressive strength (Malvar et al. 2000), i.e.,

a0 ¼ 0:2956fc ; a1 ¼ 0:4463; a2 ¼ 0:0808=fc ð7:50aÞ

a0y ¼ 0:2232fc ; a1y ¼ 0:625; a2y ¼ 0:2575=fc ð7:50bÞ

a0f ¼ 0; a1f ¼ 0:4417; a2f ¼ 0:1183=fc ð7:50cÞ

In fact, Eqs. (7.50a–c) is suitable for relatively low pressure. However, for the
projectile impact problem concerned here, the pressure at the projectile/target
interface can reach a magnitude of GPa. Consequently, the strength surface
306 7 Numerical Simulation of Projectile Impact on Concrete Targets

Fig. 7.44 Triaxial 18

compression test data 38MPa (Xie et al 1996) Eq. (7.50a)
34MPa (Xiong 2009b) Eq. (7.51a)
15 60MPa (Xiong 2009b)
47MPa (Ansari 1998)
9.8MPa (Yan et al 2007)
12 10.6MPa (Yan et al 2007)
11.4MPa (Yan et al 2007)
25MPa (Wu et al 2007)

/fc
9 27MPa (Hou 2006)
25MPa (Hou 2006)

0
0 2 4 6 8 10 12 14
p/fc

parameters should be redetermined to predict the response of concrete targets

subjected to projectile impact. Figure 7.44 shows some triaxial compression test
data. It can be observed that Eq. (7.50a) has a good agreement with the experi-
mental data when p/fc < 3, beyond which the deviation increases with the increase
of the pressure. A new set of maximum strength surface parameters is proposed by
curve fitting of the data shown in Fig. 7.44 as follows:

a0 ¼ 0:4426fc ; a1 ¼ 0:5698; a2 ¼ 0:02516=fc ð7:51aÞ

By assuming Dry ¼ 0:45Drm (Malvar et al. 1997), the initial yield surface
parameters are obtained as

a0y ¼ 0:2797fc ; a1y ¼ 0:8989; a2y ¼ 0:0685=fc ð7:51bÞ

For the determination of the residual strength surface parameters, triaxial com-
pression experiments are needed, which, however, are difficult to be found in the
relevant literatures. Noting that the residual strength surface should be parallel with
the maximum strength surface under the high-pressure condition, the residual
strength surface parameters are proposed as follows:

a0f ¼ 0; a1f ¼ a1 ; a2f ¼ a2 ð7:51cÞ

Equations (7.51a–7.51c) is used for the modified K&C model in the following
discussions. Another concern is the accuracy of the automatically generated EOS
parameters and discussed below. For the EOS of concrete material, the available
experimental data is limited. The isotropic compression data from Hanchak et al.
(1992) for 48-MPa concrete and the full-scale explosive detonation and
flyer-plate-impact data from Gebbeken et al. (2006) for 51.2-MPa concrete are
plotted in Fig. 7.45. Using the automatically generated parameters of the original
7.5 Modified K&C Model 307

Fig. 7.45 EOS test data and 8

prediction by automatically Hanchak et al. (1992)
generated EOS parameters 7 Gebbeken et al. (2006)
Automatically generated EOS
6

p (GPa)
4

0
0.00 0.05 0.10 0.15 0.20 0.25

K&C model gives a satisfied agreement with the experimental data. Therefore, the
automatically generated EOS parameters are adopted.

7.5.2 DIFt

In Sect. 7.4.6, we show that DIFt has a strong influence on the cratering and
scabbing damage of the concrete slab subjected to projectile impact. Therefore, an
accurate relationship between DIFt and strain rate is essential. Numerical analysis
(Lu and Li 2011b) concludes that DIFt observed in the dynamic tensile tests is a
genuine material effect; consequently, DIFt obtained from the curve-fitting exper-
imental data can be applied to the material model directly.
So far, there are many suggested relationships between DIFt and strain rate in the
literatures, which are plotted in Fig. 7.46 along with experimentally obtained DIFt.
It is found that CEB-FIB-recommended DIFt, which is widely used in previous
studies, greatly underestimates the test data. The empirical formula suggested by Xu
and Wen (2013) shows good agreement with test data, and therefore, it is used, i.e.,
   
DIFt ¼ ½tanhðlogðe_ =_e0 Þ  Wx ÞS Fm =Wy  1 þ 1 Wy ð7:52Þ

However, for the dynamic increase factor for compression (DIFc), as discussed
in Sect. 7.4.4, since the unavoidable inertia-induced radial confinement makes a
contribution to the dynamic compression enhancement when the strain rate is
greater than a critical transition strain rate between 101 and 102 s−1 for concrete (Li
and Meng 2003; Zhang et al. 2009; Li et al. 2009), DIFc obtained from the dynamic
compressive tests includes both the so-called structural effect and genuine strain
rate effect. Consequently, the strain rate enhancement will be overestimated if one
uses the DIFc obtained by fitting to experimental results directly as the
308 7 Numerical Simulation of Projectile Impact on Concrete Targets

10
CEB (1990)
Zhou et al. (2006)
8 Malvar and Crawford (1998)
Xu and Wen (2013)

6
DIFt

0
-9 -7 -5 -3 -1 1 3
10 10 10 10 10 10 10
Strain rate (1/s)

Fig. 7.46 Experimental DIFt and suggested relations (test data reprinted from Schuler et al. 2006)

computational input. Considering the above findings, in the latest version of K&C
model, a cutoff value of 2.94 was introduced for the DIFc, which is used in this
section.

7.5.3 Relationship Between ƞ and k

In the original K&C model, the relationship between η and k is taken as a piecewise
linear curve. If the automatically generated method is not adopted, 13 pairs of
ðki ; gi Þ must be carefully defined, which is inconvenient for practical application.
A new relationship between η and k is proposed as follows:
(
ak=km þ ð3  2aÞðk=km Þ2 þ ða  2Þðk=km Þ3 ; k  km ; strain hardening
gð kÞ ¼ k=km
ac ðk=km 1Þad þ k=km
; k [ km ; strain softening
ð7:53Þ
where a, ac, and ad are the constants governing the strain-hardening and
strain-softening branches, respectively. km is the corresponding value of k when
η = 1 which can be easily obtained by Eqs. (7.14 and 7.53) and the recommended
uniaxial stress-strain relation.
Attard and Setunge (1996) developed an empirical expression for the full
stress-strain relationship of confined and uniaxially loaded concrete, which is
shown to be applicable for a broad range of concrete strengths between 20 and
130 MPa. Comparisons of the stress-strain curves for 50-MPa concrete under dif-
ferent confinement pressure (Pcon) obtained from the modified K&C model
described in this section and the original K&C model with the empirical formula
suggested by Attard and Setunge (1996) are shown in Fig. 7.47. It is observed that
7.5 Modified K&C Model 309

Fig. 7.47 Comparisons of 140

compressive stress-strain =3, c =0.29, d =1.86, m =8.7e-5
curves of 50-MPa concrete 120
developed by Attard and
Setunge (1996) with those 100

Stress(MPa)
obtained by modified and
original K&C models 80

60

40
Empirical formular
20 Original K&C model
Modified K&C model
0
0.000 0.005 0.010 0.015 0.020 0.025 0.030 0.035 0.040
Strain

the modified model shows a good agreement with the empirical formula, while the
original K&C model is stiffer in the loading stage and has larger softening gradient
when compared with the empirical formula. In addition, the original K&C model
overestimates the peak stress, while the modified model has a good agreement with
the empirical peak stress due to the use of new strength surface parameters.

7.5.4 Tensile Damage Accumulation

A typical load-displacement curve for uniaxial tension is shown in Fig. 7.48a,

where d is the elongation of the fracture zone (crack width) and Fm is the maximum
tensile force. The subscripts m and frac denote the corresponding quantities at the
maximum strength and failure, respectively. It should be pointed out that d cannot
be represented by the total displacement in a uniaxial tension test as the fracture
occurs in a localized zone. Previous study (Weerheijm and Van Doormaal 2007)
showed that d is in the order of 0.1 mm.
In order to determine the relation between tensile stress and tensile strain that is
necessary for the material model, a relation between the displacement d and the
tensile strain e must be assumed. Following the assumption made by Weerheijm
and Van Doormaal (2007), in which e is defined as d=lfrac , the corresponding
stress-strain curve is obtained, as shown in Fig. 7.48b, where lfrac denotes the length
of the fracture zone. Extensive static and dynamic tensile tests have been conducted
previously, but there is little quantitative information about the size of the fracture
zone. Based on the recommendation by Bazant (1992) for static loading, the size of
the fracture zone is typically 1–6 times of the maximum aggregate size (Malvar
et al. 2000).
310 7 Numerical Simulation of Projectile Impact on Concrete Targets

Fig. 7.48 Typical F

load-displacement and
(a)
stress-strain curves for
uniaxial tension of concrete: Fm
a load-displacement curve
and b stress-strain curve

δm δ frac δ

(b)

G f /lfrac

ε m; λ = λ m ε frac ; λ = nλ m

Since the static fracture energy Gf required to fail the concrete is dissipated
within lfrac, the specific strain energy in the entire cracking process should meet the
following energy conservation condition (Weerheijm and Van Doormaal 2007):

Zefrac
rde ¼ Gf =lfrac ð7:54Þ
em

In FE simulation, lfrac is always related to the characteristic length of the element

hc, which can be approximated by the cube root of the volume of the element in a
3D analysis. In Sect. 7.4.2, lfrac = hc is used based on the assumption that the tensile
fracture occurs in one element, as shown in Fig. 7.49a. However, for the accurate
simulations of the cratering and scabbing phenomena, hc is usually needed to be
smaller than lfrac, which means that the fracture occurs in several elements, as
shown in Fig. 7.49b. Consequently, the element of size hc will dissipate fracture
energy of (hc/lfrac)Gf, since (lfrac/hc) of them are needed to fill the fracture zone and
develops the fracture energy of Gf.
In the original K&C model, if not provided, lfrac can be taken as the element size
hc, but no less than three times of the concrete maximum aggregate size. When
lfrac = 10 cm and Gf is around 100 N/m, Eq. (7.54) is satisfied (Malvar et al. 2000).
For other lengths of the fracture zone, in order to satisfy Eq. (7.54) and maintain the
static fracture energy around 100 N/m at the same time, a relative element size
7.5 Modified K&C Model 311

(a) (b) Microcrack

hc

hc
l frac
l frac

Fig. 7.49 Schematic diagram of the relationship between lfrac and hc, tensile fracture occurs in
a one element and b several elements

parameter rsize = lfrac/(10 cm) is introduced to obtain new damage function values
for the strain-softening stage as follows:

k0i ¼ km þ ðki  km Þ=rsize ð7:55Þ

where the subscript i ranges from 6 to 13, corresponding to the strain-softening

stage. Then, relationship between η and kfor the  strain softening is taken as the
piecewise linear curve using new pairs of k0i ; gi as shown in Eq. (7.55).
There are some weaknesses of the K&C tensile damage model discussed as
follows:
(i) For static fracture energy far from 100 N/m, b2 must be redetermined by
adjusting its value until Eq. (7.54) is satisfied, which is time-consuming.
(ii) The relationship between η and k controls both the compressive and tensile
behavior of concrete material. Correspondingly, the treatment for ensuring
the static fracture energy affects not only the tensile strain-softening branch,
but also the compressive strain-softening branch. When the length of the
fracture zone is relatively small, the compressive strain softening will be not
obvious, making the concrete material relatively “hard,” which may be
unrealistic.
(iii) As will be shown in Fig. 7.50, for the original K&C model, both the pre-
dicted fracture stain and the fracture energy increase with the increase of the
strain rate, which is inconsistent with recent experimental observation
(Weerheijm and Van Doormaal 2007) that the fracture strain at high strain
rate is actually a constant and the fracture energy increases with the increase
of strain rate in tension.
In order to address the above three issues, the tensile damage accumulation is
modified as follows:
8
X< Dep
; p\0 and k\km
rf ð1 þ p=rf T Þb2
k¼ ð7:56Þ
: Dep
els ; p\0 and k  km
312 7 Numerical Simulation of Projectile Impact on Concrete Targets

Fig. 7.50 Comparisons of 30

dynamic tensile stress-strain Original K&C model
curves between modified and 25 Modified K&C model
original K&C models
20 100s 1

Stress (MPa)
15

10

5 10s 1 =0.01
frac

1
1s
0
0.000 0.004 0.008 0.012 0.016
Strain

where els is a constant related to the length of failure zone and static fracture
energy. According to Eq. (7.56), tensile damage accumulation for the
strain-hardening stage is the same as that in the original K&C model, while a
modification is made for the strain-softening stage. Considering the elastic strain
eR
frac eR
frac
limit is relatively small, rde  rdep can be obtained. Assuming that the
em em
tensile failure happens at k = nkm, where stress approaches zero, as shown in
Fig. 7.50, the left side of Eq. (7.54) is replaced by

Zefrac Zefrac Zefrac

rde  rdep ¼ Tgdep ð7:57Þ
em em em

where η is given by Eq. (7.53).

Using Eq. (7.56), for the strain-softening stage, dep ¼ ðelsÞdk; consequently,
Eq. (7.57) can be further transformed into

Zefrac Znkm Zn
x
Tgdep ¼ TgðelsÞdk ¼ T  ðelsÞ  km dx ð7:58Þ
ac ðx  1Þad þ x
em km 1

Based on Eqs. (7.54 and 7.58), els is expressed by

Gf
els ¼ ð7:59Þ
Rn
lfrac  T  km x
ac ðx1Þad þ x
dx
1

Consequently, the fracture strain is determined according to Eq. (7.56) as

7.5 Modified K&C Model 313

efrac ¼ nkm els ð7:60Þ

The length of failure zone, static fracture energy, and fracture strain are related
directly in the proposed tensile damage model, which avoids the time-consuming
parameter calibration in the original K&C model. Besides, as shown in Eq. (7.56),
the proposed tensile damage model does not affect the compressive strain-softening
branch, which is more suitable than the original K&C model. In order to gain
insight into the modified tensile damage accumulation, dynamic tensile stress-strain
curves for 50-MPa concrete under different strain rates are calculated. The static
fracture energy is assumed to be 85 N/m which is the same as the automatic
generated one in the original K&C model. The length of fracture zone is assumed
equal to the side length of the cubic element (10 mm). Thus, els and efrac are
obtained as 1.15 and 0.01, respectively. As will be discussed in Sect. 7.5.6, the
parameter n has limited influence on the cratering and scabbing simulation results
when it is larger than 100 where the tensile stress approaches zero. Consequently,
n is set as 100. The predicted stress-strain curves at different strain rates are shown
in Fig. 7.50, where the results from the original K&C model are also presented for
comparison. It can be observed that for the modified K&C model, the fracture stain
is a constant and the fracture energy increases with the increase of the strain rate,
which is consistent with the experimental observations (Weerheijm and Van
Doormaal 2007). While for the original K&C model, both the fracture stain and the
fracture energy increase with the increase of the strain rate, which may overestimate
the fracture energy at high strain rates.
The predicted stress-strain curves for biaxial and triaxial tensions are also of
interest and shown in Fig. 7.51. For biaxial tension, it is found that the strain
softening predicted by the original K&C model is almost linear, which does not
give good prediction. For triaxial tension, although the volumetric damage is
considered in the original K&C model, the predicted stress-strain relation is perfect
elasto-plastic, indicating that there may be an error during the implementation of the

(a) (b)
4.5 4.5
Original K&C model 4.0
4.0
Modified K&C model
3.5 3.5 Original K&C model
Modified K&C model
3.0 3.0
Stress (MPa)
Stress (MPa)

2.5 2.5
2.0 2.0
1.5 1.5
1.0 1.0
0.5 0.5
0.0 0.0
0.000 0.001 0.002 0.003 0.004 0.005 0.006 0.007 0.008 0.0000 0.0002 0.0004 0.0006 0.0008 0.0010
Strain Strain

Fig. 7.51 Stress-strain curves for biaxial and triaxial tension of the modified and original K&C
models: a biaxial tension and b triaxial tension
314 7 Numerical Simulation of Projectile Impact on Concrete Targets

volumetric damage into LS-DYNA. The modified K&C model describes the strain
softening well for both biaxial and triaxial tensions.

7.5.5 Numerical Simulation

The proposed modified concrete material model is implemented in finite element

code LS-DYNA (Hallquist 2007) through the user-defined material model. In order
to ensure that the modified model can predict the cratering and scabbing phe-
nomena, as well as projectile residual velocity at the same time, corresponding
results from several experiments with varying projectile mass, impact velocity, and
slab’s thickness should be realistically reproduced. Projectile perforation experi-
ments into concrete slabs conducted by Wu et al. (2015a) are numerically simulated
to validate the modified K&C model, in which a newly proposed erosion criterion is
used. Furthermore, corresponding simulation results of the original K&C model are
also presented to show improved performances of the modified K&C model.

7.5.5.1 New Erosion Criterion

For the numerical simulation using finite element approach, an erosion algorithm
should be carefully introduced to capture the physical crush (compressive failure)
and fracture (tensile failure) of concrete material. When a specified variable of an
element used to represent the compressive or tensile failure reaches its critical
value, the element is deleted immediately. It is generally accepted that the com-
pressive failure can be represented by the effective plastic strain (Huang et al. 2005;
Liu et al. 2009) or the maximum principal strain (Leppänen 2006; Tu and Lu 2010;
Polanco-Loria et al. 2008). For the tensile failure, the criterion is still inconclusive.
Tensile damage in the modified TCK model is considered in previous studies
(Huang et al. 2005; Liu et al. 2009) as the criterion for the tensile failure. However,
the value of tensile damage criterion is determined empirically as 0.5. Recently, Xu
and Lu (2006) and Li and Hao (2014) proposed a tensile fracture strain criterion
with an empirical critical value of 0.01 for the tensile failure of concrete material,
which demonstrates good agreement with experimental results. However, it is
observed that the simulation results for the cratering and scabbing are sensitive to
the critical tensile fracture strain (Kong et al. 2016a).
To resolve the empirical and sensitive issues in the above tensile failure criteria,
a new tensile failure criterion is proposed as follows:
8
>
< h 0; pi[ 0
X
Dkt ¼ Dep = rf ð1 þ p=rf T Þb2 ; p\0 and k\km ; kt ¼ Dkt ð7:61Þ
>
:
Dep =els; p\0 and k  km
7.5 Modified K&C Model 315

where the prefix D refers to the increment of variables during a time step. When the
kt reaches nkm, tensile failure occurs, and the element should be deleted immedi-
ately. Although an empirical constant n is introduced in the present tensile failure
criterion, it will be shown in Sect. 7.5.6 that the influence of n on the cratering and
scabbing simulation results is small when it is greater than 100 as the tensile stress
approaches to zero when strain is large. The compressive failure is still represented
by the effective plastic strain ep .

7.5.5.2 Wu et al.’s Test

Recently, we conduct twenty-five shots of reduce-scaled projectile perforation tests

on five configurations of monolithic and segmented RC slabs with a layer of rear
steel plate (Wu et al. 2015a). The cratering size, scabbing size, residual velocity,
and projectile acceleration are carefully recorded, which provide experimental data
for the validation of the modified K&C model.
The finite element model and element type for the RC target and projectile are
the same as that adopted in Sect. 7.4.5.2, which will not be repeated here. As
discussed in Sect. 7.5.5.1, element erosion technique with the criteria based on ep
and kt is adopted for the modified K&C model, i.e., the element will be deleted
immediately when ep or kt is greater than critical values. Since there are no available
element erosion criteria in the original K&C model, maximum principal strain
criterion is adopted using *MAT_ADD_EROSION in LS-DYNA.
The steel projectile is modeled as rigid material as little erosion and deformation
of the projectile were observed. The reinforcement bar is modeled as von Mises
material model (Mat_003 in LS-DYNA) with elastic modulus of 210GPa and yield
stress of 400 MPa. For the concrete material, the strength surfaces are defined by
the parameters given in Eqs. (7.51a–c). The damage-related parameters, i.e., a, ac,
ad, and km, adopt the suggested values, as shown in Fig. 7.47. Constant n for the
tensile damage is set as 100. Using the recommended relation between the static
fracture energy and unconfined compressive strength of concrete (i.e., Gf ¼ 73fc0:18 ,
fc in MPa unit), Gf is found to be 140 N/m. Since the length of the fracture zone is
absent in this experiment, it is assumed to be the maximum diameter of the
aggregate (Malvar et al. 1997), i.e., 10 mm. Consequently, els can be obtained
using Eq. (6.59). Other constants are obtained from the automatically generated
method in the original K&C model. Table 7.4 presents all the parameters of the
modified K&C model.
In order to avoid repeated simulations, Shot 1-1, Shot 2-3, Shot 4-3, and Shot
5-3, in which the thickness of the slabs and the impact velocity of the projectile
vary, are used to validate the modified model. The left of Fig. 7.52 presents the
simulation results of the original and modified K&C models, including cratering
diameter, cratering depth, scabbing diameter, scabbing depth, and the projectile
residual velocity. The corresponding experimental results are shown on the right of
Fig. 7.52. It is clearly observed that the cratering size, scabbing size, and residual
316 7 Numerical Simulation of Projectile Impact on Concrete Targets

Table 7.4 Parameters of the Strength surface Damage Others

modified K&C model for
41-MPa concrete a0 1.73e7 b1 1.6 q 2300 kg/m3
a1 0.5698 b3 1.15 fc 41 MPa
a2 6.14e-10 lfrac 10 mm G 12.8 GPa
a0y 1.15e7 els 2 T 3.6 MPa
a1y 0.8989 Gf 140 N/m
a2y 1.67e-9 km 8.7e−5
a0f 0 n 100
a1f 0.8989 a 3
a2f 1.67e-9 a c; 0.29;
ad 1.86

Experiments
(a) Original K&C model
80mm
Modified K&C model
180mm
Front surface Rear surface

15mm
42mm

300mm 380mm

52mm 68mm
Cratering depth: 55mm Sacbbing depth: 55mm
120mm 410mm
Residual velocity 279m/s Residual velocity 273m/s Residual velocity 272m/s

(b) Original K&C model

65mm
Modified K&C model Experiments
216mm
17mm

55mm
220mm 270mm

54mm 48mm

Cratering depth: 50mm Sacbbing depth: 75mm

120mm 275mm

Residual velocity 304m/s Residual velocity 313m/s Residual velocity 323m/s

(c) Original K&C model

90mm
Modified K&C model Experiments
220mm
10mm
28mm

200mm 230mm

36mm 38mm

120mm Cratering depth: 50mm Sacbbing depth: 50mm

260mm
Residual velocity 489m/s Residual velocity 507m/s Residual velocity 508m/s

(d) Original K&C model

85mm
Modified K&C model Experiments
192mm

25mm
36mm
150mm 200mm

35mm 45mm
270mm
Cratering depth: 50mm Sacbbing depth: 50mm
100mm

Residual velocity 368m/s Residual velocity 386m/s Residual velocity 354m/s

Fig. 7.52 Comparisons of simulation results with experimental data: a Shot 1-1: slab thickness
300 mm, projectile initial velocity 641 m/s; b Shot 2-3: slab thickness 200 mm, projectile initial
velocity 544 m/s; c Shot 4-3: slab thickness 150 mm, projectile initial velocity 508 m/s; and
d Shot 5-3: slab thickness 100 mm, projectile initial velocity 486 m/s
7.5 Modified K&C Model 317

(a) (b)
20 -250
Experimental data
Experimental data
-300 Original K&C model 0.0
Acceleration (10 4g)

15 Original K&C model

Displacement (m)
Modified K&C model
-350

Velocity (m/s)
Modified K&C model
-0.1
10 -400 Velocity

-450
-0.2
5 -500
-550 -0.3
0
-600 Displacement
-5 -650 -0.4
-0.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Time (ms) Time (ms)

Fig. 7.53 Comparisons of simulation results with experimental data: a acceleration and b velocity
and displacement of the projectile

velocity predicted by the modified K&C model have good agreements with the
corresponding experimental data. However, both the cratering size and scabbing
size are largely underestimated by the original K&C model, which result from the
overestimation of fracture energy at high strain rate, and the absence of tensile
failure criterion. The residual velocities predicted by the modified and original
K&C models all agree well with the test data, since it is mainly dominated by the
compressive behavior of concrete material.
The predicted acceleration, velocity, and displacement of the projectile and the
corresponding experimental data for Shot 1-1 are shown in Fig. 7.53. By inte-
grating the acceleration-time curve, the velocity time and displacement time of the
projectile are obtained. It is found that the predictions by both the original and
modified K&C models have good agreements with the test data, as they are mainly
governed by the compressive behavior of the concrete material.

7.5.6 Parametric Analyses

The dynamic tensile behavior of concrete material, which is crucial for the correct
modeling of the cratering and scabbing phenomena, is governed by seven parameters
in the modified K&C model, i.e., T, Gf, DIFt, n, b2, lfrac, and els. Considering the
strain hardening is relatively small compared to the strain softening, the influence of
b2 (governing the strain hardening) on the predicted cratering size and scabbing size
should be small, and therefore, it will not be discussed further. As shown in Eq. (6.
59), T, Gf, and lfrac are related to els; consequently, sensitivity analyses of the three
independent parameters, i.e., n, els, and DIFt, are carried out by varying one
parameter by ±25 and ±50 percentages while holding the others constant. In order
to save computational time, the simulation results from Sect. 7.5.5.2 with projectile
initial velocity of 544 m/s (Shot 2-3) are chosen as a baseline.
318 7 Numerical Simulation of Projectile Impact on Concrete Targets

Fig. 7.54 Effect of n on the 50

Rate of cratering and scabbing sizes (%)

cratering size and scabbing Cratering diameter
size Cratering depth
Scabbing diameter
25 Scabbing depth

-25

-50
-60 -40 -20 0 20 40 60
Varying rate of n (%)

7.5.6.1 Influence of Parameter n

Figure 7.54 presents the influence of the parameter n on the cratering size and
scabbing size, in which their changing rate is defined as the relative changes of the
quantity to its corresponding baseline value. In all simulations, it is found that the
failure modes are always similar for the values of n examined. Furthermore, it is
interesting to observe that the cratering size and scabbing size, especially the
scabbing size, are not very sensitive to parameter n. This is because the dynamic
tensile stress approaches to zero when n reaches a relative large value, beyond
which the contribution of dynamic stress-strain curve to the fracture energy is
minor, as observed in Fig. 7.50. Therefore, without invoking the complex cali-
bration of tensile failure parameters, the proposed erosion criterion with
kt [ 100km can be further used to describe other tensile-dominated failure phe-
nomena, such as the spallation of concrete slabs subjected to blast loadings.

7.5.6.2 Influence of Parameter els

The influence of els on the cratering and scabbing is shown in Fig. 7.55. It is
observed that the cratering in the concrete slab is not very sensitive to els. This
implies that the cratering phenomenon is dominated by the compressive property of
concrete material. The influence of els on scabbing phenomenon is found to be
evident, and the simulated scabbing size decreases with the increase of els. Besides,
the crack length decreases with the increase of els. This is a result of the increase of
the fracture energy with the increase of els, as shown in Eq. (7.59).
7.5 Modified K&C Model 319

Fig. 7.55 Effect of els on cratering and scabbing

7.5.6.3 Influence of DIFt

The influence of DIFt on the cratering and scabbing phenomena is also found
interesting. Figure 7.56 presents the simulation results of various DIFt, including
the suggested ones by Malvar and Crawford (1998) and Xu and Wen (2013)
(Eq. 7.52), CEB without a cutoff value and CEB with a cutoff value of 10. It is
found that the predicted cratering and scabbing phenomena are very similar using
the suggested DIFt by Malvar and Crawford (1998) and Xu and Wen (2013), which
are actually close to each other, as shown in Fig. 7.46. However, using the
CEB-recommended DIFt cannot reproduce the scabbing phenomenon well, and
even a cutoff value of 10 is introduced, since the CEB-recommended DIFt greatly
underestimates the experimental data, as shown in Fig. 7.46. Same conclusion is
presented in Sect. 7.4.6.2, which indicates that this conclusion is independent of
concrete material model.

DIFt suggested by Malvar and Crawford (1998) Present DIFt (Eq. (7.52), baseline)

CEB recommended DIFt CEB recommended DIFt with cutoff value of 10

Fig. 7.56 Effect of DIFt on cratering and scabbing

320 7 Numerical Simulation of Projectile Impact on Concrete Targets

7.6 Summary

Numerical simulations of projectile penetration at macroscopical scale are discussed

in this chapter, and extensive attentions are drawn from concrete material models.
The general descriptions of the material models usually consist of constitutive
relation and EOS. The constitutive relation is used to describe the deviatoric
behavior of the material, and EOS is used to describe the compaction behavior of
the material. The frequently used material models (HJC model and K&C model) in
commercial hydrocodes are briefly introduced, where the disadvantages of the
above two models are discussed, which can serve as a guide for further improve-
ment of these models.
In order to accurately model the behavior of concrete materials subjected to
projectile impact, the parameters of the yield surface and EOS for all material
models are important, which rely on data fitting to relevant experimental data, such
as triaxial compression data and flyer-plate-impact test. However, these experiments
are always absent in projectile penetration tests. Consequently, parameter calibra-
tion of the yield surface and EOS for NSC and HSC is conducted, respectively,
where the original HJC model is chosen as an example. Compared with the pre-
viously suggested values of the yield surface and EOS parameters, the proposed
parameters give better predictions regarding the DOP and residual velocity, espe-
cially for the HSC target.
Another important issue for concrete material is strain rate effect. The four
definitions of the effective strain rate are summarized, and theoretical analysis is
conducted to assess the deviations of the different definitions of the effect strain rate
based on a bilinear elasto-plastic model. For metal-like material or no dilatancy
effect material, the different definitions of the effective strain rate have insignificant
impact on the numerical simulations. However, for the concrete or geological
materials, none of the four definitions of the effective strain rate can be in good
agreement with axial strain rate and a great overestimation of the dynamic strength
would arise if the test data (DIF-axial strain rate curve) is embedded directly into
material model. Based on the relation between the effective strain rate and axial
strain rate, a corrected computation of the effective strain rate in K&C model is
proposed to ensure the effective strain rate equal to the axial strain rate in uniaxial
condition. Moreover, an iterative corrective methodology is proposed to eliminate
the inertia effect, and it is shown that the real inertia effect should not be as large as
described in the previous studies.
In order to obtain the improved numerical simulations of the cratering and
scabbing phenomena in concrete slabs subjected to projectile impact, the original
HJC and K&C material models are modified. The modified models are imple-
mented in the finite element code LS-DYNA through the user-defined material
model. To validate the improved models, several sets of projectile perforation
experiments are numerically simulated. The predicted cratering size, scabbing size,
and residual velocity show reasonable agreements with the corresponding experi-
ments, while the original HJC and K&C models cannot describe the cratering and
7.6 Summary 321

scabbing phenomena well. The sensitive analysis of DIFt indicates that the strain
rate effect should be described by the “radial enhancement” approach, and it is
crucial to select DIFt accurately in order to correctly simulate the cratering and
scabbing phenomenon.
 Part II
Aircraft Impact
Chapter 8
Aircraft Impact Force

8.1 Introduction

The containment of the nuclear power plant (NPP) is the outermost and last safety
barrier of the nuclear reactor, the protective performance of which is rather critical
to prevent the leakage of radioactive fission products. However, with the increas-
ingly rampant terrorist attacks as well as the rapidly growing numbers of passenger
planes in-service, the NPP containment is confronted with more deliberate and
accidental impacting threats by commercial aircraft. However, before September
11, 2001, the existing NPP containments were not specifically designed to resist
any external impact load greater than a light aircraft crash force, and only the
probabilistic assessment method was used in the risk evaluation of large com-
mercial aircraft impact accident. In particular, after the 9/11 terrorist attack, the
commercial aircraft collision with the NPP containment has attracted much atten-
tion in the field of nuclear safety. In 2009, the USA Nuclear Regulatory
Commission (NRC 2009) required that the newly built NPP containments must take
into consideration of large commercial aircraft impact load (Jiang and Chorzepa
2014).
The aircraft impact force should be firstly investigated in detail since the impact
load directly affects the local failure and global response of the NPP containment.
To determine the aircraft impact force, Riera (1968) developed a one-dimensional
rigid perfectly plastic model to predict the normal impact force of aircraft hitting on
a rigid flat target approximately, which was recognized as the Riera function and
widely used thereafter. Dritter and Gruner (1976) put forward a differential method
to calculate the impact force by considering the elasto-plastic material properties of
aircraft. Afterward, Wolf et al. (1978) established a lumped mass-spring model, in
which the lumped mass of aircraft is distributed in N nodes connected by
elasto-plastic springs, and the predicted results fit well with the Riera function.
In the actual impact process, the scattering debris will lead to the losses of
energy and mass, which may weaken the impact force. Based on the energy

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 325
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_8
326 8 Aircraft Impact Force

conservation law, some modifications to the Riera function were performed.

Hornyik (1977) introduced a coefficient to the Riera function based on the energy
conservation law. However, the exact value of the coefficient was not obtained due
to the lack of experimental data. Bahar and Rice (1978) gave a simplified derivation
of the aircraft impact force-time history and derived the above coefficient to be 0.5.
However, Riera (1980) found that the coefficient suggested by Bahar and Rice
(1978) was a lower limit to the impact force-time history from a view of the energy
balance. The accurate modification values were not obtained until that an F-4 fighter
impact experiment on a flat concrete panel was conducted by Sugano et al. (1993a).
It was confirmed that the Riera method was a practical and reasonable approach to
evaluate the aircraft impact force and a reduction coefficient of 0.9 was proposed to
the inertial force portion of the Riera function.
Many previous researchers (Riera 1968, 1980; Wolf et al. 1978; Hornyik 1977;
Bahar and Rice 1978; Kar 1979) developed several theoretical methods for cal-
culating the aircraft impact force, which made a great contribution to study the
aircraft impact problem. However, these analytical methods required many
assumptions; thus, the target response and damage could not be obtained accurately.
In addition, conducting the aircraft impact test to derive the more reliable data needs
complex experimental techniques and great expenditure. Only one prototype F-4
fighter field test (Sugano et al. 1993a) is conducted until today. Besides, with the
rapid development of simulation software and computer performance, the numerical
simulation has become an efficient method to analyze the aircraft impact force,
which is adopted in the following investigation.
In this chapter, fine FE models of the commercial aircraft Airbus A320 and the
prestressed NPP containment are established, and the models are validated by the
available test data of an F-4 fighter impacting on a flat concrete wall (Sugano et al.
1993a) and prototype GE-J79 engine collision with concrete panels (Sugano et al.
1993b, c), respectively. The verified FE models are also used to conduct further
studies in Chap. 9. In addition, the Riera method is presented and the impact forces
derived from the Riera function and detailed numerical simulation results are
compared with each other. Finally, the influences of different containment radii on
the impact force are analyzed, and the Riera function is modified by proposing a
coefficient function to its inertial force portion.

8.2 Numerical Modeling

The existing studies (Siefert and Henkel 2014; Itoh et al. 2005) pointed out that in
order to improve the accuracy of numerical simulations, fine FE models that could
reflect the actual mass and stiffness distributions as much as possible are necessary
and critical. Thus, the fine FE models of A320 aircraft and prestressed NPP con-
tainment are established and verified by prototype impact experiments, respectively.
8.2 Numerical Modeling 327

8.2.1 Commercial A320 Aircraft

A representative commercial aircraft of A320 (Fig. 8.1a) is chosen in the following

studies, which has a maximum takeoff mass of about 78,000 kg. The length and
wingspan of the A320 aircraft are about 38 and 34 m, respectively.

8.2.1.1 Modeling

A fine FE model of A320 aircraft is shown in Fig. 8.1b. The main parts of an A320
aircraft, describing the structural behavior and representing the mass and stiffness
distributions, are considered as follows: the main load-carrying beam structures,
e.g., the floor board, wing frame, and landing gears; the secondary framework, e.g.,
stringers, ribs, and skin; aviation fuel and its tanks located in the wings; engines
fixed on the wing frame; additional payload of devices; and total 150 passengers, as
shown in Figs. 8.2, 8.3, and 8.4, respectively. The A320 aircraft FE model consists
of beam elements for main beams, ribs, stringers, and landing gears; shell elements
for skin and engines; SPH (smoothed particle hydrodynamics) elements for the
aviation fuel; and discrete mass elements for the additional payload corresponding
to the actual distribution approximately. The masses of the main portions of the
A320 model are nearly equal to the actual configuration, including the fuselage
about 19,100 kg, two main wings without fuel and tail wings about 14,500 kg,
aviation fuel about 17,000 kg, two engines about 4800 kg, and additional payload
of passengers and devices about 14,300 kg. The total mass of the A320 model is
about 72,200 kg (including the mass of 2500 kg caused by the mass scaling effect
to reduce the time step size), and the entire model includes about 2
105 elements.
The real A320 aircraft consists of some carbon fiber and glass fiber composite,
and the main structural parts are made of high-strength aluminum alloy for the ribs
and steel for girders. Since the detailed material parameters of the A320 are not
available, simplified aeronautical engineering materials are considered (Siefert and
Henkel 2014).
The following simulations are performed by the nonlinear transient dynamic
finite element software LS-DYNA. The material constitutive model of
MAT_SIMPLIFIED_ JOHNSON_COOK (MAT#098) (Johnson and Cook 1983),
reflecting the nonlinear and strain rate dependent behavior, is used for the main

Fig. 8.1 A320 aircraft a prototype b FE model

328 8 Aircraft Impact Force

Fig. 8.2 Fuselage of A320 aircraft a real fuselage b FE model

Fig. 8.3 Engine of A320 aircraft a real engine (http://iae.wpengine.com/products/) b FE model

structures (Table 8.1), including 4340 steel for the floorboard beams, wing frame,
and landing gears and 2024 aluminum for the stringers and ribs. The aircraft skin
and engine are modeled by the model of MAT_PLASTIC_KINEMATIC
(MAT#003) (Table 8.2), which includes strain rate effect and is computationally
efficient. The aviation fuel is modeled by MAT_NULL (MAT#009) (Table 8.3),
8.2 Numerical Modeling 329

Fig. 8.4 Fuel and tanks of

A320 aircraft

Table 8.1 Model parameters Material model SIMPLIFIED

of main structures JOHNSON COOK (#098)
(Johnson and Cook 1983)
Material type 4340 steel 2024 aluminum
Density qs (kg/m3) 7830 2770
Young’s modulus Es (Pa) 2.00e+11 7.00e+10
Hardening constant A (Pa) 7.92e+08 2.65e+08
Hardening constant B (Pa) 5.10e+08 4.26e+08
Hardening exponent n 0.26 0.41
Strain rate constant C 0.014 0.01
Poisson’s ratio v 0.29 0.33

and the discrete ELEMENT_MASS defining a lumped mass element assigned to a

nodal point is used for the additional payload.

8.2.1.2 Calibration

In the actual impact process, partial kinetic energy would be consumed to crush the
airframe, which may explain the mass loss effect due to the separation of debris.
A reduction coefficient is introduced into the Riera function to reflect the above
mass loss effect (Hornyik 1997; Bahar and Rice 1978; Kar 1979; Riera 1980).
In the numerical simulations, the mass loss can be achieved by using the erosion
algorithm with different failure criteria and the plastic strain is commonly used (Lee
and Kim 2013). With the decrease of the failure value, the airframe breaks up more
easily with relatively smaller impact force. Although the Riera function is modified
by the F-4 Phantom fighter impact test (Sugano et al. 1993a) with a reduction
coefficient of 0.9, the Riera method is also used to calculate the impact force and
impulse of the A320 aircraft theoretically. In order to satisfy the above theoretical
value, parametric influential analyses of erosion failure strain (FS) of the A320
330 8 Aircraft Impact Force

Table 8.2 Model parameters Material model PLASTIC KINEMATIC (#003)

of skin and engine
Density qs (kg/m3) 2700
Young’s modulus Es (Pa) 7.00e+10
Tangent modulus Et (Pa) 1.00e+10
Yield stress fy (Pa) 3.00e+08
Poisson’s ratio m 0.3
Hardening parameter b 0.5

Table 8.3 Model parameters of aviation fuel

Material model NULL (#009)
Density q (kg/m )3
750
Pressure cutoff PC (Pa) −10
Dynamic viscosity coefficient MU 1.00e−03
Relative volume for erosion in tension TEROD 0
Relative volume for erosion in compression CEROD 0
Young’s modulus E (Pa) 0

aircraft materials are conducted numerically. Afterward, a suitable erosion FS could

be determined by comparing the impact impulses obtained by simulations and the
Riera function, respectively.
1. Longitudinal mass distribution
In order to use the Riera function to calculate the impact force, the longitudinal
mass distribution of the A320 aircraft in FE model is needed, which is obtained by
measuring its mass section by section in the HyperMesh (Altair HyperWorks 2010)
software from Fig. 8.1b, as shown in Fig. 8.5.
2. Longitudinal crushing strength distribution
Before determining the crushing strength of the A320 aircraft in FE model, the
influences of the crushing strength on the impact process are firstly analyzed based
on the Riera function, which was established for the Boeing B707-320 aircraft. The
corresponding longitudinal mass and crushing strength distributions were given by
Riera (1968), as shown in Fig. 8.6.
When selecting five different percentages of the actual crushing strength of
B707-320 aircraft (Riera 1968), the effects of the crushing strength on the impact
force-time histories and impulse-time histories are shown in Figs. 8.7 and 8.8,
respectively, and no obvious difference is observed.
Figure 8.9 illustrates the influences of the crushing strength on the impact
process, which indicates that with the increase of the crushing strength, (i) the peak
value of the impact force linearly decreases in the small range (less than 5%);
(ii) the time corresponding to the peak value of the impact force slightly fluctuates
less than 2.5%; (iii) the length of crushed aircraft reduces with the increase of the
crushing strength. The aircraft is almost completely crushed when the crushing
8.2 Numerical Modeling 331

Fig. 8.5 Longitudinal mass

distribution of A320 aircraft 10
in FE model

Mass distribution(103kg/m)
8

0
0 5 10 15 20 25 30 35 40
Length(m)

Fig. 8.6 Mass and crushing 10 10

strength distributions of 9 B707-320 aircraft 9
B707-320 aircraft (Riera Mass distribution
Mass distribution (10 kg/m)

8 Crushing strength 8

Crushing strength (10 N)

1968)

6
7 7
3

6 6
5 5
4 4
3 3
2 2
1 1
0 0
0 5 10 15 20 25 30 35 40 45
Length (m)

Fig. 8.7 Influences of 90

aircraft crushing strength on Different percentages
80
the impact force-time histories of real crushing strength
70 50%
100%
Impact force (10 N)

60 150%
6

50 200%

40

30

20

10

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Time (s)
332 8 Aircraft Impact Force

Fig. 8.8 Influences of the 10

crushing strength on the 9 Different percentages
impact impulse-time histories of real crushing strength
8
50%
7 100%

Impulse (10 N·s)

6 150%

6
200%
5
4

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Time (s)

Fig. 8.9 Influences of 40

aircraft crushing strength on 35 Impact peak force
the impact process Arrival time of peak force
30
Crushed length
Percentage change (%)

25 Impact impulse
20 Riera's value
15
10
5
0
-5
-10
-15
0 20 40 60 80 100 120 140 160 180 200
Percentage of B707-320 crushing strength (%)

strength is less than 20% of the value proposed by Riera (1968); and (iv) the impact
impulse does not change and agrees well with the conservation law of momentum;
thus, the influences of crushing strength on the impact impulse may be ignorable.
Since the crushing strength of the A320 FE model is difficult to obtain, some
simplifications have to be done. As to the B707-320 aircraft shown in Fig. 8.6, it is
indicated that the linear relationship between the crushing strength distribution and
the mass distribution can be approximately assumed to be 1.2, which is also used
for the A320 aircraft in this section.
3. Determination of the erosion failure strain
In this section, based on the A320 aircraft model established above, a practical
erosion failure strain is determined by comparing the impact impulses derived by
the Riera function with numerical simulations.
8.2 Numerical Modeling 333

When the longitudinal mass distribution and the crushing strength distribution of
the A320 aircraft are determined as mentioned above, the impact force-time history
could be calculated based on Eq. (8.1), which was modified by the F-4 fighter
impact test at the velocity of 215 m/s (Sugano et al. 1993a). In order to eliminate
the influences caused by the impact velocity, the impact velocity of the A320
aircraft also is assumed as 215 m/s.

FðtÞ ¼ P½xðtÞ þ a  l1 ½xðtÞV 2 ðtÞ ð8:1Þ

where V(t) is the aircraft velocity, l1[x(t)] is the longitudinal mass distribution, and
P[x(t)] denotes the crushing force. a = 1 was originally proposed by Riera (1968),
and afterward, it was determined as 0.9 by the F-4 fighter impact test (Sugano et al.
1993a).
In numerical simulations, the concrete material constitutive model of
MAT_CSCM_CONCRETE (MAT#159) (Hallquist 2007; Lee et al. 2014) is uti-
lized, which is a continuous surface cap model with a smooth intersection between
the shear yield surface and hardening cap. The initial damage surface coincides with
the yield surface, and strain rate effect is described by viscoplasticity theory. The
yield function is based on the three invariants and the cap hardening parameter j0 ,
which is as follows

YðI1 ; J20 ; J30 ; j0 Þ ¼ J20  <2 ðJ30 ÞFf2 ðI1 ÞFc ðI1 ; j0 Þ ð8:2Þ

where I1 is the first invariant of the stress tensor, J20 ðJ30 Þ is the second (third)
invariant of the deviatoric stress tensor, Ff is the shear failure surface, Fc is the
hardening cap, and < is the Rubin three-invariant reduction factor. The cap hard-
ening parameter j0 is the value of the pressure invariant at the intersection of the cap
and shear surfaces.
Concrete exhibits softening in the tensile and low-to-moderate compressive
regimes.

rdij ¼ ð1  dÞrvp
ij ð8:3Þ

A scalar damage parameter d transforms the viscoplastic stress tensor without

damage rvp into the stress tensor with damage rd . Damage accumulation is based
upon two formulations of ductile damage and brittle damage, and the initial damage
threshold is coincident with the shear plasticity surface. The damage parameter
applied to the stresses is equal to the current maximum of the ductile or brittle
damage parameter. More detailed descriptions of MAT_CSCM_CONCRETE
(MAT#159) are introduced in the keyword user’s manual of LS-DYNA 971
(Hallquist 2007).
The impact force-time histories and impulse-time histories obtained by Eq. (8.1)
and simulations with the different values of erosion failure strain are shown in
Figs. 8.10 and 8.11, respectively. As shown in Fig. 8.10, the impact force-time
histories are basically identical in the variation shape, and the small differences of
334 8 Aircraft Impact Force

Fig. 8.10 Impact force-time 480

histories corresponding to Erosion type 1: FS=0.1
420 Erosion type 2: FS=0.4
different failure values
Erosion type 3: FS=0.2
360
Erosion type 4: FS=0.3

Impact force (10 N)

6
300 Erosion type 5: FS=0.25
Riera, a =1.0
240 Riera, a =0.9

180

120

60

0
0.00 0.05 0.10 0.15 0.20
Time (s)

Fig. 8.11 Impulse-time 16

histories corresponding to a =1.0
14
different failure values a =0.9
Impulse (10 N· s)

12

10
6

Erosion type 1: FS=0.1

8 Erosion type 2: FS=0.4
Erosion type 3: FS=0.2
6
Erosion type 4: FS=0.3
4 Erosion type 5: FS=0.25
Riera, a =1.0
2 Riera, a =0.9

0
0.00 0.05 0.10 0.15 0.20
Time (s)

the peak forces are mainly due to the fact that the Riera function could not com-
pletely reflect the actual impact process. As shown in Fig. 8.11, when the erosion
type 5 (FS = 0.25) is imposed, the impact impulse fits well with the calculated
value by Eq. (8.1) with a = 0.9. It should be pointed that the reduction coefficient
of 0.9 is determined based on the impact impulse (Sugano et al. 1993a). Figure 8.12
shows the simulated A320 aircraft impact phenomena and the corresponding
concrete target damage when the erosion type 5 is imposed. At 0.05 s, the nose and
front body of the A320 aircraft are crushed and the target damage shape is circular
corresponding to the fuselage section. Afterward, the main wings and engines
impact on the target at 0.10 s, leading to the collision debris flying and fuel
splashing after the fracture of tanks. The obvious damage area appears in the target,
which is similar to the impact zone of the main wings and engines. When the time
comes to 0.20 s, the debris and fuel scatter more widely and the tail wings crash on
the target after breaking away from the fuselage due to the shear failure of
8.2 Numerical Modeling 335

(a) (b)

t=0.05s t=0.05s

Fuel

Debris

t=0.10s t=0.10s

Fuel

Debris

t=0.20s t=0.20s

Fig. 8.12 A320 aircraft impact a phenomenon b target damage (erosion type 5)

connections. After the impact, the target zone corresponding to the vertical pro-
jection of the A320 aircraft is seriously damaged. The losses of the mass and energy
during the collision are well reproduced. Thus, the erosion type 5 is adopted in the
following numerical simulations.
One may note that the structural style of fighters and passenger planes is dif-
ferent, thus. Therefore, the validity of the A320 FE model is not completely
accurate based on the F-4 fighter impact test. However, since no other related and
more suitable aircraft impact experiments are available, the deviation due to the
difference in aircraft types is not taken into consideration. If the impact tests of
A320 aircraft or other more similar aircraft are conducted in the future, the erosion
FS for the A320 aircraft impact simulation should be readjusted.

8.2.2 Ling Ao NPP Containment

The single-layered and prestressed reinforcement concrete NPP containment is

widely applied in China, such as the Ling Ao NPP containment in Guangdong
Province. The containment mainly consists of a cylindrical barrel body and
336 8 Aircraft Impact Force

Fig. 8.13 The whole NPP

containment model

65m

0.9m
37m

approximate hemispherical dome with the total height of about 65 m and the wall
thickness of about 0.9 m, as shown in Fig. 8.13.

8.2.2.1 Modeling

The containment model principally includes the concrete, rebar (Fig. 8.14), steel
liner, prestressed tendon (Fig. 8.15), and its outer pipes, as shown in Fig. 8.16. The
whole containment model weights about 17000 t and contains nearly 1.5 million
elements.
The material constitutive model of MAT_CSCM_CONCRETE (MAT#159) is
used to model the concrete properties. The steel liner, rebar, prestressed tendon, and
its outer pipes are modeled by MAT_PLASTIC_KINEMATIC (MAT#003)
(Table 8.4).

[email protected]
[email protected]

Outward
25@220mm
25@230mm

Fig. 8.14 Rebar arrangement

8.2 Numerical Modeling 337

Steel liner
Longitudinal
r=18.50m truss
r=19.03m Circumferential
r=19.23m truss
0.22m
0.22m
r=19.40m

Fig. 8.15 Prestressed tendon arrangement

(a) (b)
Pipe Concrete

Steel liner

Prestressed
tendon

Reinforcement

Fig. 8.16 Details of prestressed NPP containment a real configuration (Lee et al. 2013) b FE
model

Table 8.4 Model parameters of steel liner, rebar, tendon, and pipes
Material model PLASTIC_KINEMATIC (#003)
Component Steel liner Rebar Tendon Pipes
Density qs (kg/m3) 7850 7850 7850 7850
Young’s modulus Es (Pa) 2.10e+11 2.00e+11 1.90e+11 2.00e+11
Tangent modulus Et (Pa) 3.00e+10 3.00e+10 2.70e+10 3.00e+10
Yield stress fy (Pa) 3.00e+08 3.00e+08 1.48e+09 3.00e+08
Poisson’s ratio v 0.29 0.29 0.29 0.29
Hardening parameter b 0.5 0.5 0.5 0.5
338 8 Aircraft Impact Force

8.2.2.2 Calibration

The concrete material constitutive model and its parameters are the most critical
factors influencing the simulation accuracy. The constitutive model of
MAT_CSCM_CONCRETE (MAT#159) is convenient to use and widely adopted
to model the concrete (Lee and Kim 2013; Lee et al. 2014). To identify the accuracy
of the above concrete constitutive model, the available prototype impact experiment
of a fighter engine is simulated.
1. Prototype engine impact test
Sugano et al. (1993b) conducted the No. L4 prototype engine GE-J79 impact
experiment, and the related test parameters are listed in Table 8.5.
As shown in Figs. 8.17 and 8.18, the fine FE models of GE-J79 engine and
concrete panel are established, respectively. The material constitutive model and
related parameters of the engine are identical with the A320 aircraft as listed in
Table 8.1.
2. Comparisons with test data
Firstly, the static compression experiment of the engine is simulated and compared
with the actual test to check the accuracy of the engine FE model. The simulation of
static compression process is shown in Fig. 8.19. The comparisons of experimental
results with numerical simulations are shown in Fig. 8.20, which indicate that the
material model and the crushing strength distribution of the engine basically fit well
with the real GE-J79 engine.
The engine velocity, the impact force, the strain of the rebar in the impact center,
and the displacement-time histories in the impact center, as well as the panel
damage contours, are shown in Figs. 8.21, 8.22, 8.23, 8.24, and 8.25, respectively.
The results of numerical simulations fit well with the actual impact test, which

Table 8.5 Test parameters of no. L4 (Sugano et al. 1993b)

Engine type Diameter (mm) Mass (kg) Velocity (m/s) Panel thickness
(mm)
Prototype GE-J79 760 1767 215 1600

Fig. 8.17 GE-J79 engine a real engine (Sugano et al. 1993b) b FE model
8.2 Numerical Modeling 339

Fig. 8.18 Panel a real configuration (Sugano et al. 1993b) b FE model

Fig. 8.19 Simulations of engine compression process (crushed length) a 0.0 m b 0.5 m c 1.0 m
d 1.5 m e 2.0 m f 2.2 m

validates the applicability and accuracy of the selected constitutive models for
concrete and rebar material to some extent.

8.3 Calculating Methods for Aircraft Impact Force

In this section, the Riera method is presented and the impact force derived from the
Riera function and numerical simulations is compared.

8.3.1 Theoretical Analysis

Riera (1968) developed a one-dimensional rigid perfectly plastic model to predict

the normal impact force of an aircraft hitting on a rigid flat target approximately,
which is recognized as the Riera function. The Riera function has served as a basis
340 8 Aircraft Impact Force

Fig. 8.20 Compression (a) 5

results of engine a crushing
strength b absorbed energy
4 Test data

Crushing strength (10 N)

Numerical simulation

6
3

0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Crushed length of the engine (m)

(b) 5

4 Test data
Absorbed energy (10 N·m)

Numerical simulation
6

0
0.00 0.25 0.50 0.75 1.00 1.25 1.50
Crushed length of the engine (m)

Fig. 8.21 Tested and 250

simulated engine Test data
velocity-time histories Numerical simulation
200
Engine velocity (m/s)

150

100

50

0
0.000 0.005 0.010 0.015 0.020
Time (s)
8.3 Calculating Methods for Aircraft Impact Force 341

Fig. 8.22 Tested and 50

simulated impact force-time Test data
histories of the engine Numerial simulation
40

Impact force (10 N)

6
30

20

10

0
0.000 0.005 0.010 0.015 0.020
Time (s)

Fig. 8.23 Tested and 8

simulated rebar strain-time
7 Test data
histories in the impact center Numerical simulation
6

5
Strain (10 )
-3

0
0.00 0.01 0.02 0.03 0.04 0.05
Time (s)

for predicting the impact force-time history of aircraft collision and has been
included in the NEI (2011) report, the DOE (2006), and IAEA (2003) guide. In the
Riera function, an aircraft impact force function is presented by summing the
crushing strength and the force required to decelerate the mass of “deformation”
zone. As shown in Fig. 8.26, Riera (1968) assumed a negligibly thin “deformation”
zone between the rigid target and the rigid portion of the aircraft. Besides, the
following assumptions are also made: (i) The target is rigid and flat; (ii) the aircraft
velocity direction and fuselage axis coincide and are normal to the target; (iii) the
deformation zone only exists at the cross section next to the target; (iv) the material
of the aircraft is regarded as rigid perfectly plastic; and (v) the crushed portion of
the aircraft is neglected, while the uncrushed portion is decelerated as a whole by
the crushing strength.
342 8 Aircraft Impact Force

Fig. 8.24 Tested and 35

simulated displacement-time
histories in the impact center 30

25

Displacement(mm)
20

15

10

5 Test data
Numerical simulation
0
0.00 0.01 0.02 0.03 0.04
Time (s)

Fig. 8.25 Panel damage after the engine impact a front side b back side

Fig. 8.26 Riera model of

aircraft impact Rigid portion Deformation zone
Rigid
target
X
Fx
Velocity of rigid portion V

Control Thickness of
Volume deformation zone

The momentum of an aircraft M(t)V(t) is determined by the initial velocity and

the total mass. The impact force of an aircraft generated in a small time interval of
dt, corresponding to the changing rate of the momentum, is defined as follows
8.3 Calculating Methods for Aircraft Impact Force 343

d d d
FðtÞ ¼ ½MðtÞVðtÞ ¼ MðtÞ ½VðtÞ þ VðtÞ ½MðtÞ
dt dt dt ð8:4Þ
¼ MðtÞaðtÞ þ VðtÞl1 ½xðtÞVðtÞ ¼ P½xðtÞ þ l1 ½xðtÞV 2 ðtÞ

where V(t) and a(t) are the velocity and deceleration of an aircraft mass M(t),
respectively. l1[x(t)] is the longitudinal mass distribution. Hence, the right-hand
side of Eq. (8.4) consists of the force P[x(t)] to decelerate the uncrushed portion and
an inertial force l1[x(t)]V2(t) due to the velocity change of the crushed mass.
As shown in Eq. (8.4), only the crushing strength, the mass distribution, and the
initial velocity are needed in the Riera function. Therefore, a numerical step-by-step
scheme can be designed to calculate the impact force-time history (Lee et al. 2013).

8.3.2 Numerical Simulation

The impact forces induced by the different parts (fuselage, engines, and wings) of
the A320 aircraft are calculated, respectively, as shown in Fig. 8.27. The impact
height is 30 m, and the impact velocity is 215 m/s perpendicular to the containment
axis.
Figure 8.27 indicates that (i) the fuselage impact force sustains the whole impact
process and vibrates with four impact force peaks, which are caused by the impact
of aircraft nose (intensive electronic equipment), connecting components between
wings and fuselage, landing gears, and aircraft tails, respectively; (ii) the wings
impact process only appears at about 0.05–0.15 s, nearly a half of the whole impact
duration. At about 0.06 s, the impact force sharply rises to about 165
106 N
corresponding to the impact of main wings, massive fuel, and tanks at the same
time. However, at about 0.125 s, a relative small impact force peak (54
106 N)

Fig. 8.27 Impact force-time 200

Unfiltered 50Hz low-pass filtered
histories of three parts of the
Fusalage Fusalage
A320 aircraft Engines Engines
150 Wings Wings
Impact force (10 N)
6

100

50

0
0.00 0.05 0.10 0.15 0.20
Time (s)
344 8 Aircraft Impact Force

Fig. 8.28 Impact of external Impact of external

broken wings at about 0.125 s broken wings

Fig. 8.29 Force-time 500

histories of A320 aircraft Unfiltered
impacting on containment and Fixed flat panel
fixed flat panel 400 NPP containemnt
Impact force (10 6N)

50Hz low-pass filtered

300 Fixed flat panel
NPP containment

200

100

0
0.00 0.05 0.10 0.15 0.20
Time (s)

appears since the wings fracture and the external broken wings impact the cylin-
drical containment subsequently (Fig. 8.28); and (iii) the duration time of the
engine impact is shorter (about 0.03 s) with a single peak (60
106 N), which
appears at the same time with the peak induced by the wings impacting.
The whole impact force-time histories and impulse-time histories of the A320
aircraft collision with the cylindrical containment and fixed flat panel are shown in
Figs. 8.29 and 8.30, respectively.
Figures 8.29 and 8.30 indicate that (i) the peak impact force of the aircraft crash
on the containment is nearly 40% less than that of the aircraft collision with the
fixed flat panel, since the cylindrical section of the containment decomposes the
normal impact velocity and the containment deformation would absorb some
kinetic energy; (ii) the collision of broken wings with the containment is later than
that of the entire wings with the flat panel; thus, there is a small peak force appears
at about 0.125 s; and (iii) for the cylindrical containment, the reduction coefficient
in the Riera function decreases to about 0.78, which is smaller than the suggested
8.3 Calculating Methods for Aircraft Impact Force 345

Fig. 8.30 Impulse-time 16

histories of A320 aircraft α=1.0
14
impacting on containment and α=0.9
fixed flat panel 12 α=0.78

Impulse (10 N·s)

10

6
Fixed flat panel
8
NPP containment
Riera， α=1.0
6
Riera， α=0.9
4 Riera， α=0.78

0
0.00 0.05 0.10 0.15 0.20
Time (s)

Fig. 8.31 Comparisons of 275

different impact force-time 250 A320_215m/s (Simulation)
histories Phantom F4_215m/s (Sugano et al., 1993a)
225 B707-320_103m/s (Riera, 1968)
200 A320_160m/s (Siefert and Henkel, 2014)
Impact force (10 N)
6

175
150
125
100
75
50
25
0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Time (s)

coefficient of 0.9 derived from the F-4 fighter impact test onto a flat concrete panel
(Sugano et al. 1993a). It indicates that the containment shape would apparently
influence the impact force, and the influences of containment radius on the impact
force will be discussed in Sect. 8.4.
As shown in Fig. 8.31, the typical impact force-time histories of aircraft given in
the existing literatures are compared with the numerical results, which indicate that
the variation shape of the impact force-time histories is basically similar. However,
the duration time and the peak load are obviously different for various aircrafts and
impact velocities.
346 8 Aircraft Impact Force

8.4 Influence of Containment Radius on Impact Force

As shown in Figs. 8.29 and 8.30, the impact force and impulse of the A320 aircraft
impacting on the flat panel and cylindrical containment are obviously different;
thus, the influences of containment radius are further investigated. Besides, the
Riera function is also based on the hypothesis of the flat panel, which limits its
application to the non-planar target. Thus, it is worth to further discuss the influ-
ences of target radius on the impact force.
In this section, the detailed numerical simulations of an A320 commercial air-
craft impacting on the containment with different radii are performed to analyze the
weakening effect of the circular sectional NPP containment on the impact force.
Thereafter, the original Riera function is modified to account for this effect.

8.4.1 Theoretical Analysis of Impact Impulse

The aircraft impact velocity during the collision will be decomposed due to the
circular sectional shape, as shown in Fig. 8.32.
According to the conservation law of momentum and the Riera function, the
impact impulse is equal to the initial momentum. Hence, calculating the impact
impulse of the A320 aircraft based on the initial momentum is a convenient option,
which does not need to consider the crushing strength and inertial force. The
impulse calculating formula including the target radius effect could be written as

Z a ;Rc Þ
minðL Z a ;Rc Þ
minðL
R2c  x2
I¼ l2 ðxÞVcos hdx ¼ 2V
2
l2 ðxÞ dx ð8:5Þ
R2c
minðLa ;Rc Þ 0

where La is half of the total aircraft wingspan, Rc is the external radius of the NPP
containment, V is the initial impact velocity, and l2(x) is the transverse mass

Fig. 8.32 Decompositions of

aircraft impact velocity
8.4 Influence of Containment Radius on Impact Force 347

Fig. 8.33 Transverse mass

distribution l2 of A320
aircraft

distribution of the aircraft (Fig. 8.33). However, the Eq. (8.5) could not consider
the influences of the losses of mass and energy, as the coefficient of the Riera
function is not introduced. The calculation results of the Eq. (8.5) are shown in
Fig. 8.38.

8.4.2 Numerical Simulation of Aircraft Impact Process

The main purpose of this section is to study the radius effect. If the integrated model
is adopted, some secondary factors, such as wall thickness, concrete strength,
tendon prestress, and reinforcement ratio, may influence the conclusions. Thus, the
target is dealt with the simplified circular sectional walls instead of an entire NPP
containment. Figure 8.34 shows the simplified concrete NPP containments with the
different radii from 10.0 to 30.0 m at the interval of 2.5 m. The containments are
built with solid elements and modeled by the concrete constitutive model of
MAT_CSCM_CONCRETE (MAT#159). Similarly, the simplified containments
are also 0.9 m in thickness with the meshing size of 0.225 m, and the inner
boundaries of the containments are fixed. The impact velocity is 215 m/s and
perpendicular to the containment axis.
The impact phenomena are shown in Fig. 8.35. It can be seen that the aircraft is
seriously crushed and the main wings fracture at its center position. Besides, the
decomposition effect of the containment attenuates with the radius increasing.
Figures 8.36 and 8.37 show the impact force-time histories and impulse-time
histories when the aircraft impacting on containments with different radii, respec-
tively. It can be derived that the peak impact force and impulse reduce with the
348 8 Aircraft Impact Force

30.0m
27.5m
25.0m
22.5m
20.0m
17.5m
15.0m
12.5m
10.0m

Fig. 8.34 Impact of the A320 aircraft on NPP containments with different radii

R c =10.0m R c =12.5m R c =15.0m

R c =17.5m R c =20.0m R c =22.5m

R c =25.0m R c =27.5m R c =30.0m

Fig. 8.35 Phenomena of A320 model impacting on containments with various radii at 0.2 s
8.4 Influence of Containment Radius on Impact Force 349

Fig. 8.36 Impact force-time 450

histories of the different Rc=10.0m
400
containment radii Rc=12.5m
350 Rc=15.0m

Impact force (10 N)

300 Rc=17.5m

6
Rc=20.0m
250
Rc=22.5m
200 Rc=25.0m
150 Rc=27.5m
Rc=30.0m
100
Rc=Infinity (flat)
50

0
0.00 0.05 0.10 0.15 0.20 0.25
Time (s)

Fig. 8.37 Impulse-time 15

histories of the different 14
containment radii 13
12
11
Impulse (106 N·s)

10
9
8
Rc=10.0m Rc=22.5m
7
6 Rc=12.5m Rc=25.0m
5 Rc=15.0m Rc=27.5m
4 Rc=17.5m Rc=30.0m
3
Rc=20.0m Rc=Infinity
2
1 (flat)
0
0.00 0.05 0.10 0.15 0.20 0.25
Time (s)

decreasing of the radius. The external radius of Ling Ao NPP containment in China
is about 20 m; for example, if the influences of the containment radius are not
considered, the peak impact force and impulse will be overestimated about 20 and
10%, respectively.
Figure 8.38 shows the simulated impact impulse and 0.9 times of the theoretical
value derived by Eq. (8.5). They are in good agreement; thus, a coefficient of 0.9 is
assumed to consider the losses of mass and energy approximately.
Comparing the simulated impact impulses with the calculation results by the
Riera function, a series of coefficients corresponding to the different containment
radii could be derived. Figure 8.39 shows the coefficients for different ratios of the
containment diameter to the aircraft wingspan as well as the fitted curves.
Consequently, the Riera function with considering the influences of the NPP
containment radius and aircraft size could be expressed as
350 8 Aircraft Impact Force

Fig. 8.38 Impact impulses of 15

different containment radii

14

Impulse (106N·s)
13

12
Theoretical calculation value
0.9 times of the theoretical value
11
Numerical simulation values

10
10.0 12.5 15.0 17.5 20.0 22.5 25.0 27.5 30.0
Radius (m)

Fig. 8.39 Coefficients for 0.90

different ratios of the
containment diameter to the
0.85
aircraft wingspan
Correction coefficient

0.80

0.75

Simulated coefficients
0.70 Fitting curve

0.65

0.6 0.8 1.0 1.2 1.4 1.6 1.8

Rc/La

FðtÞ ¼ P½xðtÞ þ aðRc =La Þ
l1 ½xðtÞV 2 ðtÞ ð8:6Þ

0:14
aðRc =La Þ ¼ 0:9  ; Rc =La  0:6 ð8:7Þ
0:39 þ ðRc =La Þ3:3

8.5 Summary

This chapter can be summarized as follows:

(1) In Sect. 8.2, the fine FE models of the commercial aircraft A320 and the
prestressed NPP containment are established. The corresponding constitutive
8.5 Summary 351

models and their parameters are further confirmed by comparing the numerical
simulation results with the available test data.
(2) In Sect. 8.3, the collision process of the A320 aircraft on the NPP is numer-
ically simulated, and the impact force is derived. The detailed information,
such as the impact forces induced by the different aircraft parts, the
high-frequency oscillations of force-time histories, and the peak values of the
impact force caused by the collision of the external broken wings, is presented
and explained.
(3) In Sect. 8.4, the weakening effect of the circular sectional NPP containment on
the impact force is analyzed. The Riera function is modified by proposing a
coefficient function that is dependent on the dimensionless ratio of the NPP
containment diameter to the aircraft wingspan.
Chapter 9
Numerical Simulation of A320 Aircraft
Impact on NPP Containments

9.1 Introduction

For the numerical analyses of the dynamic responses and damages of the NPP
containment, partial existing studies have been conducted by the loading-time
history method (decoupled crash simulation) (Abbas et al. 1996; Petrangeli 2010;
Frano and Forasassi 2011; Iqbal et al. 2012; Sadique et al. 2013). Although the
loading-time history method is simpler and more efficient without needing the
aircraft FE modeling, there are some inevitable limitations, for example, the impact
zone is difficult to determine exactly. Besides, the influences of aircraft impact
angles and the containment shape are hard to be taken into consideration. Some
literatures further indicated that this decoupled method would result in relatively
slighter local damage and smaller impact deflection of the containment (Arros and
Doumbalski 2007; Siefert and Henkel 2014), while the missile-target interaction
method (integral crash simulation) is more widely applied (Siefert and Henkel
2014; Itoh et al. 2005; Lee et al. 2013, 2014), which requires both containment and
aircraft FE modeling and could simulate the impact process much better under
various impact scenarios.
In this chapter, two kinds of analysis approaches, i.e., the missile-target inter-
action method and the loading-time history method, are both adopted to simulate
the crash process and compared with each other. Considering that different impact
conditions may influence the collision process, a series of parametric studies are
conducted with the missile-target interaction method, including the aircraft impact
position, velocity, angle, and the containment wall thickness, rebar ratio, tendon
prestressing force.

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 353
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_9
354 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

9.2 Comparison Between the Integral and the Decoupled

Impact Simulations

The fine FE models of the A320 aircraft and the prestressed NPP containment have
been built and verified in Sect. 8.2. In this section, the entire crash process of the
A320 aircraft impacting on the containment is first simulated by the missile-target
interaction method, in which the impact height and velocity are 30 m and 215 m/s,
respectively. Then, the obtained impact force-time histories of the different aircraft
parts (fuselages, engines, and wings) are loaded at the corresponding impact zones
to conduct the decoupled crash simulations.

9.2.1 Impact Phenomenon Predicted by Missile-Target

Interaction Method

The impact phenomena of the A320 aircraft collisions on the containment by the
missile-target interaction method are shown in Fig. 9.1. At about 0.05 s, the aircraft
nose is crushed, while the other portions basically keep integral. At about 0.10 s,
the main wings and fuel tanks already strike the containment and aviation fuel
scatters due to the rupture of fuel tanks. The impact process terminates at about
0.20 s and the A320 aircraft debris and fuel spread widely.

9.2.2 Damage of the Prestressed NPP Containment

In Sect. 8.3.2, the impact force-time histories of the three parts (fuselage, engines
and wings) of the A320 aircraft have been obtained, which would be loaded at the
corresponding projection area to perform the decoupled crash simulation.

Fig. 9.1 A320 aircraft impact phenomena on the containment by missile-target interaction
method a t = 0.05 s b t = 0.10 s c t = 0.15 s d t = 0.20 s
9.2 Comparison Between the Integral and the Decoupled Impact Simulations 355

The numerically simulated damages of the prestressed NPP containment are given
as follows.

9.2.2.1 Damage of Concrete

The damages of the concrete in the containment obtained by the missile-target

interaction method and loading-time history method are shown in Fig. 9.2,
respectively. With the impact process proceeding, the concrete damage is becoming
more serious, while the damage degree predicted by the integral crash simulation is
apparently greater than that obtained from the decoupled approach.

9.2.2.2 Effective Plastic Strain of Steel Liner

Figure 9.3 shows the plastic strains of the steel liner obtained by the missile-target
interaction method and loading-time history method, respectively. The corre-
sponding maximum effective plastic strains of the steel liner accumulates to about
0.1 and 0.02 for the above two methods, respectively. Besides, the damage area
predicted by the missile-target interaction method is obviously larger than that by
the loading-time history method.

(a)

t=0.05s t=0.10s t=0.15s t=0.20s

(b)

t=0.05s t=0.10s t=0.15s t=0.20s

Fig. 9.2 Damages of the concrete in the containment a missile-target interaction method
b loading-time history method
356 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

(a)

t=0.05s t=0.10s t=0.15s t=0.20s

(b)

t=0.05s t=0.10s t=0.15s t=0.20s

Fig. 9.3 Effective plastic strains of the steel liner a missile-target interaction method
b loading-time history method

9.2.2.3 Axial Force of Prestressed Tendons

The variations of the axial force in the circumferential and longitudinal prestressed
tendons during the impact process predicted by the missile-target interaction
method and loading-time history method are shown in Figs. 9.4 and 9.5, respec-
tively. Before the aircraft impact, the tendons have been exerted prestressing forces
by dynamic relaxation method provided in LS-DYNA (Hallquist 2007). In addition,
the prestressing forces of the circumferential and longitudinal tendons are set as
4000 and 8000 KN according to the actual containment, respectively. The tendons
can move in the embedded steel tubes instead of being fixed with fillers, in order to
retighten the tendons to reduce the loss of partial prestressing force. Therefore, the
circumferential tendon forces may decrease because of the containment deformation
during the impact. However, the longitudinal tendon force would become greater,
since the length of the tendon is stretched.
As illustrated in Figs. 9.4 and 9.5, with the impact process proceeding, the axial
forces of the circumferential and longitudinal tendons obtained by the missile-target
interaction method change more greatly than that obtained by the loading-time
history method. The reason lies in that the predicted displacements by the above
two methods are different as illustrated in Sect. 9.2.3, which make the different
stretches in the tendons.
9.2 Comparison Between the Integral and the Decoupled Impact Simulations 357

(a)

t =0.05s t =0.10s t =0.15s t =0.20s

(b)

t =0.05s t =0.10s t =0.15s t =0.20s

Fig. 9.4 Axial force of circumferential outer layer tendon a missile-target interaction method
b loading-time history method (unit: N)

Fig. 9.5 Axial force of longitudinal tendon at 0.20 s a missile-target interaction method
b loading-time history method (unit: N)

9.2.2.4 Effective Plastic Strain of Rebars

Figure 9.6 shows the effective plastic strain contours of the outer layer rebar
obtained by the integral and decoupled method, respectively. Due to the greater
deformation predicted by the missile-target interaction simulation, the effective
plastic strain in the rebars is correspondingly larger than that by the loading-time
history method.
358 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

(a)

t =0.05s t =0.10s t =0.15s t =0.20s

(b)

t =0.05s t =0.10s t =0.15s t =0.20s

Fig. 9.6 Effective plastic strains of the outer layer rebars a missile-target interaction method
b loading-time history method

9.2.3 Containment Displacement

Figure 9.7 shows the displacements in the impact center (along the impact direc-
tion) obtained by the missile-target interaction method and loading-time history
method, respectively. It indicates that (i) the initial displacement before impact
reaches about 15 mm due to the tension of the prestressed tendons; (ii) after about
0.07 s, the displacement rapidly rises as the increasing of impact force; (iii) the
maximum displacement in the impact center rises up to about 860 mm by
missile-target interaction method and only about 160 mm by the loading-time

Fig. 9.7 Displacement-time 900

histories in the impact center Missile-target
800
by the integral and decoupled interaction method
methods 700
Loading-time
Displacement (mm)

600 history method

500

400

300

200

100

0
0.00 0.05 0.10 0.15 0.20
Time (s)
9.2 Comparison Between the Integral and the Decoupled Impact Simulations 359

Fig. 9.8 Maximum 900

displacements along the Loading-time Missile-target
800 history method interaction method
circumferential direction
700

Displacement (mm)
600

500

400

300

200

100

0
-15 -10 -5 0 5 10 15
Distance from impact center (m)

history method; and (iv) afterward, the displacements slightly recover as the
reduction of impact load and the tension effect of prestressed tendons.
Figures 9.8 and 9.9 illustrate the maximum displacements along the circum-
ferential and longitudinal directions of the containment obtained by the
missile-target interaction method and loading-time history method, respectively. It
indicates that the maximum displacement appears in the impact center and gradu-
ally decreases along with the distance away from the impact point. When the
distance reaches about 12 and 20 m from the impact center at the circumferential
and longitudinal direction, respectively, the maximum impact displacements reduce
to about 50 mm.
According to the above analysis, obvious differences between the results pre-
dicted by the two methods are observed. For example, there are about five times

Fig. 9.9 Maximum 900

displacements along the
800 Missile-target
longitudinal direction
interaction method
700
Loading-time
Displacement (mm)

600 history method

500

400

300

200

100

0
0 10 20 30 40 50
Height of the containment (m)
360 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

differences between the maximum displacements in the impact center. As Siefert

and Henkel (2014) stated, the above differences are mainly caused by the two
reasons. The first one is that the energetic fluxes within the airplane are not con-
sidered in the decoupled method. The second one is that the loading-time history is
always lumped over the whole predetermined impact area, namely the impact force
is uniformly applied at the impact zone, which do not reflect the local stress con-
centration in actual situation. Therefore, the local damage predicted by
missile-target interaction method would be more serious than which obtained by the
decoupled method. Hence, the missile-target interaction method is more suitable to
analyze the aircraft collision with NPP containments.

9.3 Parametric Analyses

The influential factors on the impact process and containment damage mainly
include two aspects. The one is from the aircraft, such as the aircraft impact
position, velocity, and angle. The other one is from the NPP containment, such as
the wall thickness, rebar ratio, and tendon prestressing force. In this section,
numerical simulations considering different impact scenarios are conducted to study
their effects by the missile-target interaction method.

9.3.1 Influence of Impact Position

The protective ability of different containment positions (such as the cylindrical

barrel body, dome, and their junction) against the aircraft impact might be various
for the different local structure styles and constraint conditions. Numerical simu-
lations of aircraft crashing on seven positions are performed to determine the
weakest impact point based on the criterion of containment deformation.
Figure 9.10 illustrates the schematic diagram of different impact positions, in which
the normal impact velocity is set as 215 m/s.

9.3.1.1 Impact Phenomenon

The final crashing phenomena of an aircraft impacting on the different positions are
shown in Fig. 9.11. The impact phenomena are basically similar and all the con-
tainments are not perforated. Although the impact positions are different, the shapes
of impact zones are all cambered surfaces, and the aircraft failure modes are not
obviously influenced. At about 0.20 s, all the aircrafts are crushed seriously and
only the small rear portions are left. The wings fracture into some pieces and the
engines break away due to great deceleration. Meanwhile, the fuel tanks are
destroyed and the aviation fuel spreads widely.
9.3 Parametric Analyses 361

P7
P6
55m
P5
45m
P4
40m

P3
30m

P2
20m

10m P1

0m P:Position

Fig. 9.10 The schematic diagram of different impact positions

Fig. 9.11 Impact phenomena of A320 aircraft crash on the different positions at 0.20 s a P1 b P2
c P3 d P4 e P5 f P6 g P7
362 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

9.3.1.2 Containment Damage

Figure 9.12 shows the corresponding containment concrete damages under the
different impact positions at 0.20 s. The damage shapes and areas in the positions
P1–P4 are almost the same. The damage areas in the positions P5 and P6 become
relatively smaller. As to the position P7, its damage degree is more serious.

9.3.1.3 Impact Force and Impulse

Figures 9.13 and 9.14 show the impact force-time histories and impulse-time his-
tories under the different impact positions, respectively. Almost the same results are
observed, which lies in that, in these seven scenarios, the impact forces and
impulses are mainly determined by the initial collision parameters of the aircrafts
and not influenced by the impact positions.

9.3.1.4 Containment Displacement

As shown in Fig. 9.15, the displacement-time histories in the impact center under
the different impact positions are obviously different. It indicates that (i) the impact
Position 1 is strongly strengthened by the containment foundation and its maximum

Fig. 9.12 Containment damages under the different impact positions at 0.20 s a P1 b P2 c P3
d P4 e P5 f P6 g P7
9.3 Parametric Analyses 363

Fig. 9.13 Impact force-time 275

histories under the different 250 Position 1
impact positions Position 2
225
Position 3
200

Impact force (106N)

Position 4
175 Position 5
150 Position 6
Position 7
125
100
75
50
25
0
0.00 0.05 0.10 0.15 0.20
Time (s)

Fig. 9.14 Impact 13

impulse-time histories under 12
Position 1
the different impact positions 11
Position 2
10 Position 3
9 Position 4
Impulse (10 6 N·s)

8 Position 5
7 Position 6
6 Position 7
5
4
3
2
1
0
0.00 0.05 0.10 0.15 0.20
Time (s)

Fig. 9.15 Displacement-time 900

histories under the different Position 1
800
impact positions Position 2
700 Position 3
Displacement (mm)

Position 4
600 Position 5
Position 6
500 Position 7
400

300

200

100

0
0.00 0.05 0.10 0.15 0.20
Time (s)
364 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

Fig. 9.16 Distribution of 900

maximum displacements Position 1
800 Position 2
along the circumferential
Position 3
direction under the different 700
Position 4
impact positions Position 5

Displacement (mm)
600

500

400

300

200

100

0
-15 -10 -5 0 5 10 15
Distance from the impact center (m)

displacement is less than 680 mm; (ii) the foundation constraint effect to the
Position 2 becomes weaker with the increasing of impact height, and the maximum
impact displacement rises to about 795 mm; (iii) the Position 3 locates at about 2/3
height of the containment barrel body and its maximum displacement reaches about
860 mm for the weaker foundation enhancing effort; (iv) although the Position 4 is
further away from the foundation, the dome transverse stiffness effect makes an
obvious contribution to the decrease of displacement, and its maximum displace-
ment is about 780 mm; (v) the Position 5 locates at the junction of the barrel body
and dome where the dome transverse stiffness is greatest, and therefore, the dis-
placement is least (about 480 mm); (vi) the maximum displacement of Position 6
(about 550 mm) is bigger than that of Position 5 for its further distance away from
the junction; and (vii) since the dome diameter is larger than that of the cylindrical
barrel body, the shape around the Position 7 is relatively more flat and its stiffness
to resist the deformation is smaller, the maximum displacement of the Position 7 is
about 885 mm. However, this impact scenario of the Position 7 is nearly impossible
in the actual accident. By comparing with the maximum displacement, the dome top
is the weakest position of the whole containment to resist the normal crash of the
aircraft. As to the containment barrel body, its 2/3 height (Position 3) is the weakest
position.
The maximum deformations along circumferential and longitudinal directions
under the different positions (from Position 1 to Position 5) are shown in Figs. 9.16
and 9.17, respectively. It is concluded that the displacement shapes are alike,
namely the maximum displacement appears in the impact center and gradually
decays with the distance furthering from the impact point. Besides, when the dis-
tance is further than 10 m, the displacements are already rather small comparing
with the maximum displacements.
9.3 Parametric Analyses 365

Fig. 9.17 Distribution of 900

Position 1 Position 4
maximum displacements
800 Position 2 Position 5
along the longitudinal Position 3
direction under the different 700

Displacement (mm)
impact positions
600

500

400

300

200

100

0
0 5 10 15 20 25 30 35 40 45 50 55
Height of the containment (m)

9.3.2 Influence of Impact Velocity

Considering the aircraft takeoff and landing accidents, as well as the deliberate
collisions by terrorism attacks, the aircraft impact velocity ranging from 100
(takeoff and landing velocity) to 280 m/s (maximum cruising speed of the A320
aircraft) is discussed in this section. Since most of the aircraft initial kinetic energy
would be transferred to the containment, the aircraft impact velocity would greatly
influence the local damage and global response of the containment. The integral
crash simulations of the A320 aircraft collision with the containment with six
different impact velocities are performed. The weakest position in the containment
barrel body (Position 3) is chosen as the impact point. Figure 9.18 illustrates the
schematic diagram of collision with the containment under the different normal
impact velocities.

Fig. 9.18 Schematic diagram

of different impact velocities V: Impact velocity
V1 =100m/s
V2 =120m/s
V3 =150m/s

V4 =215m/s
V5 =250m/s
30m V6 =280m/s
366 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

9.3.2.1 Impact Phenomenon

Figure 9.19 shows the crashing phenomena of the aircraft impacting on the con-
tainment with different velocities at 0.15 s. It indicates that at the same time, the left
lengths of aircrafts become shorter and the wings fracture more seriously with the
impact velocities increasing. Correspondingly, the aviation fuel spreads into larger
spatial region.

9.3.2.2 Containment Damage

Figure 9.20 shows the maximum damage contours of the containment concrete
under the different impact velocities. It can be concluded that, with the crash
velocity increasing, the damage degree becomes more and more serious. Besides,
when the impact velocities are less than about 150 m/s, the damages are relatively
slighter.

Fig. 9.19 A320 aircraft impact phenomena on the containment under the different velocities at
0.15 s a V1 b V2 c V3 d V4 e V5 f V6
9.3 Parametric Analyses 367

Fig. 9.20 Maximum containment damages under the different impact velocities a V1 b V2 c V3
d V4 e V5 f V6

9.3.2.3 Impact Force and Impulse

Figures 9.21 and 9.22 show the impact force-time histories and impact
impulse-time histories corresponding to the different impact velocities, respectively.
It indicates that, with the impact velocity increasing, the peak impact forces increase

Fig. 9.21 Impact force-time 400

280m/s
histories under the different
impact velocities 350
250m/s
300
Impact force (10 6 N)

250 215m/s

200

150
150m/s
100 120m/s
100m/s
50

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Time (s)
368 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

Fig. 9.22 Impact 15

impulse-time histories under 14 Impact velocity
the different impact velocities 13 100m/s 215m/s
12 120m/s 250m/s

Impact impulse (10 6 N·s)

11 150m/s 280m/s
10
9
8
7
6
5
4
3
2
1
0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40
Time (s)

greatly and their maximum values appear earlier, while the overall variation shape
of the impact force-time histories is basically similar. Besides, the impact impulses
rise with the impact velocities increasing, and the durations become shorter at the
same time.

9.3.2.4 Containment Displacement

The displacements in the impact center under the different impact velocities are
shown in Fig. 9.23. As indicated above, the displacements become rather small
when the impact velocity is less than 150 m/s. The maximum displacements along
the circumferential and longitudinal directions under the different impact velocities
are shown in Figs. 9.24 and 9.25, respectively. It can be concluded that the
deformation zone becomes larger with greater crash speed.

Fig. 9.23 Displacement-time 1600

histories under the different Impact velocity
1400
impact velocities 100m/s
1200 120m/s
Displacement (mm)

150m/s
1000 215m/s
250m/s
800 280m/s

600

400

200

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Time (s)
9.3 Parametric Analyses 369

Fig. 9.24 Containment 1600

displacements along the Impact velocity
1400 100m/s
circumferential direction 120m/s
under the different velocities 1200 150m/s

Displacement (mm)
215m/s
1000 250m/s
280m/s
800

600

400

200

0
-15 -10 -5 0 5 10 15
Distance from the impact center (m)

Fig. 9.25 Containment 1600

displacements along the Impact velocity
1400
longitudinal direction under 100m/s
the different velocities 120m/s
1200
150m/s
Displacement (mm)

215m/s
1000 250m/s
280m/s
800

600

400

200

0
0 5 10 15 20 25 30 35 40 45
Height of the containment (m)

9.3.3 Influence of Impact Angle

Obviously, the maximum aircraft impact force would occur under the normal
collision with the coincidence of the crashing direction and aircraft axis. However,
the normal impact is rather rare in the actual accident. Therefore, the oblique impact
accidents are more likely and need to be elaborately studied.
Riera (1980) proposed the oblique impact force equations by considering the
decomposition effect of crash angles. However, the theoretical analysis method
could not obtain the local damage and global response of the containment. The
existing F-4 fighter impact experiment (Sugano et al. 1993a) belongs to the normal
collision on the target. Furthermore, the loading-time history method is incapable of
considering the impact angles. In this section, the missile-target interaction simu-
lations of the A320 aircraft impact on the containment with six different impact
angles are performed. In addition, the weakest position in the containment barrel
370 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

Fig. 9.26 The schematic A 6 =50°

diagram of different impact A 5 =40°
angles
A 4 =30°
10°
10° A 3 =20°
10° A 2 =10°
10°
10° A 1 =0°

A: Impact angle
30m

body (Position 3) is chosen as the impact point. Figure 9.26 illustrates the sche-
matic diagram of the collisions with different impact angles (the angle between the
aircraft axis and the horizontal level) and the impact velocity is set as 215 m/s.

9.3.3.1 Impact Phenomenon

Figure 9.27 illustrates the crashing phenomena of the aircraft impacting on the
containment with the above-mentioned six different impact angles at 0.20 s,
respectively. It indicates that all the aircrafts are destroyed seriously and the avi-
ation fuel spreads widely. With the impact angles increasing, the direction of air-
craft impact velocity changes more greatly during the collision, and the fuselages
impact on the containment in the lateral direction.

9.3.3.2 Containment Damage

Figure 9.28 shows the final containment concrete damage contours under the dif-
ferent impact angles. It indicates that the damage degree becomes slighter with the
impact angles increasing, since more kinetic energies of the aircrafts would be
decomposed by the containment under the larger oblique angles.

9.3.3.3 Impact Force and Impulse

Figures 9.29 and 9.30 show the impact force-time histories and impulse-time his-
tories corresponding to different impact angles, respectively. With the impact angle
9.3 Parametric Analyses 371

Fig. 9.27 A320 aircraft impact phenomena on the containment with different angles at 0.20 s a A1
b A2 c A3 d A4 e A5 f A6

increasing, the decomposition effect becomes more obvious, which leads to the
gradual descending of the peak impact force and impulse. Besides, when the impact
angle is larger than 30°, the influence of every 10° interval on the peak impact force
and impulse becomes more and more obviously.

9.3.3.4 Containment Displacement

The displacements in the impact center under the different impact angles are shown
in Fig. 9.31. As expected, the displacement induced by the normal crash is larger
than that induced by the non-normal crash. The maximum displacements along the
circumferential and longitudinal directions under the different impact angles are
shown in Figs. 9.32 and 9.33, respectively. It could be concluded that the defor-
mation zones are similar, and the magnitudes of the displacements increase with the
decreasing of oblique angles. Besides, the height of the maximum displacement
point along the longitudinal direction slightly reduces due to the decomposition
effect.
372 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

Fig. 9.28 Maximum containment damages under the different impact angles at 0.20 s a A1 b A2
c A3 d A4 e A5 f A6

9.3.4 Influence of Containment Wall Thickness

The containment wall thickness is an important aspect to resist the aircraft crash and
other external impact loads. The protective performance of the containment would
be improved with the wall thickness increasing. However, considering the

Fig. 9.29 Impact force-time 275

histories under the different 250 Impact angel
impact angles 225 0
10
200
Impact force (106N)

20
175 30
150 40
50
125
100
75
50
25
0
0.00 0.05 0.10 0.15 0.20
Time (s)
9.3 Parametric Analyses 373

Fig. 9.30 Impact 13

impulse-time histories under 12
Impact angle
the different impact angles 11 0

Impact impulse (10 6 N·s)

10 10
9 20
8 30
7 40
6 50
5
4
3
2
1
0
0.00 0.05 0.10 0.15 0.20
Time (s)

Fig. 9.31 Displacement-time 900

histories under the different Impact angle
800
impact angles 0
700 10
20
Displacement (mm)

600 30
500 40
50
400

300

200

100

0
0.00 0.05 0.10 0.15 0.20
Time (s)

Fig. 9.32 Displacements 900

along the circumferential Impact angle
800 0
direction under the different
angles 700 10
20
Displacement (mm)

600 30
40
500 50
400

300

200

100

0
-15 -10 -5 0 5 10 15
Distance from the impact center (m)
374 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

Fig. 9.33 Displacements 900

along the longitudinal Impact angel
800
direction under the different 0
angles 700 10

Displacement (mm)
20
600 30
500 40
50
400

300

200

100

0
0 5 10 15 20 25 30 35 40 45
Height of the containment (m)

construction cost and technology, as well as the concrete workability, the wall
thickness should be limited to a reasonable range.
In this section, the missile-target interaction simulations of the A320 aircraft
impacting on the containment with six different wall thicknesses (0.6, 0.9, 1.0, 1.2,
1.5 and 1.8 m) are performed. The weakest position of the containment barrel body
(Position 3) is chosen as the impact point, and the normal impact velocity is also
215 m/s. The impact phenomena under the different containment wall thicknesses
are similar with the typical impact phenomena shown in Fig. 9.1, which are not
given again here.

9.3.4.1 Containment Damage

Figure 9.34 shows the final containment concrete damage contours under the dif-
ferent wall thicknesses. It indicates that, with the increasing of wall thickness, the
damage degree and the local damage area decrease greatly.

9.3.4.2 Impact Force and Impulse

Figures 9.35 and 9.36 show the impact force-time histories and impulse-time his-
tories corresponding to the different wall thicknesses, respectively. With the wall
thickness increasing, the containment stiffness becomes greater. Therefore, the
deformation descends while the peak impact force and impulse slightly rise
up. Generally, the impact force-time histories and impulse-time histories are basi-
cally similar, since they are mainly determined by the initial collision parameters of
the aircrafts.
9.3 Parametric Analyses 375

Fig. 9.34 Maximum concrete damages under the different wall thicknesses at 0.20 s a 0.6 m
b 0.9 m c 1.0 m d 1.2 m e 1.5 m f 1.8 m

9.3.4.3 Containment Displacement

The displacements in the impact center under the different wall thicknesses are
shown in Fig. 9.37. As expected, the displacement rapidly increases with the
reduction of wall thickness. The maximum displacements along the circumferential
and longitudinal directions under the different wall thicknesses are shown in

Fig. 9.35 Impact force-time 300

histories under the different Wall thickness
wall thicknesses 250 0.6m
0.9m
1.0m
Impct force (10 6 N)

200 1.2m
1.5m
1.8m
150

100

50

0
0.00 0.05 0.10 0.15 0.20
Time (s)
376 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

Fig. 9.36 Impact 13

impulse-time histories under 12 Wall thickness
the different wall thicknesses 11 0.6m
0.9m

Impact impulse (10 6 N·s)

10
1.0m
9
1.2m
8 1.5m
7 1.8m
6
5
4
3
2
1
0
0.00 0.05 0.10 0.15 0.20
Time (s)

Fig. 9.37 Displacement-time 2250

Displacement of impact center (mm)

histories under the different 2000

wall thicknesses Wall thickness
1750 0.6m
0.9m
1500
1.0m
1250 1.2m
1.5m
1000 1.8m
750

500

250

0
0.00 0.05 0.10 0.15 0.20 0.25
Time (s)

Figs. 9.38 and 9.39, respectively. In particular, when the wall thickness is larger
than 0.9 m, the deformation zones are similar. When the thickness is 0.6 m, the
maximum impact displacement reaches about 2.2 m, which may induce severe
damages (even collapse) of the containment, and the containment could not resist
the aircraft normal impact at the velocity of 215 m/s.

9.3.5 Influence of Rebar Ratio

The rebars and tendons are installed in the NPP containment to improve its tensile
properties (Lee et al. 2013); the actual containment rebar arrangement is shown in
Fig. 8.14, and the rebar ratio is about 0.32%. In this section, parametric studies
using the missile-target interaction method are conducted by changing the rebar
9.3 Parametric Analyses 377

Fig. 9.38 Containment 2250

Wall thickness
displacements along the 0.6m
2000
circumferential direction 0.9m

Impact displacement (mm)

under the different wall 1750 1.0m
thicknesses 1.2m
1500 1.5m
1250 1.8m

1000

750

500

250

0
-15 -10 -5 0 5 10 15
Distance from the impact center (m)

Fig. 9.39 Containment 2250

displacements along the Wall thickness
2000 0.6m
longitudinal direction under
Impact displacement (mm)

1750 0.9m
the different wall thicknesses 1.0m
1500 1.2m
1.5m
1250 1.8m
1000
750
500
250
0
0 10 20 30 40 50
Height of the containment (m)

ratio. Six percentages (0, 50, 100, 200, 300, and 400%) of the actual rebar ratio are
applied to both the vertical and horizontal rebars. The impact location and velocity
are in accordance with the above sections (Position 3, 215 m/s). The impact phe-
nomena of collisions with the different rebar ratios are similar with the typical
impact phenomena shown in Fig. 9.1, which are not given again.

9.3.5.1 Containment Damage

Figure 9.40 shows the final containment concrete damage contours under the dif-
ferent percentages of the actual rebar ratio. It indicates that, with the rebar ratio
increasing, the damage degree becomes slighter.
378 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

Fig. 9.40 Maximum concrete damages under the different percentages of actual rebar ratio at
0.20 s a 0% b 50% c 100% d 200% e 300% f 400%

9.3.5.2 Impact Force and Impulse

Figures 9.41 and 9.42 show the impact force-time histories and impulse-time his-
tories under the different percentages of the actual rebar ratio, respectively. It
indicates that the impact force and impulse curves are almost the same and not

Fig. 9.41 Impact force-time 275

histories under the different Rebar ratiopercentage
250
percentages of actual rebar of actual value
225 0%
ratio 50%
200
Impact force (10 6 N)

100%
175 200%
150 300%
400%
125
100
75
50
25
0
0.00 0.05 0.10 0.15 0.20
Time (s)
9.3 Parametric Analyses 379

Fig. 9.42 Impact 13

impulse-time histories under 12 Rebar ratio percentage
the different percentages of 11 of actual value
0%

Impact impulse (10 N·s)

actual rebar ratio 10
9 50%

6
100%
8
200%
7 300%
6 400%
5
4
3
2
1
0
0.00 0.05 0.10 0.15 0.20
Time (s)

obviously influenced by the rebar ratios. The reason also lies in that the impact
force-time histories and impulse-time histories are mainly determined by the initial
collision parameters of the aircrafts.

9.3.5.3 Containment Displacement

The displacements in the impact center under the different percentages of actual
rebar ratio are shown in Fig. 9.43. As expected, the displacement increases with the
reduction of rebar ratio. In particular, when there is no any rebar in the containment,
the displacement is about two times of the value corresponding to the actual rebar
ratio. The maximum displacements along the circumferential and longitudinal
directions under the different percentages of actual rebar ratio are shown in
Figs. 9.44 and 9.45, respectively.

Fig. 9.43 Displacement-time 2000

histories under the different 1800 Reinforcement ratio
percentages of actual rebar percentage of actual value
1600
Impact displacement (mm)

ratio 0%
1400 50%
100%
1200 200%
300%
1000 400%
800
600
400
200
0
0.00 0.05 0.10 0.15 0.20
Time (s)
380 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

Fig. 9.44 Displacements 2000

Reinforcement ratio percentage
along the circumferential 1800 of actual value
direction under the different 0%

Impact displacement (mm)

1600
percentages of actual rebar 50%
ratio 1400 100%
200%
1200 300%
1000 400%

800
600
400
200
0
-15 -10 -5 0 5 10 15
Distance from the impact center (m)

Fig. 9.45 Displacements 2000

along the longitudinal Reinforcement ratio
1800 percentage of actual value
direction under the different
0%
Impact displacement (mm)

percentages of actual rebar 1600

50%
ratio 1400 100%
200%
1200
300%
1000 400%

800
600
400
200
0
0 5 10 15 20 25 30 35 40 45
Height of the containment (m)

When the percentage of actual rebar ratio is less than 50%, the containment
displacement is significantly great and basically could not resist the aircraft normal
crash at the velocity 215 m/s. However, when the percentage is bigger than 200%,
the strengthening effect is not obvious.

9.3.6 Influence of Prestressing Forces in Tendons

In order to improve the tensile resistance of the containment, the tendon is installed
in the containment and tightened to decrease the concrete tensile stress (Lee et al.
2013). The tendon arrangement is shown in Fig. 8.15, and the actual prestressing
forces of each tendon along circumferential and longitudinal directions are 4000
and 8000 KN, respectively. Parametric studies by the missile-target interaction
method are conducted by altering the tendon axial prestressing force. Six
9.3 Parametric Analyses 381

percentages (0, 25, 50, 75, 100, and 125%) of the actual prestressing axial force are
applied to both the vertical and horizontal tendons. The impact point and velocity
are in accordance with the above sections (Position 3, 215 m/s). Likewise, the
impact phenomena of the A320 aircraft collision with the different prestressing
forces in tendons are similar with the typical impact phenomena shown in Fig. 9.1,
which are not given here.

9.3.6.1 Containment Damage

Figure 9.46 shows the maximum containment concrete damage contours under the
different percentages of the actual prestressing force. It indicates that, with the
prestressing force increasing, the damage degree becomes slighter.

9.3.6.2 Impact Force and Impulse

Figures 9.47 and 9.48 show the impact force-time histories and impulse-time his-
tories corresponding to the different percentages of the actual prestressing force in
tendons, respectively. It could be seen that, for the identical collision parameters of

Fig. 9.46 Containment concrete damages under the different percentages of actual prestressing
force at 0.20 s a 0% b 25% c 50% d 75% e 100% f 125%
382 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

Fig. 9.47 Impact force-time 275

histories under the different 250 Tendon prestressing
percentages of actual percentage of the actual value
225
prestressing force 0%
200 25%

Impact force (10 N)

6
175 50%
75%
150
100%
125 125%
100
75
50
25
0
0.00 0.05 0.10 0.15 0.20
Time (s)

Fig. 9.48 Impact 13

impulse-time histories under 12 Tendon prestressing
the different percentages of 11 percentage of actual value
0%
Impact impulse (10 N·s)

actual prestressing force 10

25%
9
6

50%
8 75%
7 100%
6 125%
5
4
3
2
1
0
0.00 0.05 0.10 0.15 0.20
Time (s)

the aircraft, the impact force and impulse curves are almost the same and basically
not influenced by the variation of the prestressing force.

9.3.6.3 Containment Displacement

The displacements in the impact center under the different percentages of actual
prestressing force are shown in Fig. 9.49. As expected, the impact displacement
increases with the reduction of the prestressing force. The maximum displacements
along the circumferential and longitudinal directions under the different percentages
of actual prestressing force are shown in Figs. 9.50 and 9.51, respectively. It could
be found that when the percentage of actual prestressing force is less than 25%, the
displacement is significantly great and basically could not resist the aircraft normal
crash at the velocity of 215 m/s. However, when the percentage is bigger than 50%,
the strengthening effect of the prestressing force is not obvious.
9.3 Parametric Analyses 383

Fig. 9.49 Displacement-time 2500

Tendon prestressing
histories under the different

Displacement of impact center (mm)

percentage of actual value
percentages of actual 0%
prestressing force 2000
25%
50%
75%
1500 100%
125%

1000

500

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Time (s)

Fig. 9.50 Displacements 2500

Tendon prestressing
along the circumferential
percentage of
direction under the different actual value
Impact displacement (mm)

percentages of actual 2000

0%
prestressing force 25%
50%
1500 75%
100%
125%
1000

500

0
-15 -10 -5 0 5 10 15
Distance from the imapct center (m)

Fig. 9.51 Displacements 2500

along the longitudinal Tendon prestressing
direction under the different percentage of actual value
Impact displacement (mm)

2000 0%
percentages of actual 25%
prestressing force 50%
1500 75%
100%
125%
1000

500

0
0 5 10 15 20 25 30 35 40 45
Height of the containment (m)
384 9 Numerical Simulation of A320 Aircraft Impact on NPP Containments

9.4 Summary

This chapter can be summarized as follows:

(1) In Sect. 9.2, the detailed numerical simulations of the collisions are performed
by using the missile-target interaction (integral) and loading-time history
method (decoupled), respectively. It derives that the predicted results of
loading-time history method are significantly inclined to a small damage
extent. The missile-target interaction method is relatively more reliable and
reasonable.
(2) In Sect. 9.3, the parametric analyses are performed, including the impact
position, velocity, and angle of the aircraft, as well as the wall thickness, rebar
ratio, and tendon force of the containment. It derives that (i) the impact force
and impulse are obviously influenced by the aircraft impact velocity and
impact angle, while the effects of the impact position, wall thickness, rebar
ratio, and tendon force are not significant; (ii) the containment damage and
deformation are deeply influenced by all the considered influential factors;
(iii) the weakest position of the barrel body to resist the external aircraft impact
is confirmed at about its 2/3 height; and (iv) when the wall thickness is less
than 0.9 m, and the rebar ratio or the tendon force is less than 50% of their
actual values, the containment would suffer rather significantly serious
damages.
Chapter 10
Aircraft Engine Impact
on UHP-SFRC Slabs

10.1 Introduction

In the process of aircraft collision, the massive engines would quite likely detach
from the aircraft wings due to the great deceleration and lead to the deepest local
penetration among all aircraft components (Riera et al. 1982). Hence, the impact
damage caused by the aircraft engine missile needs to be investigated separately.
Sugano et al. (1993b, c) conducted systematic experiments of prototype and
scaled aircraft engine impacting on concrete panels, in which the designed com-
pressive strength of concrete ranged from 23.5 to 35.3 MPa. Based on the three sets
of impact tests, i.e., small-, intermediate-, and full-scaled tests, not only the detailed
local damages of the concrete panel caused by the aircraft engine missile were
studied, but also the similarity law for the aircraft engine was validated.
Furthermore, an F-4 Phantom fighter impact test was also carried out by Sugano
et al. (1993a), which showed that the corresponding engine impacted zone suffered
larger penetration depth with a maximum value of about 60 mm, while only about
20 mm induced by the impact of the fuselage was observed. In addition, in order to
provide test data to calibrate and validate computer models, Lawver et al.
(2002) performed a TF-30 engine impacting test on concrete panels with the
velocity of about 107 m/s, in which the thickness and compressive strength of the
concrete panels were about 0.61 m and 33.3 MPa, respectively. In the above engine
impact experiments, all the targets were casted with normal strength concrete
(NSC), which could hardly meet the ever-increasing protection requirements.
Aiming to improve the resistance of protective structures, extensive develop-
ments have been achieved in the concrete material technology, for example, the
ultra-high-performance steel fiber-reinforced concrete (UHP-SFRC), in which the
incorporating fibers could bridge the cracks by the bonding force between fibers
and concrete matrix (Brandt 2008). UHP-SFRC has high compressive strength of
up to 200 MPa and tensile strength of about 20–40 MPa. The fracture energy
of UHP-SFRC is around 20,000–40,000 J/m2, which is nearly two orders of

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 385
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_10
386 10 Aircraft Engine Impact on UHP-SFRC Slabs

magnitude higher than that of NSC (Barnett et al. 2010). Thus, UHP-SFRC is an
excellent and promising material for the construction of important buildings, such
as the NPP containment, to resist the aircraft and its engine impact.
Riedel et al. (2010) conducted a series of engine collision test with UHP-SFRC
panels with various velocities ranging from 194.7 to 368.6 m/s. The engine missile
was 1/10 scaled model according to the work of Sugano et al. (1993b, c), and the
panels were 100 mm thick with the steel fiber volumetric ratio of 1% and the
average compressive strength of about 185 MPa. Siefert and Henkel (2014), Itoh
et al. (2005), and Lee et al. (2013, 2014) numerically studied the aircraft impacting
on the NPP containment, while the engine collision was not studied separately. Thai
and Kim (2015) numerically simulated the engine model impacting test conducted
by Riedel et al. (2010), while the panels were modeled as a homogeneous material
without modeling the discrete fiber elements and the actual fiber effect may not be
reproduced reasonably. Thai and Kim (2016) further numerically discussed the
influences of panel thickness on the damages of UHP-SFRC panels under the
aircraft engine impact. For the simulations of fiber-reinforced concrete under
intense dynamic loadings, the following two numerical approaches are frequently
utilized. The first one assumes the matrix as the homogenous material, and the
influences of fibers are reflected by changing the matrix compressive and tensile
strengths (Rong and Sun 2012; Mahmud et al. 2013; Thai and Kim 2015, 2016) or
adjusting the parameters which control the strain-softening behavior (Teng et al.
2008; Wang et al. 2010b; Mao et al. 2014). The second one simulates the concrete
matrix and discrete fibers separately, and the fiber bridging effect is reproduced
more directly. The mesoscale model established by Xu et al. (2012) was two
dimensional for the spiral fiber-reinforced concrete, which may not reflect the
damage mechanism reasonably. Fang and Zhang (2012, 2013) established a
three-dimensional mesoscale model for steel fiber-reinforced concrete, considering
the random size, orientation, and spatial distribution of the fibers, and a good
agreement between the numerical results and test data is obtained.
In this chapter, considering the aircraft takeoff and landing velocity, as well as
the higher impact velocity during the terrorist attacks, two series of engine collision
experiments with the UHP-SFRC panels with the impact velocities of about 120
and 250 m/s are conducted. The thickness of the UHP-SFRC panels varies from 40
to 80 mm to evaluate their influences on the panel damages. The volumetric ratio of
the mixing steel fiber is designed as 2%, and the straight brass-coated steel fiber is
chosen since it provides a good trade-off between workability and mechanical
properties of matrix (Máca et al. 2014; Wu et al. 2015c; Sovják et al. 2015; Peng
et al. 2016a). The average compressive strength of the UHP-SFRC is 150 MPa.
Then, by using the keyword CONSTRAINED_LAGRANGE_IN_SOLID in
LS-DYNA, the discrete fiber beam elements are combined with the UHPC hexa-
hedron solid elements together, and the fiber bridging effect is reproduced more
directly. In addition, the bonding and sliding effects between the fiber and concrete
matrix are reflected approximately by modifying the fiber material properties
according to the single-fiber pullout test. The deformation and residual velocity of
engine missile, as well as the damage of the UHP-SFRC panel, are numerically
10.1 Introduction 387

predicted and compared with the test results. Furthermore, the parametric influential
analyses of panel thickness and engine missile impact velocity are discussed.
Finally, a modified empirical formula for predicting the engine missile residual
velocity to guide the practical protective structure design is proposed.

10.2 Test Setup

The experimental objectives are to investigate the resistance and damage modes of
UHP-SFRC panels subjected to the soft engine missile impact, and provide
experimental data to calibrate the proposed numerical modeling approach.

10.2.1 Aircraft Engine Model

Sugano et al. (1993b, c) performed three sets of different scaled impact tests, i.e.,
1/7.5, 1/2.5, and full scale, in which the similarity law and the method of simpli-
fying the aircraft engine were verified. The 1/10 scaled model with the mass of
1.5 kg is adopted in this test. The full-scaled and 1/10 scaled engine models are
shown in Fig. 10.1, respectively. In order to distinguish the engine missile more
clearly in the pictures from high-speed videos, the surface of the engine model is
sprayed with five circumferential red strips, and the location and width of three

(a)
760

75 1270 90 830 113

7.5 20
2378mm
(b)
Velo
Front city
slice
7.5 Middle
127 slice
5 9 Rear
0.8 83 slice
11.3
76

238mm 2 6.3

Fig. 10.1 Dimensions of engine model a full scale (Sugano et al. 1993b) b 1/10 scale
388 10 Aircraft Engine Impact on UHP-SFRC Slabs

Table 10.1 Material properties of engine model

Material parameter Present experiment Sugano et al. (1993b) experiment
Yield strength fy (MPa) 355 349.1
Ultimate strength fu (MPa) 600 687.4
Young’s modulus E (GPa) 210 201

relatively narrow strips are corresponding to the three lumped mass slices, as shown
in Fig. 10.1b.
As shown in Fig. 10.1b, the steel pipe of the engine missile with the wall
thicknesses of 0.8 and 2 mm is machined integrally from the seamless steel pipe.
The three lumped mass slices, which are connected to the pipe by welding around
their edges, are sliced from the steel rod. The material of the engine model is the
Chinese standard 45# steel with the main properties listed in Table 10.1, and the
corresponding parameters adopted by Sugano et al. (1993b) in the full-scaled tests
are also given for comparison.
The barrel caliber of the air gas gun used in the experiments is 250 mm and
larger than the engine diameter; thus, the sub-caliber technology is utilized in the
test. The engine model is fitted with a specially designed aluminum sabot and oak
obturator that fit snugly into the gun bore, as shown in Fig. 10.2. The sabot is
mainly comprised of six flabellate slices and six rods. In order to separate the sabot
after flying out of the muzzle, three springs are set between the upper three slices.

10.2.2 UHP-SFRC Slabs

The width of the casted UHP-SFRC panel is 700 mm, which is also 1/10 of the
panel used in the experiment by Sugano et al. (1993b). In order to investigate
the critical perforation thickness under the engine impact, the thicknesses of the
panels vary from 40 to 80 mm, and two panels are casted for each thickness.

Engine front slice Engine

model

Aluminum Springs
sabot
Steel
rods
Sabot slices
Oak
obturator

Fig. 10.2 Sabot of engine missile

10.2 Test Setup 389

The reinforcing mesh of the panels is ignored for the following three reasons:
(a) Sugano et al. (1993b, c) experimentally concluded that the reinforcement ratio
nearly has no effect on the local damage of the panels without perforation; (b) ig-
noring the rebar effect is more conservative for the design of protective structures;
(c) the panels are too thin to place the rebar, especially for the 40-mm-thick panel.

10.2.2.1 Material Compositions

The mixing proportions of the UHP-SFRC are listed in Table 10.2, in which all the
ingredients are commercially available in China. The volumetric ratio of the mixing
fiber is designed as 2%, and the straight brass-coated steel fiber is chosen since it
provides a good trade-off between workability and mechanical properties of con-
crete. The equivalent diameter, length, and tensile strength of the steel fibers are
0.2 mm, 13 mm, and 2800 MPa, respectively. The water-reducing ratio of the
PCA®-I polycarboxylic type HRWR (high-range water reducer) is not less than
35%, which is developed by Jiangsu SOBUTE new material Co., Ltd.

10.2.2.2 Mixing Procedure

To achieve good workability, particle distribution, and packing density, the mixing
procedure for the UHP-SFRC must be controlled rigorously. The detailed mixing
procedure of UHP-SFRC will be introduced in Sect. 11.3.2.
The prepared matrix is poured into 700 mm
700 mm
(40–80) mm wooden
molds and then vibrated on the shaking table. The freshly casted panels are firstly
kept in the molds for 24 h, and then they are demounted and cured in a standard
climate with temperature of 17.7–21.7 °C and relative humidity of 98.4% for
28 days. Afterward, the specimens are taken out of the curing room and cured
atmospherically. The impact test is conducted at 84d after the specimens are casted,
and the specimen density and average compressive strength are about 2480 kg/m3
and 150 MPa, respectively.
Aiming to obtain the compressive and tensile strengths of the UHP-SFRC with
2% steel fibers as well as plain UHPC without fibers, the uniaxial compressive and
tensile tests are conducted, as shown in Figs. 10.3 and 10.4, respectively. The
uniaxial tensile curves are shown in Fig. 10.5, which indicate that the average tensile
strengths for the UHP-SFRC and UHPC are 7.2 and 6.5 MPa, respectively. In
addition, the UHP-SFRC specimen remains definite tensile resistance to some extent
owing to the fiber bridging effect even after the matrix fracturing. The average

Table 10.2 Mixtures of UHP-SFRC (kg/m3)

Cement Silica fume Fine sand Admixture Water HRWR Steel fibers
700 140 1200 110 152 22.8 156
390 10 Aircraft Engine Impact on UHP-SFRC Slabs

(b)

(a)
100mm 100mm

100mm

Fig. 10.3 Uniaxial compressive test a specimen b SUNS YAW-5000 system

(a) (b)
100
100
100
500mm

50

Fig. 10.4 Uniaxial tensile test a specimen b INSTRON-8803 system

unconfined compressive strengths (100 mm
100 mm
100 mm cube) for the

UHP-SFRC and UHPC are measured as 150 and 110 MPa, respectively.

10.2.3 Setup

As shown in Fig. 10.6, an air gas gun with 250 mm caliber is used to accelerate the
engine missile. A laser velocimeter is set at the muzzle to measure the missile
10.2 Test Setup 391

(a) (b)
8 7
7 6
6
5

Stress (MPa)
5
Stress (MPa)

4
4
3
3
2
2

1 1

0 0
0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6
Strain Strain (×10-4 )

Fig. 10.5 Uniaxial tensile stress–strain curves a UHP-SFRC with 2.0% fibers b plain UHPC

(a)
Air vessel Air gas gun Laser velocimeter Sabot baffle Panel
Target
Missile frame
Sabot
1.2m 1.2m High
speed
camera

(b)
Laser Target
Air gas gun velocimeter frame
Panel Panel
Sabot
baffle

High speed
camera

Fig. 10.6 Experimental setup a diagram b photographs

velocity. The impact process and missile deformation are recorded by a high-speed
camera, which is placed at the side of the panel. In addition, the initial and
residual/rebound velocities of the engine missile could be obtained approximately
from the high-speed camera. The panels are fixed to a steel frame.
The engine missile with the sabot and obturator is firstly located in the barrel
properly, and the air is then pushed into the sealed vessel by a compressor until to a
designed pressure. Afterward, the valve of air vessel is opened and the engine
missile is accelerated forward by the released compressed air. When the engine
missile reaches the muzzle, two laser beams are interrupted successively to measure
its velocity. Once the engine missile comes out from the muzzle, the three springs
as shown in Fig. 10.2 will bounce off the front sabots and the engine missile
continues to fly freely. Aiming to prevent the sabot impacting on the panel, a steel
392 10 Aircraft Engine Impact on UHP-SFRC Slabs

sabot baffle is designed and fixed on the ground. A hole with the diameter of
150 mm is cut out from the baffle and faces the gun barrel, which can allow the
engine missile to pass freely. Thus, the sabot is stopped by the steel baffle and the
engine missile will impact on the panel. The whole collision process is recorded by
the high-speed camera located aside the panel.

10.3 Test Results

Aiming to analyze the influences of different velocities and panel thicknesses on the
damages of UHP-SFRC panels and engine missiles during the collisions, two series
of experiments with the impact velocities of about 250 and 120 m/s are conducted
with the panel thickness varying from 40 to 80 mm, respectively. The related
experimental scenarios are listed in Table 10.3. According to the experimental
results, some typical shots, i.e., Shots 4-1, 6-1, 2-2, 3-2, and 5-2, are selected to
represent the following typical panel damage modes: perforation, just perforation,
just scabbing, and penetration, which are explained in Sect. 10.3.4.

10.3.1 Impact Process

Figure 10.7 illustrates the impact processes of Shots 4-1, 6-1, 2-2, 3-2, and 5-2
recorded by the high-speed camera, respectively. In Shot 4-1 (Fig. 10.7a) and Shot
6-1 (Fig. 10.7b), when the aluminum sabots impact on the steel baffle with the
velocity of about 250 m/s, plenty of sparks generate for the high-speed collision
and friction between sabot and baffle. However, in Shot 2-2 (Fig. 10.7c), Shot 3-2
(Fig. 10.7d), and Shot 5-2 (Fig. 10.7e), the sparks do not appear due to the rela-
tively low impact velocity of about 120 m/s.

Table 10.3 Experimental scenarios

Test series Shot no. Engine velocity (m/s) Panel thickness (mm)
1 1-1 222 40
2-1 254 50
3-1 244 60
4-1 253 65
5-1 254 70
6-1 262 80
2 1-2 120 40
2-2 139 50
3-2 119 60
4-2 126 65
5-2 121 70
6-2 128 80
10.3 Test Results 393

Generally, during the impact process, when the first slice of the engine missile
impacts on the panel, some small pieces of the UHPC debris would spall from the
front surface. Then, the steel pipe with the thickness of 0.8 mm may crush and
absorb partial kinetic energy to decelerate the remaining engine missile.
Subsequently, the second slice and the 2-mm-thick steel pipe would collide with the
panel, leading to the different damage modes of the panels based on different impact
velocities and panel thicknesses. For the perforation mode, plenty of the UHPC
debris would eject away from the rear side, while for the penetration damage mode,
only some radial minor cracks appear.

10.3.2 Deformation of Engine Missiles

After the impact, the major residual parts of the engine missile in Shots 4-1, 6-1,
2-2, 3-2, and 5-2 are collected, as shown in Fig. 10.8. As indicated, the front
0.8-mm-thick steel pipes are all crushed and broken into small fragments. Under the
relatively high impact velocity of about 250 m/s, the rear 2.0-mm-thick steel pipes
are buckled and torn at the edge (Fig. 10.8a, b). While for the relatively low impact

Fig. 10.7 Typical impact processes of shots a 4-1 b 6-1 c 2-2 d 3-2 e 5-2
394 10 Aircraft Engine Impact on UHP-SFRC Slabs

(a) (b)

70mm Tail 60mm Tail

(c) (d) (e)

115mm Tail
118mm Tail
111mm Tail

Fig. 10.8 Engine missile deformations in shots a 4-1 b 6-1 c 2-2 d 3-2 e 5-2

velocity of about 120 m/s, the 2.0-mm-thick steel pipes are remained intact without
obvious deformation (Fig. 10.8c–e). However, the rear slice, which is welded on
the 2.0-mm-thick steel pipe, breaks away because of the tensile stress wave and
deformation of the engine missile.

10.3.3 Damage of UHP-SFRC Slabs

Figure 10.9 illustrates the damages of the UHP-SFRC panels in Shots 4-1, 6-1, 2-2,
3-2, and 5-2, in which the “F” and “R” written on the label pasted on the panels
denote the front and rear surfaces of the panels, respectively.
For Shot 4-1 (Fig. 10.9a) and Shot 6-1 (Fig. 10.9b), the panel damages are both
perforation modes with obvious radial cracks on the rear surface. The just perfo-
ration damage mode occurs in Shot 2-2 (Fig. 10.9c) with the impact velocity of
139 m/s and the panel thickness of 50 mm, in which a large piece of scabbed
concrete on the rear side remains connecting to the whole panel by steel fibers. The
damage mode of Shot 3-2 (Fig. 10.9d) belongs to the just scabbing mode, where
only some small debris are peeled off from the rear surface with an obvious bulge.
The Shot 5-2 (Fig. 10.9e) represents the penetration damage mode, in which only a
small crater is formed on the front surface and some tiny cracks appear on the rear
face.
10.3 Test Results 395

Fig. 10.9 Panel damages in shots a 4-1 b 6-1 c 2-2 d 3-2 e 5-2
396 10 Aircraft Engine Impact on UHP-SFRC Slabs

10.3.4 Test Data

The experimental data is summarized in Table 10.4, and some letters are used to
stand for different parameters, viz. H for panel thickness, M for missile mass, V0 and
Vr for the initial and residual velocities of engine missile, a and Lr for the oblique
angle and residual length of engine missile, respectively.
In addition, according to the different damage modes (DM) of the panel, the
following terminologies defined by Sugano et al. (1993b) are adopted here: per-
foration mode (P) indicates that the missile passes through the panel completely;
just perforation mode (JP) indicates that the missile does not pass through the panel
and a large opening occurs at the rear surface; scabbing mode (S) indicates that
considerable concrete debris is peeled off from the rear surface of the panel and a
small opening appears; just scabbing mode (JS) indicates that few small pieces of
concrete debris are peeled off and the shear cone cracks form on the rear surface of
the panel; and penetration mode (C) indicates that a crater forms on the front surface
of the panel, and no scabbing occurs on the rear surface.
It should be pointed out that, for Shot 1-1, the engine missile is rubbed by the
sabot when it flies through the hole on the baffle, which leads to the impact velocity
decreasing from 241 to 222 m/s and a larger oblique angle of 30°, as shown in
Fig. 10.10.
In Table 10.5, the cratering damages at the front and rear faces of the
UHP-SFRC panels are listed, in which the damage areas are obtained by image
pixels method illustrated in Werner et al. (2013). For Shot 2-2, the panel is just
perforated while the engine missile falls directly in front of the panel with nearly no
rebound.

Table 10.4 Experimental data

Test Shot H M (g) V0 Vr DM a Lr
series no. (mm) (m/s) (m/s) (°) (mm)
1 1-1 40 1475 222 108 P 30 95
2-1 50 1459 254 84 P −10 93
3-1 60 1476 244 97 P 2 75
4-1 65 1426 253 82 P 2 70
5-1 70 1463 254 76 P 2 74
6-1 80 1460 262 46 P 2 60
2 1-2 40 1460 120 21 P 2 192
2-2 50 1457 139 −2 JP 2 118
3-2 60 1454 119 −4 JS −1 115
4-2 65 1449 126 −5 C 8 106
5-2 70 1451 121 −4 C 3 111
6-2 80 1442 128 −6 C 10 95
10.3 Test Results 397

Fig. 10.10 Oblique impact

of engine missile in shot 1-1

40mm
30° Panel

Table 10.5 Damages of panels

Test Shot Front crater Rear crater DM
series no. Width Height Area Depth Width Height Area
(mm) (mm) (m2) (mm) (mm) (mm) (m2)
1 1-1 109 129 0.011 – 200 186 0.028 P
2-1 133 146 0.013 – 298 215 0.047 P
3-1 113 120 0.010 – 274 257 0.048 P
4-1 100 107 0.008 – 198 197 0.030 P
5-1 105 104 0.009 – 201 219 0.034 P
6-1 101 109 0.013 – 216 225 0.037 P
2 1-2 124 98 0.009 – 174 187 0.027 P
2-2 110 97 0.009 – 233 235 0.047 JP
3-2 83 81 0.005 5.0 93 101 0.006 JS
4-2 90 97 0.008 9.5 – – – C
5-2 80 86 0.006 4.0 – – – C
6-2 94 103 0.008 9.5 – – – C

Overall, under the relatively high velocity of about 250 m/s, all the panels in
Test 1 are perforated and their residual velocities gradually decrease with the panel
thickness increasing from 40 to 80 mm. For Test 2, the impact velocity is about
120 m/s; thus, only the panel with 40 mm thickness is completely perforated and
the engine missile passes through the panel. The 50-mm-thick panel is just perfo-
rated, and the engine missile rebounds back and falls in front of the panel. The
damage mode of the 60-mm-thick panel belongs to the just scabbing mode with
little debris peeling off its rear side, as presented in Sect. 10.3.3. For the 65, 70,
80 mm panels, the velocity of about 120 m/s only leads to the penetration mode
and the engine missiles rebound back.
398 10 Aircraft Engine Impact on UHP-SFRC Slabs

10.4 Numerical Simulation and Comparison

with the Test Data

In this section, by proposing an approach to model the UHP-SFRC with randomly

distributed steel fibers, the impact process of aircraft engine model collision with
the UHP-SFRC panels is numerically simulated by LS-DYNA and compared with
the existing and above test data.

10.4.1 Finite Element Model

Taking the 40-mm-thick UHP-SFRC panel for instance, the FE model is shown in
Fig. 10.11. The steel pipe and three lumped mass slices of the engine missile are
modeled by 8848 shell elements and 10,560 solid elements with the size of about
2.5 mm, respectively. Total of 640,000 solid elements are used to model the matrix
of the UHP-SFRC (plain UHPC) panel, where the element size is 2.5 mm for the
central impacted zone (300 mm
300 mm) and 5.0 mm for the marginal region,
respectively. Besides, aiming to reduce the calculation time, the steel fibers are only
distributed in the above impacted zone. Regarding the UHP-SFRC as the inho-
mogeneous material, for the geometry of the present straight fiber and 2.0% vol-
umetric ratio, through writing a program with the software MATLAB, nearly
180,000 beam elements are established and embedded randomly and uniformly into
the UHPC matrix to realize the discretely distributed fibers. Each fiber is modeled
with one beam element, and the element size is identical with the actual fiber.
The process of establishing beam elements for modeling steel fibers is described
as follows:
Analyzing the writing format of the keyword ELEMENT_BEAM in LS-DYNA,
it could be found that the keyword primarily consists of five variables, e.g.,
Element ID, Part ID, Node 1, Node 2, and Node 3. The Element ID is the exclusive
number of one element, and the Part ID is a set number for the elements. In the
finite element model, one beam element is determined by two end nodes (Node 1

mm
Fibers 300
40mm
+ t er
cen
300mm

act e
Imp zon
300mm

mm
Fixed 300
support

Fig. 10.11 Finite element model

10.4 Numerical Simulation and Comparison with the Test Data 399

and Node 2) and a third node for orientation (Node 3). It should be pointed that, for
the circular cross section of a beam element, the coordinate of Node 3 is arbitrary,
except for locating on the axis of Node 1 and Node 2. Now, the writing format of
ELEMENT_BEAM is clear, and the software MATLAB can be used to write a
program for establishing a large number of beam elements based on the following
steps:
(i) Specifying the initial conditions, i.e., the length Lf, diameter Df, and vol-
umetric ratio Vf of the fibers, as well as the spatial volume Sv to be filled
with the fibers;
(ii) Calculating the total number of fibers Nf based on the above conditions, i.e.,
Nf = (4
Sv
Vf)/(p
Lf
Df2). Obviously, the number of Node 1,
Node 2, and Node 3 are all Nf, respectively;
(iii) Generating the first set of Nf nodes as the Node 1 uniformly and randomly
distributed in the region Sv using the function of “rand” provided by the
MATLAB software;
(iv) Generating the second set of Nf nodes as the Node 2, which are limited by
the conditions that the distance between each Node 1 and the corresponding
Node 2 must be equal to the length Lf, and the angles between the connected
line of the two nodes and the coordinate axis are also set randomly;
(v) Generating the third set of Nf randomly distributed nodes as the Node 3 to
determine the orientations of the beam elements;
(vi) Designating the Element IDs for all Nf beam elements, e.g., the serial
number of 1 to Nf is a feasible selection and each ID is exclusive;
(vii) Designating an arbitrary integral number as the Part ID for all the beam
elements;
(viii) Following the format of the beam element in LS-DYNA, the
above-generated data is assembled as the order of Element ID, Part ID,
Node 1, Node 2, and Node 3, and then output a file that contains all the
necessary information of the Nf beam elements.
Thus, all the randomly distributed beam elements have been established, while
the fiber diameter Df and material parameters are to be specified by the
SECTION_BEAM and MAT in LS-DYNA, respectively. The bonding and sliding
effects between the fiber and matrix can be approximately modeled which will be
presented in Sect. 10.4.2.3.
The option CONTACT_ERODING_SURFACE_TO_SURFACE is adopted to
realize the contact between the engine missile and concrete panel. In addition, the
interaction between the steel fibers and UHPC matrix is dealt with the keyword
CONSTRAINED_LARGRANGE _IN_SOLID, by which the end nodes of the fiber
elements are fixed in the UHPC solid elements.
400 10 Aircraft Engine Impact on UHP-SFRC Slabs

10.4.2 Material Models

10.4.2.1 Engine Missile

The MAT_SIMPLIFIED_JOHNSON_COOK material model (MAT#098),

including the strain rate effect and failure, is used to model the behavior of 45# steel
for the engine missile, and the detailed parameters are listed in Table 10.6 (Hu et al.
2003). The flow stress ry is expressed as

n
ry ¼ A þ Bep ð1 þ C ln e_  Þ ð10:1Þ

_ e0 is
where A, B, C, and n are input constants, ep is effective plastic strain, e_  ¼ e=_
1
effective strain rate for e_ 0 ¼ 1 s . The element failure strain (FS) is determined as
0.25 by trial and error based on the engine missile deformations.

10.4.2.2 UHPC Panel

The UHPC panel is modeled by MAT_WINFRITH_CONCRETE material model

(MAT#084), which is simple to input parameters, as listed in Table 10.7. The strain
rate effect is automatically considered in the model, and the concrete strength is
calculated by multiplying the original value with the strain rate enhancement factor.
The tensile ðET Þ and compressive ðEC Þ enhancement factors are calculated using
the equations below (Schwer 2010).
If the incremental effective strain rate e_ is less than 30 s−1,
0
ET ¼ ðe_ =_e0T Þ1:016d ; EC ¼ ðe_ =_e0C Þ1:026a ð10:2Þ

and if e_ is greater than 30 s−1,

Table 10.6 Material parameters of MAT#098 for 45# steel

Material Density qp Young’s Poisson’s A (MPa) B (MPa) C n Failure
(kg/m3) modulus ratio v strain
Es (GPa)
45# steel 7800 210 0.3 496 434 0.040 0.307 0.25

Table 10.7 Material parameters of MAT#084 for UHPC

Material Density q0 Young’s Poisson’s fc ft Fracture Aggregate
(kg/m3) modulus E (GPa) ratio v (MPa) (MPa) energy size (mm)
Ef (J/m2)
UHPC 2480 55 0.19 110 6.5 165 2
10.4 Numerical Simulation and Comparison with the Test Data 401

ET ¼ g_e1=3 ; EC ¼ c_e1=3 ð10:3Þ

where d ¼ 1=ð10 þ 0:5fcu Þ, a0 ¼ 1=ð5 þ 0:75fcu Þ, log10 g ¼ 6:933d  0:492,

log10 c ¼ 6:156a0  0:492, e_ 0T ¼ 30
106 s1 , e_ 0C ¼ 3
106 s1 , and fcu is the
concrete cube compressive strength in MPa.
The strain rate enhancement of Young’s modulus EE is calculated as an average
of tensile and compressive strain rate enhancements,
h i
EE ¼ 0:5
ðe_ =_e0T Þ0:016 þ ðe_ =_e0C Þ0:026 ð10:4Þ

If any of the above strain rate enhancement factors are less than one, they are set
equal to one, i.e., no strain rate enhancement.
In addition, the fracture energy of UHP-SFRC is rather great (about 15,000 J/m2
in this test), while the fracture energy of UHPC only reaches about 165 J/m2 in this
test.
The MAT_WINFRITH_CONCRETE material model (MAT#084) could not
consider the element erosion itself; therefore, the keyword MAT_ADD_EROSION
is used to delete the failure elements and reproduce the damage modes. The prin-
cipal strain is selected to be the erosion criterion (Sagals et al. 2011; Thai and Kim
2014). Based on the results of this experiments and previous tests (Riedel et al.
2010), the failure strain is determined as 0.04.

10.4.2.3 Steel Fiber

Aiming to reflect the bridging effect of discrete fibers explicitly, Xu et al.

(2012) modeled the spiral fibers by small shell elements, which were connected
with the matrix elements by joint nodes, i.e., the bonding and sliding effects of
fibers could not be realized. Fang and Zhang (2012, 2013) established the discrete
fibers by beam elements and considered their bonding and sliding effects with the
matrix by the keyword CONTACT_1D in LS-DYNA. Based on the relationship of
axial stress with slip displacement from the single-fiber pullout test, the bonding
and sliding effects of fibers are reflected by the constitutive model of the fibers by
adjusting the material parameters and failure strain. The constitutive model of
MAT_PIECEWISE_LINEAR_ PLASTICITY (MAT#024) is adopted to simulate
the steel fiber.
The fiber tensile strength reaches about 2800 MPa, which implies that almost all
the fibers could not reach their maximum strength even though they are pulled out
from the matrix. For the simulation of single-fiber pullout test, since the two end
nodes of one fiber element are fixed in the UHPC solid elements, the behaviors of
steel fibers in the actual pullout test and simulation are different, as shown in
Fig. 10.12. When the fiber is completely pulled out, in the actual pullout test shown
in Fig. 10.12a, the fiber length nearly remains unchanged (13 mm). While in the
402 10 Aircraft Engine Impact on UHP-SFRC Slabs

Load
(a) (b) Load

Load
13mm Load
6.5mm

6.5mm
13mm 19.5mm

Specimen Specimen UHPC

elements

Fig. 10.12 Fiber behaviors a pullout test b numerical simulation

Fig. 10.13 Relationship of 1200

axial stress in fiber with slip Pullout test (Wille and Naaman, 2012)
displacement 1000 Simulation of the pullout test
Fiber axial stress (MPa)

800

600
For present impact test
400

200

0
0 1 2 3 4 5 6 7
Slip (mm)

simulations shown in Fig. 10.12b, the fiber element is stretched obviously from the
original length (13 mm) to a specific length (19.5 mm = 13 mm + 6.5 mm). When
the fiber length is stretched up to 19.5 mm in the simulations, the fiber element is
deleted to reflect that it is completely pulled out. Thus, the relationship between the
actual axial stress and slip displacement could be approximately reproduced by the
simulation.
Wille and Naaman (2012) derived the relationship of the axial stress with slip
displacement based on the single-fiber pullout test on UHP-SFRC, and the steel
fiber is identical with that of the present test. In the single-fiber pullout test, the
embedded fiber length is 6.5 mm and the maximum axial stress of the fiber is about
1100 MPa. Figure 10.13 illustrates the experimental and simulated dependences of
axial stresses of the fiber on slip displacement, respectively. A good agreement is
observed, and the bonding and sliding effects of fibers are well reproduced.
In addition, in the actual impact scenario, not all the fibers, which bridge the
crack, are exactly crossed at their middle points, as sketched in Fig. 10.14a.
10.4 Numerical Simulation and Comparison with the Test Data 403

(a)
(b)
Matrix
Fibers
a Crack
b
Steel fibers
Matrix

Fig. 10.14 Bridging fibers across the crack a photograph b sketch map

Table 10.8 Material parameters of MAT#024 for modeling the steel fiber
Material q (kg/m3) E (GPa) m fy (MPa) Et (MPa) Failure strain
Steel fiber 7800 60 0.28 550 −600 0.22

For the fiber with 13 mm length, the embedded length of 6.5 mm in the pullout test
is the maximum value that could appear in the impact test, since the fiber would be
generally pulled out from the shorter side. As illustrated in Fig. 10.14, the fibers
corresponding to the red dashed lines will be pulled out.
For simplicity, based on the theory of probability, the average pulled out length
is expressed as

E½minða; bÞ ¼ lf =4 ð10:5Þ

where a and b are the fiber length separated by the crack and lf = a + b is the total
length of one fiber. As to the fiber of 13.0 mm length used in the test, the average
pulled out length is 13/4 = 3.25 mm.
Assuming that the maximum axial stress of the fiber is dependent on its sur-
rounding contact area with the matrix, and the influences of matrix strength and the
angle between fibers and cracks are not considered. Thus, according to Wille and
Naaman (2012), the maximum axial stress of the fiber rmax in present test could be
calculated by

pð0:2=2Þ2
3:25
rmax ¼
1100 ¼ 550 MPa ð10:6Þ
pð0:2=2Þ2
6:5

Therefore, for the present test, the relationship of axial stress with slip dis-
placement is shown in Fig. 10.13 and the corresponding parameters are listed in
Table 10.8, in which the Young’s modulus E, tangent modulus Et, and FS are
obtained based on the single-fiber pullout test (Wille and Naaman 2012).
404 10 Aircraft Engine Impact on UHP-SFRC Slabs

10.4.3 Numerical Simulation of the Experiment

Some typical impact scenarios are simulated and compared with the experimental
results, i.e., Shots 4-1, 6-1, 2-2, 3-2, and 5-2 shown in Sect. 10.3, corresponding to
the different damage modes of the panels (perforation, just perforation, just scab-
bing, and penetration). In the simulations, the nodes of the eroded elements are
retained without mass loss and continued to be active in contact, which will be
shown in modeling of the fragments. In addition, the oblique angles of the engine
missiles are set identically with the actual values listed in Table 10.4.

10.4.3.1 Impact Phenomenon

Figure 10.15 illustrates the impact phenomena of the experiments and simulations
for Shots 4-1, 6-1, 2-2, 3-2, and 5-2 at 0.004 s since the engine missile just
impacting on the panel, respectively.
In the simulations of Shot 4-1 (Fig. 10.15a) and Shot 6-1 (Fig. 10.15b), the
numerical simulated phenomena show a good agreement with the experimental
observations. Both in the experiments and simulations, the engine missiles perforate
the panel at 0.004 s and are obscured by the concrete debris and liberated fibers.
While for the relatively low velocity impact such as Shot 2-2 (Fig. 10.15c), Shot
3-2 (Fig. 10.15d), and Shot 5-2 (Fig. 10.15e), all the engine missiles are left in front
of the panels, which are well reproduced by the corresponding simulations. Not
only the impact phenomena of the just perforation, just scabbing, penetration, and
engine rebound, but also the poses of the engine missiles are numerically reflected.

10.4.3.2 Deformation of Engine Missile

The recovered deformed engine missiles from the experiments and simulations in
the selected five tests are shown in Fig. 10.16. For the relatively high velocity
impact, such as Shot 4-1 (Fig. 10.16a) and Shot 6-1 (Fig. 10.16b), the steel pipe of
the engine missile crushes more seriously and the left length is shorter. While for
the relatively low velocity impact, such as Shot 2-2 (Fig. 10.16c), Shot 3-2
(Fig. 10.16d), and Shot 5-2 (Fig. 10.16e), only the 0.8-mm-thick steel pipe crushes
and the 2-mm-thick steel pipe basically keeps intact. The above numerical pre-
dictions of the deformations and the average left lengths of the engine missiles in
the selected five tests all basically coincide with the experimental results.
However, the slices would quite likely separate during the experiments, since
they are welded to the steel pipe, probably leading to larger lateral deformation of
the pipe without the supporting of the slices. However, this kind deformation of the
engine missile would not be reproduced in the simulations. The reason is that the
solid elements of the slices are always connected to the shell elements of the steel
pipe by the joint nodes during the collision. The above difference between the
experiments and simulations is illustrated by Fig. 10.16a.
10.4 Numerical Simulation and Comparison with the Test Data 405

(a)
Engine debris

(b)
Concrete debris

(c)

(d)

(e)

Fig. 10.15 Comparisons of impact phenomena between experiments and simulations for shots
a 4-1 b 6-1 c 2-2 d 3-2 e 5-2 (t = 0.004 s)
406 10 Aircraft Engine Impact on UHP-SFRC Slabs

Fig. 10.16 Experimental and (a)

simulated results of engine
deformations in shots a 4-1
b 6-1 c 2-2 d 3-2 e 5-2
70mm Tail
65mm Tail

(b)

60mm Tail 57mm Tail

(c)

118mm Tail 122mm Tail

(d)

115mm Tail
114mm Tail

(e)

111mm Tail 107mm Tail

10.4 Numerical Simulation and Comparison with the Test Data 407

10.4.3.3 Panel Damage

Figure 10.17 illustrates the experimental and simulated rear face damages of the
panel for Shots 4-1, 6-1, 2-2, 3-2, and 5-2, respectively. The dimensions of the
scabbing zones are basically well reproduced. In Shot 4-1 (Fig. 10.17a) and Shot

(a)
65mm
197mm

202mm

90mm
198mm
203mm

(b)
80mm
225mm

216mm

96mm

216mm 222mm

(c)
50mm
161mm

170mm
235mm

90 mm

175mm
171mm
233mm

Fig. 10.17 Experimental and simulated rear face damages of shots a 4-1 b 6-1 c 2-2 d 3-2 e 5-2
408 10 Aircraft Engine Impact on UHP-SFRC Slabs

(d)
60mm

101mm

88mm

85mm
93mm
88mm

(e)
70mm

Tiny
Tiny
cracks
cracks

90mm

Fig. 10.17 (continued)

6-1 (Fig. 10.17b), the perforation damage modes of the 65- and 80-mm-thick panels
are well simulated. However, for the just perforation damage mode in Shot 2-2
(Fig. 10.17c), the rear scabbing area in the experiment is obviously larger than that
in the simulation. It is observed that, although the whole spalling zone is rather
large, the thickness of marginal spalling layer is thin without obvious gradient,
which is very different from the other perforation modes. If the above marginal
spalling zone is ignored, the spalling dimension observed in the test is still similar
to the numerical results (Fig. 10.17c). The just scabbing damage mode (Shot 3-2)
and penetration damage mode (Shot 5-2) are well simulated numerically, as shown
in Fig. 10.17d, e, respectively.
Overall, the number of the cracks on the rear surface of the panel in the simu-
lations is less than that formed in the experiments. The reason is that the appearance
of the cracks in the simulations is dependent on the erosion of solid elements with
the failure criterion in which the principal strain equals to 0.04. Therefore, some
small cracks appear in the experiments could not be reproduced numerically if their
principal strains do not reach 0.04.
10.4 Numerical Simulation and Comparison with the Test Data 409

10.4.3.4 Residual Velocity

The residual velocities of the experiments and simulations for Shots 4-1, 6-1, 2-2,
3-2, and 5-2 are shown in Fig. 10.18. The velocity-time histories of the three steel
slices (Fig. 10.1b) are numerically obtained, and their average value at 0.004 s is
regarded as the final velocity of the engine missile. The numerical results agree well
with the test data.

(a) (b)
300 300
Numerical time histories
Numerical time histories
250 Rear slice 250 Rear slice
Middle slice
Middle slice
Velocity (m/s)

Front slice

Velocity (m/s)
200 200 Front slice

Experimental value 82m/s

Experimental value 46m/s
150 150

100 100

50 50

0 0
0.000 0.001 0.002 0.003 0.004 0.000 0.001 0.002 0.003 0.004
t (s) t (s)
(c) (d)
140 140
Numerical time histories Numerical time histories
120 120
Rear slice Rear slice
100 Middle slice 100 Middle slice
Velocity (m/s)
Velocity (m/s)

Front slice Front slice

80 80
Experimental value -2m/s Experimental value -4m/s
60 60
40 40
20 20
0 0
-20 -20
0.000 0.001 0.002 0.003 0.004 0.000 0.001 0.002 0.003 0.004

t (s) t (s)

140
(e)
Numerical time histories
120
Rear slice
100 Middle slice
Velocity (m/s)

Front slice
80
Experimental value -4m/s
60
40
20
0
-20
0.000 0.001 0.002 0.003 0.004
t (s)

Fig. 10.18 Experimental and simulated residual velocities of shots a 4-1 b 6-1 c 2-2 d 3-2 e 5-2
410 10 Aircraft Engine Impact on UHP-SFRC Slabs

Table 10.9 Comparisons of experimental (Exp.) with simulated (Sim.) results

Shot no. Method V0 (m/s) Vr (m/s) DM Scabbing area (m2) Lr (mm)
4-1 Exp. 253 82 P 0.030 70
Sim. 253 83 P 0.032 65
6-1 Exp. 262 46 P 0.037 60
Sim. 262 49 P 0.037 57
2-2 Exp. 139 −2 JP 0.047 118
Sim. 139 −3 JP 0.023 122
3-2 Exp. 119 −4 JS 0.006 115
Sim. 119 −7 JS 0.006 114
5-2 Exp. 121 −4 C – 111
Sim. 121 −7 C – 107

Generally, when the front slice impacts on the panel, its velocity is quickly
decelerated to nearly zero. Subsequently, the 0.8-mm-thick steel pipe is crushed
gradually, leading to the deceleration of the middle and rear slices. When the
middle slice impacts on the panel, its velocity is reduced to about 30 m/s by the
0.8-mm-thick steel pipe for the relatively high impact velocity such as Shots 4-1
and 6-1, while the reduced velocity is about 100 m/s for the relatively low impact
velocity such as Shots 2-2, 3-2, and 5-2. Although the above decelerated velocities
are different, the values of the kinetic energy absorbed by the 0.8 mm steel pipe are
nearly equal for both the high- and low-speed impact tests.
When the 0.8-mm-thick steel pipe is completely crushed, the middle slice impacts
on the panel and is decelerated. In Shot 4-1 (perforation) and Shot 6-1 (perforation),
the front slice passes through the panel; thus, it is accelerated and goes together with
the middle slice at the same speed. In Shot 2-2 (just perforation), the front slice is
also accelerated while the engine missile do not pass the panel. In Shot 3-2 (just
scabbing) and Shot 5-2 (penetration), the front slice basically keeps still. In addition,
in Shot 4-1 and Shot 6-1, the 2-mm-thick steel pipe is buckled due to the high
velocity impact, which will absorb the kinetic energy and lead to the deceleration of
the rearmost slice. However, for the relatively low velocity impact (such as Shots
2-2, 3-2, and 5-2), the middle slice, the 2-mm-thick steel pipe and the rearmost slice
basically move as a whole since the 2-mm-thick steel pipe is hardly damaged.
The simulated results of Shots 4-1, 6-1, 2-2, 3-2, and 5-2 are summarized in
Table 10.9, which show good agreement with the experimental results. Therefore, the
present numerical approach could be regarded as an effective and practical method to
simulate the collisions of aircraft engine missile with the UHP-SFRC panels.

10.4.4 Comparison with Riedel et al.’s Test

Aiming to further study the collision of aircraft engines on the UHP-SFRC panels
and verify the above numerical approach, the aircraft model impact test conducted
10.4 Numerical Simulation and Comparison with the Test Data 411

by Riedel et al. (2010) is simulated in this section. The scale of the engine model is
set as 1/10, and the panel thickness is 100 mm with the steel fiber volumetric ratio
of 1%. The fiber is 9 mm in length and 0.15 mm in diameter. The details of the
engine model and panel are shown in Figs. 10.19 and 10.20, respectively. In
addition, the parameters of the scaled engine missile and UHP-SFRC panel are
listed in Table 10.10.
Considering the symmetrical condition, a 1/4 FE model is established to save the
computational time, as shown in Fig. 10.21. The analysis approach is identical with
that presented in Sect. 10.4.1. Total of 784,000 solid elements are used to model the
UHPC panel, and the element size is 2.5 mm for the impact center zone
(200 mm
200 mm) and 5.0 mm for the marginal region. The FE model of the
engine is one-fourth of which was presented in Sect. 10.4.1. The fiber elements are
only distributed in the impact center zone in order to reduce the computational time.

(a)
(b)
76mm

2.0 0.8
82 9 127 8

237mm

Fig. 10.19 Aircraft engine model a geometry b scaled engine model (Riedel et al. 2010)

(a) (b)
100 50

1000mm

Support
700
50 100

Rebar Diameter 5.5mm

50 100 700 100 50 Rebar Distance 50mm/50mm
Concrete Cover 10mm
1000mm Fiber Content 1.0-Vol.-%
Fiber Size 9mm/0.15mm

Fig. 10.20 Panel a geometry and supports b reinforcements (Riedel et al. 2010)
412 10 Aircraft Engine Impact on UHP-SFRC Slabs

Table 10.10 Parameters for scaled engine missile and UHP-SFRC panel (Riedel et al. 2010)
Shot no. Engine missile UHP-SFRC panel
M (g) V0 (m/s) Yield strength (MPa) Ultimate fc (MPa) E (GPa)
strength (MPa)
1 1564 194.7 355 630 182.8 56.242
2 1569 258.7 355 630 172.1 55.577
3 1570 320.0 355 630 186.0 55.085
4 1570 332.0 355 630 174.5 54.922
5 1575 248.9 355 630 193.2 55.139
6 1575 368.6 355 630 196.0 54.579

Fibers Rebar
+
200

Impact
center
zone
Fixed
support

Fig. 10.21 Finite element model for Riedel et al. (2010) test

The size of the fiber elements is 9 mm in length and 0.15 mm in diameter, which is
identical with the actual situation. The steel rebar is modeled by the beam element,
and the interaction between the UHPC matrix and rebar is dealt with the keyword
CONSTRAINED_LARGRANGE_IN_SOLID.
The constitutive models of the scaled engine missile, UHPC panel, and steel
fiber are completely identical with that presented in Sect. 10.4.2. However, some
material parameters are different, as listed in Tables 10.11, 10.12, and 10.13.
The compressive and tensile strengths of the UHPC in the Riedel et al.
(2010) test are not measured, which are needed in the simulations. Therefore, the
following simplified method is used to determine those values. The compressive
strengths of the UHP-SFRC and UHPC, as well as the tensile strength of UHPC in
our tests, are measured as about 150, 110, and 6.5 MPa, respectively. Thus,
according to their simple proportional relationship and the compressive strength of
UHP-SFRC (190 MPa) in the test, the compressive and tensile strengths of the

Table 10.11 Model parameters of MAT#098 for engine missile

Density q Young’s modulus Poisson’s A (MPa) B (MPa) C n Failure
(kg/m3) Es (GPa) ratio v strain
7800 210 0.3 496 434 0.040 0.307 0.25
10.4 Numerical Simulation and Comparison with the Test Data 413

Table 10.12 Model parameters of MAT#084 for UHPC

Density q Young’s modulus Poisson’s fc ft Fracture energy Aggregate
(kg/m3) Ec (GPa) ratio v (MPa) (MPa) Ef (J/m2) size (mm)
2480 55 0.19 140 8.5 180 2

Table 10.13 Model parameters of MAT#024 for steel fiber

Density q Young’s modulus Poisson’s Yield stress fy Tangent modulus Et Failure
(kg/m3) Es (GPa) ratio v (MPa) (MPa) strain
7800 60 0.28 214 −350 0.16

Table 10.14 Model parameters of MAT#003 for steel rebar

Density q Young’s Poisson’s Yield stress fy Tangent modulus C P Failure
(kg/m3) modulus ratio v (MPa) Et (MPa) strain
Es (GPa)
7800 210 0.30 447.2 0 40.4 5 0.15

UHPC are determined as 140 and 8.5 MPa, respectively. In addition, the fracture
energy of the UHPC is also not measured; thus, a slightly greater value of 180 J/m2
is used for the UHPC in the following simulations. Although the aggregate is
limited to a maximum size of 8 mm, the average aggregate size of 2 mm is used in
the simulations, based on the general compositions of UHP-SFRC and the fracture
surface of the panels in the test. In addition, the failure strain is also set as 0.04.
According to Wille and Naaman (2012) and Eq. (10.6), the maximum axial
stress of the fiber rmax in the tests by Riedel et al. (2010) could be calculated
accordingly to Eq. (10.7). The failure strain of the steel fiber is determined as 0.16
based on its pulled length of 9/4 = 2.25 mm.

pð0:15=2Þ2
2:25
rmax ¼
1100  214 MPa ð10:7Þ
pð0:2=2Þ2
6:5

In addition, the steel rebar is modeled by the MAT_PLASTIC_KINEMATIC

(MAT#003) and the parameters are listed in Table 10.14. The constants C = 40.4
and P = 5 proposed by Jones (2011) are used to consider the effect of strain rate.
The failure strain is determined as 0.15 accordingly to the experimental results.
The tested panels used by Riedel et al. (2010) are identical in strength and
geometry, but the impact velocities of the engine missiles are ranged from 194.7 to
368.6 m/s. The numerical results of the panel damages and engine deformations at
0.006 s for the test are shown in Fig. 10.22.
In the numerical simulations of Shot 4 (Fig. 10.22d) and Shot 6 (Fig. 10.22f),
both the panels are perforated under the impact velocities of 332.0 and 368.6 m/s.
Moreover, the engine missiles are severely crushed and pass through the panels. In
414 10 Aircraft Engine Impact on UHP-SFRC Slabs

(a) (b)
ile
ile le
ngth Miss th
Miss80mm leng m
Rebar Rebar 62m

Fibers
Fibers

(c) (d)

ile ile
Miss th Miss th
leng m leng m
Concrete 15m Engine Rebar 20m
debris debris

Fibers
Rebar Fibers
Concrete
debris

(e) (f)
ile
Miss th Engine
leng m debris
52m
ile
Miss
h Rebar
lengt
m
23m

Fibers
Fibers
Concrete
Fibers debris

Fig. 10.22 Simulations results of shots a 1 b 2 c 3 d 4 e 5 f 6 in Riedel et al. (2010) (t = 0.006 s)

addition, the rebars are broken, and the steel fibers are liberated from the matrix and
fly away together with the concrete debris. It is numerically indicated that the
impact velocity of 320.0 m/s in Shot 3 (Fig. 10.22c) leads to the panel damage
mode of just perforation, which agrees well with the experimental observation
(Fig. 10.23). Therefore, when the impact velocity is less than 320 m/s, the panel
would not be perforated, as shown in Shot 1 (Fig. 10.22a), Shot 2 (Fig. 10.22b),
and Shot 5 (Fig. 10.22e). The damage of the panel in Shot 1 belongs to the pen-
etration mode, and the 2-mm-thick steel pipe of the engine missile basically remains
10.4 Numerical Simulation and Comparison with the Test Data 415

(a) (b)

Fig. 10.23 Damage of panel in shot 3 a experiment b simulation

Table 10.15 Comparisons of experimental (Exp.) with simulated (Sim.) results

Shot no. Method V0 (m/s) Vr (m/s) DM Lr (mm)
1 Exp. 194.7 −8.6 C 83
Sim. 194.7 −8.1 C 80
2 Exp. 258.7 −7.3 JS 56
Sim. 258.7 −6.1 JS 62
3 Exp. 320.0 0.0 JP 28
Sim. 320.0 0.0 JP 15
4 Exp. 332.0 11.1 P 28
Sim. 332.0 5.0 P 20
5 Exp. 248.9 −9.4 JS 58
Sim. 248.9 −5.4 JS 52
6 Exp. 368.6 16.1 P 28
Sim. 368.6 53.7 P 23

intact. While in Shot 2 (258.7 m/s) and Shot 5 (248.9 m/s), the just scabbing modes
with the engine missile slightly rebounding are reproduced numerically.
The comparisons of experimental and simulated results for Riedel et al.
(2010) test are summarized in Table 10.15, which basically show a good agree-
ment. However, there is an obvious difference in the residual velocity of the engine
missile in Shot 6 (V0 = 368.6 m/s), showing 16.1 m/s in the experiment and
53.7 m/s in the simulation. Comparing with other experimental results, especially
Shot 4 (V0 = 332.0 m/s, Vr = 11.1 m/s), it is suspected that the test result is
influenced by the impact oblique angle of the engine missile.
Generally, the test conducted by Riedel et al. (2010) is basically reproduced by
the proposed numerical approach, although there are some differences between the
416 10 Aircraft Engine Impact on UHP-SFRC Slabs

tests and simulations. Better simulation results would be obtained if more detailed
material parameters and impact conditions are required, e.g., the accurate tensile
and compressive strengths of UHPC, the rebar property, and the impact oblique
angle of the engine missile.

10.5 Parametric Analyses

In this section, the parametric influential analyses of panel thickness and impact
velocity are performed and analyzed. In Sect. 10.6, the simulated data is used to
establish an empirical formula for the design of protective structures impacted by
the engine missile.

10.5.1 Influence of Slab Thickness

Considering the deliberate hijacking accident with a relatively high velocity of

about 250 m/s, the protective performance of UHP-SFRC panels with different
thicknesses (40–100 mm corresponding to Shots H1–H8 in Table 10.16) is simu-
lated. Besides, considering the most serious damage, the engine missiles impact the
panels normally without initial oblique angle. Figure 10.24 shows the impact
phenomena at 0.004 s after the engine missile impacting on the panel as well as the
corresponding panel damages. The simulated results, including the missile residual
velocity, panel damage mode, rear scabbing area, and left engine length, are
summarized in Table 10.16. Generally, the impact resistance of UHP-SFRC panel
is improved with the thickness increasing, which leads to the lower residual velocity
of engine missile. In particular, for the Shot H7 and Shot H8, the panel damages
belong to the just perforation mode and just scabbing mode, respectively. In
addition, with the panel thickness increasing, the rear scabbing areas gradually
become larger and the engine missiles crush more seriously with smaller left
lengths.

10.5.2 Influence of Impact Velocity

Aiming to study the influences of the missile velocity, as listed in Table 10.17, six
different striking velocities are set from 150 to 280 m/s corresponding to Shots V1–
V6 in Table 10.17. The thickness of the panels is fixed as 60 mm, and the per-
pendicular impact is considered. The impact phenomena at 0.004 s and the corre-
sponding panel damages are shown in Fig. 10.25. Table 10.17 lists the related
numerical results of Shots V1–V6 for the panels and engine missiles. When the
impact velocity is larger than 150 m/s, all the UHP-SFRC panels are perforated. In
10.5 Parametric Analyses 417

(a)

145mm

86mm
144mm

50mm
(b)

157mm

87mm
153mm

(c)
202mm

89mm

202mm

70mm
(d)
209mm

87mm

204mm

Fig. 10.24 Simulated impact phenomena and panel damages of shots a H1 b H2 c H3 d H4 e H5

f H6 g H7 h H8 (t = 0.004 s)
418 10 Aircraft Engine Impact on UHP-SFRC Slabs

(e)

231mm

94mm
227mm

90mm
(f)

253mm

92mm
246mm

(g)
268mm

96mm

269mm

100mm
(h)
266mm

93mm

269mm

Fig. 10.24 (continued)

10.5 Parametric Analyses 419

Table 10.17 Simulated results for different engine velocities

Shot no. H (mm) V0 (m/s) Vr (m/s) DM Scabbing area (m2) Lr (mm)
V1 60 150 25 P 0.025 137
V2 60 180 57 P 0.026 129
V3 60 200 75 P 0.029 125
V4 60 230 99 P 0.027 104
V5 60 250 106 P 0.030 101
V6 60 280 149 P 0.024 90

Table 10.16 Simulated results for different panel thicknesses

Shot no. H (mm) V0 (m/s) Vr (m/s) DM Scabbing area (m2) Lr (mm)
H1 40 250 154 P 0.020 203
H2 50 250 140 P 0.022 184
H3 60 250 106 P 0.033 110
H4 70 250 80 P 0.035 75
H5 80 250 51 P 0.043 66
H6 90 250 27 P 0.049 58
H7 95 250 7 JP 0.055 55
H8 100 250 0 JS 0.055 52

addition, with the striking velocity increasing, the residual velocities of the engine
missiles increase, while the corresponding residual lengths decrease. Besides, the
rear scabbing areas are basically the same and not obviously influenced by the
missile striking velocity.

10.6 Modified Empirical Formula for Residual

Velocity of Engine Missiles

To predict the ballistic performance of rigid projectile impact on concrete slabs,

many empirical formulas (NDRC 1946; Li and Tong 2003; Li et al. 2005, etc.) were
proposed within a specific scope of application. However, there are still few
effective formulae to predict the terminal ballistic parameters of relatively soft
aircraft engine missiles, especially for the UHP-SFRC panels. Taking the NPP
containment for example, the residual velocity is mostly concerned to further assess
the security of the reactor pressure vessel. Aiming to guide the practical design of
the protective structures more conveniently, a simple formula for predicting the
residual velocity of the engine missile after perforation needs to be proposed.
A feasible method is to modify the existing representative empirical models
based on the simulated results in Sect. 10.5, such as the modified NDRC formula
(NDRC 1946) and UMIST formula (Li and Tong 2003; Li et al. 2005) for rigid
420 10 Aircraft Engine Impact on UHP-SFRC Slabs

projectiles. It has been illustrated by Li et al. (2005) that the perforation limits
predicted by both the modified NDRC and the UMIST formulae are in good
agreement with the test data of rigid flat-nosed projectile impacting on the thin
concrete slab. Comparing with the UMIST formula, the modified NDRC formula is
more concise and widely used; thus, the modified NDRC formula is adopted. Based
on the modified NDRC formula, Grisaro and Dancygier (2014a) further developed
an energy balance approach for predicting the residual velocity, in which the energy
dissipated in the fracture of the ejecting crater into concrete fragments as well as the

60mm
(a)

176mm

78mm
173mm

60mm
(b)
183mm

84mm

178mm

60mm
(c)
192mm

84mm

189mm

Fig. 10.25 Simulated impact phenomena and panel damages of shots a V1 b V2 c V3 d V4 e V5

f V6 (t = 0.004 s)
10.6 Modified Empirical Formula for Residual Velocity of Engine Missiles 421

60mm
(d)

191mm

84mm
184mm

60mm
(e)

88mm
200mm
200mm

60mm
(f)
170mm

85mm

170mm

Fig. 10.25 (continued)

additional cracking of the concrete panel was considered. We propose the following
formula to predict the residual velocity Vr as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
     
VBL 2 VBL 2a VBL 2b
Vr ¼ V0 1   þ ð10:8Þ
V0 V0 V0

The coefficients a and b in Eq. (10.8) are empirically obtained as a = 0.9 and
b = 1.6 by fitting the experimental results reported by Hanchak et al. (1992).
Ballistic limit VBL is determined according to the NDRC model (NDRC 1946)
422 10 Aircraft Engine Impact on UHP-SFRC Slabs

 1=1:8
dGðHÞ
VBL ¼ 59:53  1000d ð10:9aÞ
KN  M
.pffiffiffiffiffiffiffiffiffiffiffiffiffi
K ¼ 14:95 fc =106 ð10:9bÞ

where d and M are the diameter and mass of the projectile, respectively. N* is the
projectile nose geometry factor, which equals to 0.72, 0.84, 1.0, and 1.14 for flat,
hemispherical, blunt, and sharp noses, respectively. G(H) is expressed as

2
GðHÞ ¼ 0:25 hpen =d ; for hpen =d  2 ð10:10aÞ

GðHÞ ¼ hpen =d; for hpen =d [ 2 ð10:10bÞ

where hpen is the penetration depth of the projectile and has the expressions of
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
hpen =d ¼ 2:2214  4:9348  1:3928H=d ; for H=d  3 ð10:11aÞ

hpen =d ¼ 0:8065H=d  1:0645; for 3\H=d\18:06 ð10:11bÞ

The above formulae are proposed for rigid projectile. Considering the soft air-
craft engine missile impacting on the UHP-SFRC panels, a reduced coefficient η is
proposed to the initial velocity V0, which is a function of H/d and expressed as

0:5
g ¼ 0:5 þ ð10:12Þ
ð1 þ H=dÞ2

Therefore, the original formula Eq. (10.8) could be modified as

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
     
VBL 2 VBL 2a VBL 2b
Vr ¼ gV0 1   þ ð10:13Þ
gV0 gV0 gV0

In order to validate the above modified empirical formulae, the residual veloc-
ities of the engine missile impacting on the UHP-SFRC panels under the condition
of V0 = 150 m/s and H = 40 mm are further numerically simulated, except for the
simulation data listed in Table 10.16 (V0 = 250 m/s) and Table 10.17
(H = 60 mm).
Figure 10.26 shows the numerically simulated results of engine missile residual
velocities and the predicted curves by Eqs. (10.9a, 10.9b–10.13), and a good
agreement is observed. Therefore, the modified formulae could be used to predict
the residual velocity of engine missile perforating on the UHP-SFRC panels. Also,
the perforation limit (the minimum thickness of the panel required to prevent the
perforation for a given engine missile striking velocity) and ballistic limit
(the minimum initial impact velocity of the engine missile to perforate the panel
10.6 Modified Empirical Formula for Residual Velocity of Engine Missiles 423

Fig. 10.26 Dependences of (a)

engine missile residual 250
velocities on a panel Initial velocity 250m/s
thicknesses and b initial Simulated resutls

Residual velocity (m/s)

200 Eqs. (10.9-10.13)
velocities

150

100

50 Initial velocity 150m/s

Simulated resutls
Eqs. (10.9-10.13)
0
0 20 40 60 80 100
Panel thickness (mm)
(b)
200
Panel thickness 40mm
160 Simulated resutls
Eqs. (10.9-10.13)
Residual velocity (m/s)

120

80
Panel thickness 60mm
40 Simulated resutls
Eqs. (10.9-10.13)

0
120 150 180 210 240 270 300
Initial velocity (m/s)

with the constant panel thickness) of engine missile impacting on the UHP-SFRC
panel can be derived from Eqs. (10.9a, 10.9b–10.13) implicitly, which provide the
theoretical basis for the design of UHP-SFRC protective structures under the air-
craft engine missile impact.

10.7 Summary

Aircraft engine impact needs to be specially studied during the entire aircraft col-
lision with the NPP containment since it is more concentrated and massive and
would lead to the deepest local penetration among all aircraft components.
Frequently used NSC material could hardly meet the requirement of protection for
the containment, while UHP-SFRC material with excellent mechanical properties
424 10 Aircraft Engine Impact on UHP-SFRC Slabs

could greatly improve the impact resistance of the containment. In this chapter, the
effect of aircraft engine missile collision with UHP-SFRC panels is investigated
experimentally and numerically. This chapter can be summarized as follows:
(1) Two series of experiments with impact velocities of about 120 and 250 m/s are
conducted, in which the 1/10 scaled panels and engine missiles are adopted
according to Sugano et al. (1993b, c). Meanwhile, the panel thickness varies
from 40 to 80 mm to evaluate its influences on the panel damages. Under the
velocity of about 250 m/s, the thickest panel is perforated, and the residual
velocity decreases with the increasing of the panel thickness. The 50-mm-thick
panel is just perforated by the engine missile with the relatively low velocity of
139 m/s. The experiment provides useful data to reveal the damage mechanism
of the UHP-SFRC panels subjected to soft engine missile impact, which is quite
different from the impact of rigid solid projectiles.
(2) The corresponding numerical simulations of the above two series of tests are
performed. The fiber bridging effect is more directly realized by establishing a
modeling approach for the randomly distributed fibers embedded in the matrix,
and the fiber bonding and sliding effects in the matrix are reflected approxi-
mately by modifying the fiber material properties. The impact process, panel
damage, engine deformation, and residual velocity are well reproduced by the
proposed numerical approach. In addition, the engine impact test carried out by
Riedel et al. (2010) is also well numerically simulated and a good agreement is
observed between the simulated results and test data.
(3) Based on the parametric influential analyses of the panel thickness and engine
missile impact velocity by the validated numerical approach, an empirical
formula for predicting the residual velocity of aircraft engine missile impacting
on the UHP-SFRC panel is proposed.
 Part III
Protective Materials
and Structures
Chapter 11
UHPCC Targets Under Projectile Impact

Concrete is widely used in the construction of both ground and underground pro-
tective structures, such as military fortifications, underground shelters, and nuclear
containments, which are designed to withstand the intentional or accidental impact
and blast loadings caused by projectiles, fragments, aircrafts, explosives, etc. The
response of concrete [especially for the normal strength concrete (NSC)] target hit
by kinetic energy projectiles is studied for several decades. Compared with the
engineered cementitious composites (ECC) (Maalej et al. 2005), multiscale cement
composite (MSCC) (Rossi et al. 2005), reactive powder concrete (RPC) (Richard
and Cheyrezy 1995), and slurry infiltrated fiber concrete (SIFCON) (Naaman and
Homrich 1989), the ultra-high-performance cement-based composite (UHPCC)
(Rong et al. 2010; Rossi et al. 2005; Graybeal 2007; Habel and Gauvreau 2008;
Cauberg et al. 2008; Karen et al. 2008) have the following characteristics:
(i) Ordinary curing temperature and pressure as well as preparation procedure;
(ii) Prominent static and dynamic properties, such as high compressive strength
and tensile strength, and high fracture energy;
(iii) Self-consolidating workability, very low permeability, and excellent
durability.
Therefore, UHPCC may be the most prospective construction material for
protective structures to resist intense dynamic loadings.
As a special type of UHPCC, ultra-high-performance steel fiber-reinforced
concrete (UHP-SFRC) possesses very low water-to-binder ratio, high amount of
high-range water reducer (HRWR), fine aggregate with the maximum size less than
1 mm and high-strength steel fiber with the diameter of 0.15–0.20 mm (Graybeal
2007). Besides, UHP-SFRC has high compressive strength (150–200 MPa), tensile
strength (20–50 MPa), and fracture energy (20,000–40,000 J/m2). Rebentrost and
Wight (2009), Wu et al. (2009), Mao et al. (2014), Ellis et al. (2014), and Aoude
et al. (2015) conducted experimental and numerical investigations on the blast
resistance of UHP-SFRC members such as panels and columns. In this chapter, the
projectile impact resistance of UHPCC targets is mainly concerned.
© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 427
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_11
428 11 UHPCC Targets Under Projectile Impact

The projectile high-speed impact resistance of concrete is mainly dependent on

the compressive strength as well as the size and strength (hardness) of the coarse
aggregate (Zhang et al. 2005; Dancygier et al. 2007, 2014; Fang and Zhang 2013;
Bludau et al. 2006). However, in order to achieve high compressive strength, the
decreasing or even eliminating ingredient sizes in concrete is required to improve
the homogeneity and packing density, such as UHP-SFRC (seen in Table 11.1),
ECC, RPC, and MSCC. Therefore, there is no contribution of coarse aggregate to
resist the projectile penetration. In addition, as for some high-strength concrete such
as RPC, the curing conditions of high temperature and pressure must be provided to
expel the void air and extract the water in the fresh matrix, which limits its
application, especially for the construction on-site.
Table 11.1 lists the compositions of UHP-SFRC in the existing publications,
where fc is the uniaxial compressive strength, ft is the uniaxial tensile strength, and
Ef is the effective fracture energy. lf and df are the length and diameter of steel fiber,
respectively. Wf and Vf are the proportions of steel fiber in weight and volume,
respectively.
Furthermore, based on the projectile penetration tests, Zhang et al. (2005) and
Langberg and Markeset (1999) found that the DOP of the projectiles is no longer
decreasing obviously when the compressive strength of the target reaches a certain
value. The concrete materials with the compressive strengths of 90–150 MPa are
often recommended for design of protective structures based on the comprehensive

Table 11.1 Mixtures of UHP-SFRC (kg/m3)

Ref. Graybeal Máca et al. Wille et al. Habel and Farnam et al.
(2007) (2013) (2014) Gauvreau (2008) (2010a)
Portland cement 710 800 864 967 1020
Fine sand 1020 976 795 675 850
Silica fume 230 200 216 251 180
Ground quartz 210 0 0 0 0
Coarse aggregate 0 0 0 0 0
Water 110 176 164 244 385
HRWR 31 40 9.5 35 18
Accelerator 30 0 0 0 0
Glass powder 0 200 216 0 0
Fibers lf (mm) 13 13 13 5, 25 30
df (mm) – 0.15 0.2 0.25, 0.3 0.25
Wf (kg/m3) 156 160 241 430 –
Vf (%) 2 2 3 5 2
fc (MPa) 119 151.7 250 128 96
ft (MPa) – 10.3 18 11 6.9
Ef (J/m2) – 19,847 – – –
11 UHPCC Targets Under Projectile Impact 429

considerations of protective efficiency and production cost. Therefore, adding the

coarse aggregate into the matrix of the UHP-SFRC will enhance the projectile
impact resistance, although the compressive strength may decrease to some extent.
Besides, as for the shapes (straight, hooked, twisted) and volumetric fractions of
steel fibers in UHP-SFRC, Máca et al. (2014) and Sovják et al. (2015) proposed that
the straight steel fiber with the volume fraction of 1.5–2% is an optimum choice
with respect to the mechanical properties, workability, crack prohibiting ability, and
costs.
However, limited studies are concentrated on the preparation of UHPCC
including the high-strength coarse aggregate without using curing conditions of
heat treatment and high pressure. Furthermore, the existing high-speed projectile
penetration/perforation tests mainly focus on the plain or steel fiber-reinforced
high-strength concrete (SFRHSC), the related medium caliber projectile penetration
tests on UHPCC targets are scarce, especially for the influences of the strength
(hardness), size, and volumetric ratio of coarse aggregate on the impact resistance
of the target.
In this chapter, the existing studies on the projectile impact resistance of HSC
targets as well as the static and dynamic properties of UHP-SFRC material are
firstly reviewed. The preparation of UHPCC material and the triaxial compressive
experiment are then presented. Finally, the experimental and theoretical investi-
gations on the impact resistance of two types of UHPCC, e.g., basalt- and
corundum-aggregated UHP-SFRC (denoted as UHP-BASFRC and UHP-CASFRC,
respectively) panels under high-speed (up to 1320 m/s) projectile (ogival-nosed and
flat-nosed) penetrations are introduced.

11.1 A Review on Projectile Impact Resistance of HSC

Investigations on the projectile impact resistances of HSC and SFRHSC have been
drawn much attention in the past decades.
Based on the comparative projectile perforation tests on NSC (48 MPa) and
HSC (140 MPa) slabs, Hanchak et al. (1992) found that the threefold increase in
concrete compressive strength resulted in 20% decrease of the residual velocities of
the projectile after perforating concrete slabs, as shown in Fig. 11.1. In the tests, the
ogive-nosed projectiles with a diameter of 25.4 mm and a mass of 0.5 kg are
utilized, and the thickness of the slabs is 178 mm.
To investigate the performance of HSC targets under the projectile penetration,
O’Neil et al. (1999) carried out penetration tests into 157 MPa concrete targets by
using ogive-nosed projectiles with the same mass and dimensions as the projectiles
used in the tests of 36.2 and 96.7 MPa concrete targets from Forrestal et al. (1994).
Each projectile has an ogive-nose with a caliber-radius-head of 2.0, a shank
diameter of 26.9 mm and nominal mass of 0.906 kg. Based on the experimental
results shown in Fig. 11.2, it is deduced that the DOP from the experiments of 159
and 96.7 MPa concrete targets is about 46 and 33% less than the experiments of
430 11 UHPCC Targets Under Projectile Impact

Fig. 11.1 Residual velocity 1000

versus striking velocity
48 MPa
(Hanchak et al. 1992) 140 MPa
800

Vr (m/s)
600

400

200

0
200 400 600 800 1000 1200
V0 (m/s)

Fig. 11.2 DOP test data and 1.0

Test data Prediction (Forrestal et al. (1994))
model predictions (O’Neil 0.9 159 MPa
et al. 1999) 96.7 MPa
0.8
36.2 MPa
0.7
0.6
Hpen (m)

0.5
0.4
0.3
0.2
0.1
0.0
200 300 400 500 600 700 800 900
V0 (m/s)

36.2 MPa concrete targets, respectively. Also in Fig. 11.2, the experimental DOPs
are compared to the predictions by the model from Forrestal et al. (1994). The value
of parameter S in the model is determined by S ¼ 77:28fc0:513 by curve fitting as
shown in Fig. 11.3. Figure 11.2 shows that the model predictions agree well with
the experimental DOPs for the three concrete targets with different strength.
Figure 11.4 presents the predicted curves of DOP–fc by the model from Forrestal
et al. (1994). It indicates that the DOPs do not decrease greatly when the com-
pressive strength of the target is larger than 100 MPa. Langberg and Markeset
(1999) also derived that the impact resistance did not increase greatly when the
cylinder compressive strength is beyond 150 MPa.
Meanwhile, O’Neil et al. (1999) presented experimental results from the same
projectile penetrating into the SFRHSC targets with the compressive strength of
90 MPa and fiber volume fraction of 2%. Figure 11.5 gives the frontal face dam-
ages of the SFRHSC and HSC targets. It is indicated that the fibers obviously
11.1 A Review on Projectile Impact Resistance of HSC 431

Fig. 11.3 Dimensional 30

empirical parameter S versus Test data (Frew at al. (1998))
unconfined compressive 25 Test data (O'Neil at al. (1999))
strength S=82.6 fc
-0.544
(Frew at al. (1998))
-0.513
20 S=77.28 fc (Curve fitting)

15

S
10

0
0 20 40 60 80 100 120 140 160 180
fc (MPa)

Fig. 11.4 Predictions of 45

dimensional DOP versus
40
unconfined compressive 400 m/s
strength 600 m/s
35
750 m/s

30
H pen / d

25

20

15

10

5
20 40 60 80 100 120 140 160 180
fc (MPa)

prohibit visible damage of the concrete surrounding the penetration crater, although
it do not significantly improve the penetration resistance.
Dancygier and Yankelevsky (1996) and Dancygier (1998) experimentally
studied the effects of different ductility improvement measures (steel fibers, small
diameter steel wire mesh, and woven steel fence mesh of various diameters) on the
impact resistance of the HSC targets (95–100 MPa), where the projectile striking
velocities are ranged from 85 to 230 m/s. They find that adding steel fiber is an
effective measure to reduce the spalling and scabbing areas of both front and rear
faces of the targets.
Based on the projectile impact test at 364.9–378.3 m/s, Luo et al. (2000) found
that the reinforced HSC targets (72.4 MPa) exhibit smash failure, while the
SFRHSC targets (107.1–116.1 MPa) remain intact with only several radial cracks
on the front face and some minor cracks on the side face.
432 11 UHPCC Targets Under Projectile Impact

Fig. 11.5 Front-face damage of targets impacted at 800 m/s a HSC (96.7 MPa) b SFRHSC
(90 MPa) (O’Neil et al. 1999)

Liu et al. (2002) experimentally investigated the impact resistance of SFRHSC

targets, in which the fiber volumetric fraction ranged from 0% up to 12%, and the
projectile striking velocities are 314 and 414 m/s, respectively. They find that the
strength and toughness of the target increase tremendously with the rising of fiber
mixing ratio, especially when the fiber volume fraction is beyond 4%, as shown in
Fig. 11.6a. The DOP of the projectile penetrating into 100 MPa concrete targets
decreases by 33% when the fiber volume fraction rises from 0% up to 10%, as
shown in Fig. 11.6b.
Based on the projectile penetration tests on the concrete targets with the com-
pressive strengths of 45–235 MPa and the impact velocities of 610–710 m/s, Zhang
et al. (2005, 2007a) discussed the effects of the content, type and length of the

(a) (b)
200 6.0
fc=100 MPa M=0.529 kg d=30 mm
180
5.5
160
Strength (MPa)

140 5.0 V0=414m/s

120
hpen / d

4.5
100
Compressive strength 4.0
80 Flexural strength V0=314m/s
60 3.5 fc=100 MPa
40
3.0
20
0 2.5
0 2 4 6 8 10 12 0 2 4 6 8 10
Fiber fraction (%) Fiber fraction (%)

Fig. 11.6 Effect of steel fiber volume fraction on a concrete strength b DOP (Liu et al. 2002)
11.1 A Review on Projectile Impact Resistance of HSC 433

incorporated fibers, as well as the compressive and flexural tensile strengths of the
concrete on the impact resistances. It indicates that both the compressive strength of
the target and the coarse granite aggregate are beneficial to reduce DOP, while the
incorporated fibers mainly contribute to reduce crater diameter and prohibit crack
propagation. Zhang et al. (2005) further proposed that the SFRHSC with a com-
pressive strength of 100 MPa is likely one of the best choices in the design of
protective structures subjected to projectile penetration. From a series of compar-
ative projectile impact tests on NSC (40 MPa) and HSC (93–119 MPa) targets with
the striking velocities of 200–315 m/s, Dancygier et al. (2007) evaluated the
enhancing effects of compressive strength of the concrete as well as the type and
size of coarse aggregate on the impact resistance of the target. Tai (2009) carried
out the projectile low-velocity (27–104 m/s) impact tests on plain or
fiber-reinforced NSC (about 25 MPa) and RPC (161.9–192.8 MPa) targets,
respectively. It indicates that adding steel fibers slightly increases the compressive
strength of RPC, but markedly increases its toughness and fracture energy. For
example, the scabbing area decreased by approximately 50% when 1% steel fiber is
mixed into RPC.
Based on the above studies and analysis, the following conclusions may be
drawn:
(i) The impact resistance of the concrete target enhances with the increase in
compressive strength. However, it does not enhance greatly when the
compressive strength of the target reaches a certain value such as 100 MPa.
(ii) Mixing steel fibers into HSC can significantly improve the toughness and
fracture energy, and restrain the initiation and propagation of cracking by the
bridging effect, thus reducing the damages of cratering and the scabbing.
However, too much steel fibers may impair the workability of concrete. The
upper limit of the fiber volume ratio of 4% is suggested.
(iii) The type and size of coarse aggregate, especially the high-strength aggregate,
are also beneficial to decrease DOP.

11.2 Static and Dynamic Properties of UHP-SFRC

For the static mechanical properties of UHP-SFRC, Graybeal (2007) experimentally

established the relationships between the uniaxial compressive strength and the
curing ages as well as the modulus of elasticity. Based on the conclusions drawn by
Lu and Hsu (2006) that the fibers have no effect on the triaxial compressive strength
of UHP-SFRC, Sovják et al. (2013b) further studied the variations in the triaxial
compressive strength of UHPC with the confinement pressure. By comparing the
compressive strength, bending strength, direct tensile strength, fracture energy, and
the modulus of elasticity of UHP-SFRC, HPC, FRC, and NSC, Máca et al.
(2013) proposed that UHP-SFRC has excellent performance for its tensile strain
hardening accompanied by high energy-absorption capacity. Figure 11.7 shows the
434 11 UHPCC Targets Under Projectile Impact

Fig. 11.7 Uniaxial tensile behavior of HSC and UHPFRC a linear-elastic part and strain
hardening b softening (Máca et al. 2013)

typical uniaxial tensile behavior of UHP-FRC, and it indicates that UHP-FRC

exhibits strain hardening prior to reach its peak stress followed by strain softening,
unlike HPC with sudden and brittle failure shortly after the peak load. Wille et al.
(2014) also found the above phenomenon based on the direct tensile tests on
UHP-SFRC specimens with various fiber types and volume fractions (1.5–3%).
They further found that the tensile properties of UHP-SFRC are strongly dependent
on the fiber volume fraction and weakly rely on the fiber types. Habel and Gauvreau
(2008) investigated the strain rate-dependent characteristics of UHP-SFRC by
conducting the uniaxial tensile tests with the strain rate of 10−7–10−2 s−1, the
quasi-static bending tests with the strain rate of 10−6 s−1, as well as the drop weight
tests with the impact speed of about 4.2 m/s (maximum strain rate of 2 s−1).
Rong et al. (2010) conducted SHPB tests on UHP-SFRC and obtained that the
dynamic compressive strength of UHP-SFRC increases with the rising of the fiber
volume fraction (0–4%) as well as the strain rate (25.9–93.4 s−1). Farnam et al.
(2010a) conducted drop weight test on UHP-SFRC slabs periodically until the
failure occurs, in which the impact velocity is nearly 4.23 m/s. The test results
indicate that UHP-SFRC has much higher impact resistance than plain concrete.
Riedel et al. (2010) compared the impact resistances of UHP-SFRC panels with
NSC panels hit by a deformable projectile-equivalent model of the aircraft engine,
in which the weight and striking velocity of the projectile are 1.56 kg and 180 m/s,
respectively. Damages of the UHP-SFRC panels show significant decreasing
compared with the NSC panels.
From the above studies, we know that:
(i) UHP-FRC has superior static mechanical properties such as high flexural
capacity, high direct tensile strength, and fracture energy;
11.2 Static and Dynamic Properties of UHP-SFRC 435

(ii) The dynamic compressive behavior of UHP-FRC is significantly sensitive to

the strain rate, and its toughness increases with the increase in the volume
fraction of steel fibers;
(iii) Compared with NSC, UHP-FRC has much higher impact resistance.

11.3 Preparation of UHPCC

11.3.1 Compositions

To reduce the porosity of cement and increase the strength of the interface zone
between the coarse aggregate and cement paste, the following four cementitious
materials are used in the preparation of UHPCC: the Chinese standard Graded 52.5
P.II type Portland cement (20–30 lm in particle diameter), silica fume (particle size
of 0.10–0.26 lm, density of 2.1 g/cm3, specific surface area of 20500 m2/kg),
ultra-fine fly ash (density of 2.7 g/cm3), and ultra-fine slag (particle size of 3–6 lm,
density of 2.8 g/cm3). Compared with the traditional high-performance concrete,
ground fine quartz sand (the size is less than 0.6 mm) is substituted by easily
obtained natural river sand with maximal diameter of 2.5 mm and coarse aggregate.
In addition, the silica fume is partially replaced by the ultra-fine industrial waste
powder, e.g., fly ash and ultra-fine slag. Since the production of cement and fine
quartz sand will consume much more energy and resources, and silica fume is more
expensive than fly ash and ultra-fine slag, the present UHPCC is of low cost and
energy-saving.
Straight brass-coated steel fibers are used owing to their good trade-off between
the workability and mechanical properties of concrete. The equivalent diameter,
length, and tensile strength of the steel fibers are 0.175, 13 mm, and 3000 MPa,
respectively. The water-reducing ratio of the PCA®-I polycarboxylic type HRWR
(high-range-water-reducer) is no less than 35%.

11.3.2 Mixing Procedure

Aiming to achieve good workability, particle distribution and packing density, the
mixing procedure for UHPCC is controlled rigorously. A vertical pan mixer with a
capacity of 600L and a constant mixing speed (60 rpm) is used. The dry cemen-
titious materials (cement, fly ash, silica fume, slag) and sands are firstly put together
simultaneously and mixed uniformly for 3 min, and the basalt or corundum coarse
aggregate is then added and mixed uniformly for 1 min. Subsequently, the water
and superplasticizer are added together and mixed for 3–5 min. Finally, the steel
fibers are incorporated into the mixture and mixed for another 3–5 min in order to
guarantee the fibers well distributed among the matrix. The slump and spread are
436 11 UHPCC Targets Under Projectile Impact

about 230 and 610 mm, respectively, which guarantee the workability requirements
of on-site construction.

11.4 Triaxial Compressive Test of UHP-BASFRC

Concrete material under impact and blast loadings always undergoes multiaxial
compressions with a high confinement; thus, the investigations of the triaxial
compressive behavior of UHPCC are important and essential to provide valuable
information (e.g., strength criteria, post-peak response) for the structural design and
calibration/validation of the constitutive model. The existing studies on the
mechanical behavior of UHPCC are mainly concentrated on the uniaxial com-
pressive or tensile properties (Naaman and Reinhardt 1996; Petr et al. 2013; Wille
et al. 2014). The tests on the triaxial compressive behavior of UHPCC with rela-
tively high confinement ratio are limited.
In this section, based on the previous studies on the impact resistance of cement
composites, we firstly prepare UHP-BASFRC material under the ordinary proce-
dures introduced in Sect. 11.3, which has the straight steel fiber with the volumetric
ratio of 2%, high-strength basalt aggregate with the size of 5–10 mm, and the
uniaxial compressive strength of *100 MPa. Triaxial compressive tests on two
batches of 50 mm  100 mm cylindrical specimens with uniaxial compressive
strength of 95 and 129 MPa are then conducted with the maximal confinement ratio
(the ratio of confining pressure to the uniaxial compressive strength) of 0.84 and
0.78, respectively. The deviatoric stress–strain curves for the various confinement
pressures (up to 100 MPa) are derived, and the failure criteria and toughness of
UHP-BASFRC under the triaxial compression are discussed (Ren et al. 2016).
Further applications of the obtained triaxial compression data in the analyses of
projectile impact are given in Sect. 11.5 (Kong et al. 2016b).

11.4.1 A Review on Triaxial Compressive

Behavior of Concrete

Existing related studies are mainly on NSC and HSC. Wang et al. (1987) performed
a series of triaxial compressive tests on 100-mm cubic specimens, the unconfined
compressive strength is around 10 MPa, the confinement ratio reaches up to 3, and a
nearly linear relationship between the normalized triaxial compressive strength and
confinement ratio is obtained. Sfer et al. (2002) conducted the triaxial compressive
test on 150 mm  300 mm cylindrical specimens with the unconfined compressive
strength of 30 MPa, and the confinement ratio varied from 0 to 2. It is observed that
the failure envelope correlated well with the nonlinear Etse and Willam (1994)
model. Sirijaroonchai et al. (2010) experimentally studied the influences of fiber
11.4 Triaxial Compressive Test of UHP-BASFRC 437

types (high-strength hooked steel fiber and ultra-high-molecular-weight poly-

ethylene fiber), fiber volumetric fraction (1–2%), and confining pressure (41 and
52 MPa) on the passive triaxial compressive behavior of high-performance
fiber-reinforced cement composites with the unconfined compressive strength of
about 50 MPa. It indicates that the confining effect induced by the fibers with various
types and fractions is minor for relatively high confinement ratio. Chern et al.
(1992) experimentally investigated the influence of steel fiber volumetric ratio (0–
2%) and confinement pressure (0–70 MPa) on the triaxial compressive strength of
NSC with the uniaxial compressive strength of about 20 MPa. They concluded that
the addition of steel fibers has insignificant effect on the mechanical behavior of
concrete at the confining pressure up to 70 MPa.
Correspondingly, Lu and Hsu (2006) compared the triaxial compressive stress–
strain relations and failure criteria of HSC and SFRHSC, in which the unconfined
compressive strengths of HSC and SFHSC are 67 and 69 MPa, and the maximum
confinement ratios are 0.74 and 1, respectively. It is also found that the slight
reinforcing effect on the uniaxial and triaxial compressive strengths of 100 mm
 150 mm cylindrical specimens. Xie et al. (1995) experimentally confirmed the
three-parameter parabolic relationships of the confined pressure with the maximum
and residual strengths of HSC specimens (55.5 mm  110 mm cylinder), in which
the variation ranges of the unconfined compressive strengths and confinement ratio
are 60.2–119 MPa and 0–0.504, respectively. Ansair and Li (1998) and Li and
Ansair (1999, 2000) experimentally studied the triaxial constitutive relationships of
100 mm  200 mm and 75 mm  150 mm cylindrical concrete specimens with the
uniaxial compressive strengths ranged from 47.7 (maximum confinement ratio of
0.88) to 107.3 MPa (maximum confinement ratio of 0.78). The less pronounced
effect of confining pressure on the failure strength of HSC than that of NSC, and the
distinct effect of confining pressure on the failure strain are found. In addition, the
specimen size of HSC has a slight effect on the triaxial compressive failure strength
and failure surface. Furthermore, an empirical failure criterion for HSC based on the
constitutive model proposed by Ottosen (1977) is established. Attard and Setunge
(1996) and Candappa et al. (2001) investigated the stress–strain relationships of
concrete specimens (100 mm  200 mm cylinder) with the unconfined compressive
strengths of 60–130 MPa and 41.9–103.3 MPa, respectively. An empirical model
for the full triaxial stress–strain relationship is developed, and the parameters for
Ottosen (1977) constitutive model are suggested. By comprehensively assessing the
validities of the existing nine empirical formulae for triaxial ultimate compressive
strength of HSC, Girgin et al. (2007) proposed a new relationship to predict the
ultimate strength of concrete with the unconfined strength ranged from 60 to
132 MPa. Sovják et al. (2013b) experimentally studied the triaxial compressive
strengths of ultra-high-performance concrete cylinder (100 mm  200 mm) and
cubic (100 mm3) specimens; the unconfined compressive strengths are 123 and
148 MPa, and the corresponding confined ratios are 0.25 and 0.6, respectively.
Additionally, Imran and Pantazopoulou (1996), Farnam et al. (2010b), Vu et al.
(2011), and Piotrowska et al. (2014) systematically studied the influences of
438 11 UHPCC Targets Under Projectile Impact

moisture content, loading path, steel fiber (volumes, size, shape), and composition
of coarse aggregate as well as the cement paste volume on the triaxial compressive
behavior of concrete.

11.4.2 Specimens and Test Setup

Table 11.2 lists two series mixing proportion of UHP-BASFRC in tests. It indicates
that the ultra-fine industrial waste powder (silica fume, ultra-fine fly ash, and slag)
accounts for 40% of the cementitious, and the ratio of expensive silica fume is only
10%, the water–binder ratio is 0.3. The volumetric ratio of the mixing fiber is 2%.
The UHP-BASFRC specimens listed in Table 11.2 have the density of 2570 kg/m3,
and the unconfined 50 mm  100 mm cylindrical compressive strengths of the
specimen CF90 and CF130 are 95 and 129 MPa, respectively.
As shown in Fig. 11.8a, the triaxial compression test is performed by MTS
815.03 material test system. A confining pressure intensifier is used to fill and
pressurize the triaxial cell with the confining fluid, and the confining pressure r3 is
measured by an inside pressure transducer. In-vessel pressure and displacement
transducers are employed, and the axial load is applied by an axial actuator. In order
to keep the confining pressure constant, the concrete specimens are jacketed with a
rubber membrane (2 mm in thick) to prevent the penetration of fluid into the porous

Table 11.2 Mixtures of UHP-BASFRC (kg/m3)

Designing Cement Silica Fly Ultra-fine Fine Basalt Water HRWR Steel
strength fume ash slag sand aggregate (%) fibers (%)
CF90 300 50 50 100 900 700 150 1.2 2
CF130 366 61 61 122 794 700 140 1.6 2

(a) (b) (c)

Upper loading cap

Rubber
LVDT
membrane

Circ. extensometer Specimen

Rubber
membrane
Lower loading cap

Fig. 11.8 Test setup a MTS 815.03 system b specimen with rubber membrane c loading cell
11.4 Triaxial Compressive Test of UHP-BASFRC 439

Fig. 11.9 Loading path for

triaxial compression test

specimen during pressurization, as shown in Fig. 11.8b. The specimen surrounded

by the rubber membrane is placed between the upper and lower steel loading caps.
Figure 11.8c shows the instrumentation of the specimen, including one linear
variable differential transformer (LVDT) and one circumferential extensometer with
the displacement ranges of ±2.5 and 5 mm, respectively. The LVDT is assembled
at the side of the specimen to measure the overall axial displacement, and the axial
strain e1 is obtained by subtracting the LVDT current and initial readings and then
divided by the length of the specimen (100 mm). The circumferential extensometer
is used at the middle of the height of the specimen to measure the circumferential
strain e3 .
The loading path for the triaxial compression is shown in Fig. 11.9, in which the
displacement control with a rate of 0.001 mm/s is used to obtain the complete
stress–strain curve.

11.4.3 Test Results and Analyses

11.4.3.1 Failure Pattern

Figure 11.10 shows the failure patterns of the recovered UHP-BASFRC specimens
after test. It indicates obviously that all the specimens maintain their integrity.
Under the uniaxial compression, although relatively large cracks emerged, the
major splitting failure that always occurred for HSC specimens (Lu and Hsu 2006;
Xie et al. 1995) did not take place due to the addition of steel fibers. For the
specimens with relatively lower confining pressure, the shear failure induced by
several minor cracks occurred. With the further increase in confining pressure, no
visible cracks emerged and apparent swelling occured around the specimens due to
the axial contraction. Therefore, it indicates that adding fibers can improve the
failure mode of UHP-BASFRC when the confining pressure is relatively low.
While for the high confining pressure, the fiber enhancing effect is becoming slight.
440 11 UHPCC Targets Under Projectile Impact

(a)

(b)

Fig. 11.10 Failure patterns of recovered UHP-BASFRC specimens a 95 MPa b 129 MPa

It should be pointed out that the broken specimen TC1-4 is induced by the careless
operations in taking off the rubber membrane after test.

11.4.3.2 Stress–Strain Curve

Figure 11.11 shows the behaviors of UHP-BASFRC specimens subjected to vari-

ous levels of confining pressures. It is obviously that the triaxial compression
strength and the ductility of the specimens increase with the rising of confining
pressures. In the prepeak region, the slopes of the curves also gradually increase
with the rising of confining pressures. When the confining pressure r3  60 MPa,
the deviatoric stress-axial and circumferential strain curves exhibit softening. As the
confining pressures increase, the behavior of the deviatoric stress–strain curves
become more ductile, and the stress levels drop slightly after the peak stress.
Figure 11.12 illustrates the relationships of the volumetric strain ev with the axial
and circumferential strains under various confining pressures. It is found that the
softening is observed for the specimens in the case of r3  60 MPa, which is
attributed to more crack development. The volumetric strain changes from con-
traction to dilation as the cracking develops. For the higher confining pressures
larger than 60 MPa, the specimen deforms plastically and the axial contraction
dominates, since the crack development is not so pronounced.
The triaxial compression test results are listed in Table 11.3. From Table 11.3,
the relation of normalized peak axial strain ratio eu1 =euc with confinement ratio r3 =fc
11.4 Triaxial Compressive Test of UHP-BASFRC 441

Fig. 11.11 Deviatoric (a) 400

stress-axial and
circumferential strain curves 350
80MPa 80MPa
of UHP-BASFRC specimens 300
a 95 MPa b 129 MPa 60MPa 60MPa
250

(MPa)
40MPa 40MPa
200 30MPa 30MPa
150 20MPa
20MPa
10MPa 10MPa
100

50 0MPa 0MPa
0MPa 0MPa
0
-0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06

(b) 450
400 100MPa 100MPa

350 80MPa 80MPa

300 60MPa
60MPa
(MPa)

250 40MPa
30MPa 40MPa
200
30MPa
150
20MPa 20MPa
100
10MPa
10MPa
50 0MPa
0MPa
0MPa 0MPa
0
-0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06

is shown in Fig. 11.13, in which euc ¼ 0:006 and euc ¼ 0:0058 are the peak axial
strains of UHP-BASFRC under the uniaxial compression for the 95 and 129 MPa
specimens, respectively. The following nearly linear relationship between eu1 =euc and
r3 =fc is given

eu1 r3
¼ 1 þ 6:24 ð11:1Þ
euc fc

The slope of the fitted straight line is 6.24 for UHP-BASFRC. While Lu and Hsu
(2006), Ansari and Li (1998) and Candappa et al. (2001) gave the corresponding
values of 19.21, 15.15, and 20 for concrete specimens with fc ranged from 32.6 to
107.3 MPa. It indicates that for UHPCC material, the relative increasing magnitude
of the peak strain with the rising of the confinement ratio is not obviously com-
paring with the NSC and HSC.
442 11 UHPCC Targets Under Projectile Impact

Fig. 11.12 Volumetric (a) -0.05

strain-axial and 0MPa
circumferential strain curves -0.04 0MPa

of UHP-BASFRC specimens -0.03 10MPa

a 95 MPa b 129 MPa 10MPa
-0.02 40MPa
40MPa
-0.01 20MPa
20MPa
0.00

0.01 60MPa
60MPa

0.02 80MPa 80MPa

0.03
-0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06

(b) -0.05
0MPa
-0.04 0MPa

-0.03 0MPa
0MPa

-0.02 20MPa 20MPa

40MPa 40MPa
-0.01
10MPa 10MPa
0.00
80MPa
0.01 80MPa

0.02 60MPa 100MPa 60MPa 100MPa

0.03
-0.04 -0.03 -0.02 -0.01 0.00 0.01 0.02 0.03 0.04 0.05 0.06

11.4.3.3 Failure Criteria

In this section, the triaxial compression tests of UHP-BASFRC from Table 11.3 are
used to calibrate the existing three frequently used failure criteria.
(1) Mohr–Coulomb failure criterion
The Mohr–Coulomb failure criterion is a classical failure criterion for describing
the failure envelops and is still used extensively due to its simplicity and relatively
good accuracy. The pressure-dependent Mohr–Coulomb model is expressed as

ru1 r3
¼ 1þk ð11:2Þ
fc fc

where k ¼ ð1 þ sin uÞ=ð1  sin uÞ and u represents the internal-friction angle of

concrete. The existing studies indicate that the value of k is ranged from 2.6 to 5.3
for concrete specimens with diversified strengths (Lu and Hsu 2006; Xie et al.
11.4 Triaxial Compressive Test of UHP-BASFRC 443

Table 11.3 Triaxial compression test results

Test fc Confinement Peak axial Peak axial Peak lateral Ec v
no. (MPa) ratio r3 =fc strength ratio strain eu1 strain eu3 (GPa)
ru1 =fc
TC1-1 95 0 0.95 0.0060 −0.0045 24.0 0.25
TC1-2 0 1.04 0.0067 −0.0056 25.9 0.25
TC1-3 0.11 1.67 0.0141 −0.0075 24.2 0.38
TC1-4 0.21 2.20 0.0176 −0.0063 26.2 0.23
TC1-5 0.32 2.60 0.0229 −0.0055 24.9 0.15
TC1-6 0.42 3.02 0.0256 −0.0136 28.1 0.33
TC1-7 0.63 3.23 0.0262 −0.0068 24.9 0.21
TC1-8 0.84 4.46 0.0358 −0.0101 27.3 0.13
TC2-1 129 0 0.98 0.0062 −0.0009 30.8 0.07
TC2-2 0 1.03 0.0058 −0.0012 32.0 0.05
TC2-3 0.08 1.63 0.0086 −0.0026 31.8 0.16
TC2-4 0.16 1.90 0.0101 −0.0040 30.9 0.27
TC2-5 0.23 2.22 0.0135 −0.0036 32.7 0.28
TC2-6 0.31 2.56 0.0172 −0.0069 32.0 0.16
TC2-7 0.47 2.89 0.0226 −0.0046 30.1 0.15
TC2-8 0.62 3.54 0.0302 −0.0082 34.0 0.25
TC2-9 0.78 3.74 0.0318 −0.0078 28.0 0.22

Fig. 11.13 Relationship of 8

eu1 =euc with r3 =fc based on test Test data (95MPa)
on UHP-BASFRC 7 Test data (129MPa)
Fitting curve
6

0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
/fc

1995; Ansari and Li 1998; Candappa et al. 2001; Lan and Guo 1997). The present
tests give k = 4.1, according to Fig. 11.14a. Figure 11.14b also gives the existing
compressive test data on HSC, including UHPCC, HPFRC, and SIFCON as well as
the curve of Eq. (11.2) with k = 4.1.
444 11 UHPCC Targets Under Projectile Impact

Fig. 11.14 Mohr–Coulomb (a) 5.0

failure criterion for HSC Test data (95MPa)
a present test data b existing 4.5 Test data (129MPa)
test data Eq. (11.2) and k=4.1
4.0

3.5

1 c
3.0

/f
u
2.5

2.0

1.5

1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
/f
3 c

(b) 10

6
1 c
/f
u

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
/f
3 c

It indicates that (i) for low confinement pressures (r3 =fc  1), the relationship of
ru1 fc versus r3 =fc has nearly the same tendency and Eq. (11.2) with k = 4.1 agree
well with the tests data. The influences of strength, size, and compositions of
specimens are not pronounced. (ii) for high confinement pressures (r3 =fc [ 1), the
Mohr–Coulomb failure criterion overpredicts the peak triaxial stress.
(2) Willam–Warnke failure criterion
The Willam–Warnke failure criterion (Chen 1982) is a five-parameter model widely
used in commercial finite element software. The compressive meridians equation is
expressed by:
 2
sm rm rm
¼ a0 þ a1  a2 ð11:3Þ
fc fc fc
11.4 Triaxial Compressive Test of UHP-BASFRC 445
h
where rm ¼ ðr1 þ r2 þ r3 Þ=3 and sm ¼ ðr1  r2 Þ2 þ ðr2  r3 Þ2 þ ðr3 
pffiffiffiffiffi
r1 Þ2 1=2 = 15 are the main normal stress and main shear stress; r1 , r2 , and r3 are the
stresses in the three principal directions. The coefficients a0 , a1 , and a2 can be
determined based on the test data of UHP-BASFRC from Table 11.3 and are given in
the following Eq. (11.4) as shown in Fig. 11.15a.
 2
sm rm rm
¼ 0:177 þ 0:656  0:062 ð11:4Þ
fc fc fc

Figure 11.15b also gives the existing compressive test data on HSC, including
UHPCC, HPFRC, and SIFCON as well as the curve of Eq. (11.4). It indicates that
compared with the Mohr–Coulomb failure criterion, Willam–Warnke failure cri-
terion with the expression of Eq. (11.4) gives excellent predictions for HSC, and
the influences of strength, size, and compositions of specimens are not pronounced.

Fig. 11.15 Willam–Warnke (a) 1.8

failure criterion for HSC Test data (95MPa)
a present test data b existing 1.6
Test data (129MPa)
test data 1.4 Eq. (11.4)

1.2
/fc

1.0
m

0.8

0.6

0.4

0.2
0.0 0.5 1.0 1.5 2.0 2.5
/f
m c

(b) 2.0
1.8
1.6
1.4
1.2
/fc

1.0
m

0.8
0.6
0.4
0.2
0.0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
/f
m c
446 11 UHPCC Targets Under Projectile Impact

(3) Power-law failure criterion

The power-law failure criterion can be expressed by:
 b
ru1 r3
¼ 1þa ð11:5Þ
fc fc

where a and b are the coefficients.

Figure 11.16a shows the present test data on UHPCC as well as the power-law
fitted curves with the form of
 0:72
ru1 r3
¼ 1 þ 3:50 ð11:6Þ
fc fc

Fig. 11.16 Power-law failure (a) 5.0

criterion for HSC a present Test data (95MPa)
test data b existing test data 4.5 Test data (129MPa)
Eq. (11.6)
4.0

3.5
1 c
/f

3.0
u

2.5

2.0

1.5

1.0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
/f
3 c

(b) 8

5
1 c

4
/f
u

0
0.0 0.5 1.0 1.5 2.0 2.5 3.0
/f
3 c
11.4 Triaxial Compressive Test of UHP-BASFRC 447

Similarly, Fig. 11.16b gives the present and existing test data on HSC, as well as
the curve of Eq. (11.6). It can be seen that compared with the Mohr–Coulomb
failure criterion, power-law failure criterion with the expression of Eq. (11.6) gives
excellent predictions for HSC, and the influences of strength, size, and composi-
tions of specimens are also not pronounced.

11.4.3.4 Toughness

The toughness of concrete (RT) generally refers to the energy-absorbing perfor-

mance in compression and has been defined as the area under the stress–strain curve
calculated up to a specified value of strain (Taerwe 1992). Most researchers
(Nataraja et al. 1999; Fanella and Naaman 1985; Poon et al. 2004) obtained this
strain of 0.015, although the specimens still have some resistance left. In this
section, in order to calculate the toughness of UHP-BASFRC under triaxial loading,
the following convenient method suggested by Farnam et al. (2010b) is used.
A ratio named toughness index (TI) that identifies the toughness of material up to
the selected strain criteria in compression to the unconfined concrete specimen is
proposed. The energy in all three directions is described by:
Z Z
RT ¼ ðr1  r3 Þde1  2 r3 de3 ð11:7Þ

where r1 , r3 , e1 , and Re3 are the axial stress, confining stress, axial strain, and lateral
strain, respectively. ðr1  r3 Þde1 is the area under the deviatoric stress-axial
strain curve. Since the confining stress is constant, Eq. (11.7) can be further sim-
plified as:
Z
R ¼
T
ðr1  r3 Þde1  2r3 e3 ð11:8Þ

Farnam et al. (2010b) introduced the following four toughness indices (TI1, TI3,
TI5, and TI7) for the triaxial compressive test.

RTcr RT3cr RT5cr RT7cr

TI1 ¼ ; TI3 ¼ ; TI5 ¼ ; TI7 ¼ ð11:9Þ
RTcr0 RTcr0 RTcr0 RTcr0

where RTcr , RT3cr , RT5cr , and RT7cr are the toughness up to the corresponding first-crack
strain (ecr ), three times of ecr , five times of ecr , and seven times of ecr , respectively.
RTcr0 is the toughness of UHP-BASFRC with no confining pressure up to the
first-crack strain. As shown in Fig. 11.17, ecr is calculated from the deviatoric
stress-axial strain curve where the elastic stage ends. According to the above
448 11 UHPCC Targets Under Projectile Impact

definitions, the calculated results of the toughness of UHPCC in the test are listed in
Table 11.4. Here, the lowest value of Rcr0 ¼ 0:123 MPa for the specimen TC1-2 is
adopted. “—” denotes that the maximum strain in the experimental deviatoric
stress-axial strain curve could not reach its specified value. As is seen in Table 11.4,
the maximum toughness indices (RT7cr ) of UHP-BASFRC under the triaxial com-
pression can reach 82.52 times of the toughness indices (RTcr0 ) under the uniaxial
compression. Therefore, it is concluded that the confining pressure has significant
effect on increasing of the toughness indices.

Δσ (MPa)
E′ E
G′ G

C′ C

B′ B Rcr = SOAB − 2σ 3ε3 A′

ε 7cr
ε 5cr R3cr = SOBCDA − 2σ 3ε3 D ′
ε 3cr R5cr = SOBCEFDA − 2σ 3ε3 F ′
ε cr R7 cr = SOBCEGHFDA − 2σ 3ε3 H ′
H′ ε3 F ′ D′ A′ O A D F ε1 H

Fig. 11.17 Definition of toughness in terms of different multiples of first-crack strain (Farnam
et al. 2010b)

Table 11.4 Toughness indices of UHP-BASFRC under triaxial compression

No r3 (MPa) First-crack strain ecr TI1 TI3 TI5 TI7
TC1-1 0 0.0040 1.05 5.89 – –
TC1-2 0 0.0039 1.00 6.50 – –
TC1-3 10 0.0045 1.40 10.27 19.05 –
TC1-4 20 0.0044 1.60 11.70 23.80 33.49
TC1-5 30 0.0053 2.13 15.95 32.75 49.31
TC1-6 40 0.0043 1.35 11.60 24.90 36.56
TC1-7 60 0.0047 1.24 12.10 26.63 41.94
TC1-8 80 0.0056 2.83 21.39 45.24 69.25
TC2-1 0 0.0053 1.62 7.29 – –
TC2-2 0 0.0049 1.87 8.24 – –
TC2-3 10 0.0057 3.22 15.94 23.74 30.56
TC2-4 20 0.0058 3.54 20.14 32.71 –
TC2-5 30 0.0057 3.07 23.15 43.08 59.07
TC2-6 40 0.0057 3.30 23.34 40.84 –
TC2-7 60 0.0058 3.28 19.46 49.71 71.94
TC2-8 80 0.0058 2.50 25.63 55.10 82.52
TC2-9 100 0.0058 1.78 22.88 51.92 75.81
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile Penetration 449

11.5 UHP-BASFRC Targets Under Ogive-Nosed

Projectile Penetration

The projectile penetration efficiency and the damages of both projectiles and con-
crete targets are strongly dependent on the striking velocity. Most of the existing
reduce-scaled projectile penetration tests were conducted under the relatively low
striking velocities (less than *1000 m/s). While for the high-speed advanced earth
penetration weapons (the striking velocity could reach 1500 m/s), UHP-BASFRC
may be the proper construction materials for protective structures. However, the
projectile high-speed penetration resistance of UHP-BASFRC targets is still not
clear.
In this section, the projectile penetration test on UHP-BASFRC targets with the
striking velocities from 510 to 1320 m/s is firstly presented in Sect. 11.5.1. The
damages of both the UHP-BASFRC target (impact crater dimensions, DOP) and
projectile (structural destruction and ballistic trajectory deviation) are then exam-
ined. Furthermore, an improved dynamic cavity expansion model is proposed to
predict the rigid projectile penetration in Sect. 11.5.3. Finally, by using the abrasive
projectile penetration model introduced in Chap. 4, the predicted limit striking
velocities that induce the structural destruction of mass abrasive projectiles are
compared with the test data.

11.5.1 Projectile Penetration Test

11.5.1.1 UHP-BASFRC Targets

The compositions of UHP-BASFRC in the penetration test are listed in Table 11.5,
in which the weight fractions of basalt coarse aggregate Wca are up to 1042 kg/m3,
and the steel fiber volumetric ratios Vf vary from 0 to 4%.
The targets are cured atmospherically for 28 days before test, as shown in
Fig. 11.18. The compressive strength and the maximal size of basalt aggregate are
120 MPa and 10 mm, respectively. The weight fractions of basalt aggregate Wca
(kg/m3) and steel fiber ratio Vf (%), unconfined cylindrical compressive strength fc
(MPa), and the thickness H (cm) of the UHP-BASFRC targets in each shot are
listed in Table 11.6.

Table 11.5 Compositions of UHP-BASFRC

Cement Silica fume Fly ash Slag Sand Wca w/cm HRWR Vf
(kg/m3) (kg/m3) (kg/m3) (kg/m3) (kg/m3) (kg/m3) ratio (%) (%)
240–654 60–109 80–109 160– 810– 0–1042 0.17– 2 0–4
218 1090 0.2
450 11 UHPCC Targets Under Projectile Impact

Fig. 11.18 UHP-BASFRC

targets

Table 11.6 Penetration test results

Test Shot M0 V0 fc H (cm) Vf Wca DOP Crater size
no. (g) (m/s) (MPa) (%) (kg/m3) (mm) Dcf Hcf
(mm) (mm)
Test 1 A-1-1 333.1 510 67.5 65 1.5 978 148 189 50
A-1-3 344.8 67.5 65 1.5 978 168 185 51
A-2-1 336.7 87.3 55 0 1042 115 430 50
A-2-3 341.9 87.3 55 0 1042 132 485 76
A-3-1 345.6 99.3 55 1.5 870 133 335 65
A-3-3 341.7 99.3 55 1.5 870 116 200 71
A-4-1 341.0 125.2 50 2 800 124 172.5 42
A-4-3 341.0 125.2 50 2 800 120 152.5 56
A-5-1 341.1 114.0 50 3 0 134 153 52
A-5-3 341.6 114.0 50 3 0 133 130 37
Test 2 A-1-2 350.1 850 67.5 100 1.5 978 345 216 62
A-1-4 341.3 67.5 100 1.5 978 346 250 69
A-2-2 340.6 87.3 65 0 1042 260 490 110
A-2-4 341.1 87.3 65 0 1042 278 486 91
A-3-2 342.3 99.3 65 1.5 870 255 314.5 58
A-3-4 346.4 99.3 65 1.5 870 260 220.5 56
A-4-2 342.7 125.2 60 2 800 252 152.5 59
A-4-4 341.5 125.2 60 2 800 261 212.5 60
A-5-2 345.6 114.0 60 3 0 288 144 48
A-5-4 344.1 114.0 60 3 0 268 189 43
Test 3 B-1-1 392.3 1150 35 150 0 750 – 516.75 120
B-2-1 395.0 88 100 2 870 564 309 110
B-3-1 394.5 128 100 3 800 402 252.5 115
B-4-1 393.8 142 100 4 0 420 241.25 70
(continued)
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile Penetration 451

Table 11.6 (continued)

Test Shot M0 V0 fc H (cm) Vf Wca DOP Crater size
no. (g) (m/s) (MPa) (%) (kg/m3) (mm) Dcf Hcf
(mm) (mm)
Test 4 B-1-2 392.3 1250 35 150 0 750 673 420 150
B-2-2 393.3 88 100 2 870 540 250 –
B-3-2 394.3 128 100 3 800 503 238.75 119
B-4-2 394.3 142 100 4 0 473 167.5 75
Test 5 B-1-3 392.3 1320 35 150 0 750 748 403 95
B-2-3 395.0 88 100 2 870 681 295.5 102
B-3-3 393.6 128 100 3 800 130 448.75 130
B-4-3 394.1 142 100 4 0 490 251.25 88

11.5.1.2 Projectiles

Two series of reduce-scaled ogive-nosed projectiles A and B are used in the pen-
etration tests. The scheme of longitudinal section of the projectiles is given in
Fig. 11.19, where the numbers in the brackets are the dimensions of the projectile
B. As shown in Fig. 11.20a, the projectile A is made of DT300 (SiMnCrNiMoV)
steel rods with the yield strength of 1500 MPa. The projectile cartridge

Fig. 11.19 Scheme of the projectile

Fig. 11.20 Projectiles a series A b series B

452 11 UHPCC Targets Under Projectile Impact

(a)

(b)

projectile debris of the sabot and obturator

target

Fig. 11.21 Projectile penetrating into concrete targets a schematic diagram b photograph from
high-speed camera system

thickness-to-diameter ratio is 0.14, and the average mass of the projectile A is

342.2 g. The projectile B shown in Fig. 11.20b is made of D6A (45CrNiMoV) steel
rods with the yield strength of 1420 MPa. The projectile cartridge is thickened, and
the cartridge thickness-to-diameter ratio is 0.2. The average mass of the projectile B
is 393.7 g. Subcaliber technology is used. After the projectile is propelled out of the
barrel, the sabot and obturator are stripped aerodynamically into debris prior to the
projectile striking the target, as shown in Fig. 11.21. During each shot, the initial
mass of the projectile M0 is measured and the abrasion and deformation of the
recovered projectiles are recorded, which are listed in Table 11.6.

11.5.1.3 Test Setup

A schematic diagram of the projectile penetration test is shown in Fig. 11.21a.

During the test, the cylindrical target is placed with the top face perpendicular to
and centered on the longitudinal axis of the barrel. The flight process of the pro-
jectile is recorded by a high-speed camera system, and Fig. 11.21b shows the
photograph at the instant of the projectile impacting on the target. After each shot,
the damages of the targets are recorded, including the maximum DOP of the
projectile, as well as the depth Hcf and the equivalent diameter Dcf of the impact
crater.
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile Penetration 453

11.5.1.4 Test Results

Twenty shots of the projectile A on the UHP-BASFRC targets in Test 1 and Test 2,
and twelve shots of the projectile B on the UHP-BASFRC and NSC targets in Test
3–Test 5 are carried out, respectively. The penetration test results are listed in
Table 11.6, where “—” indicates that the data is not recorded.
(1) Damage of targets
There are no visible cracks emerged on the rear faces of the total 32 targets,
and the photographs of the damage on the impacted faces in each test are
shown in Figs. 11.22, 11.23, 11.24, 11.25 and 11.26.

Fig. 11.22 Photographs of target damages in Test 1 a Shot A-1-1 b Shot A-1-3 c Shot A-2-1
d Shot A-2-3 e Shot A-3-1 f Shot A-3-3 g Shot A-4-1 h Shot A-4-3 i Shot A-5-1 j Shot A-5-3

Fig. 11.23 Photographs of target damages in Test 2 a Shot A-1-2 b Shot A-1-4 c Shot A-2-2
d Shot A-2-4 e Shot A-3-2 f Shot A-3-4 g Shot A-4-2 h Shot A-4-4 i Shot A-5-2 j Shot A-5-4
454 11 UHPCC Targets Under Projectile Impact

Fig. 11.24 Photographs of target damages in Test 3 a Shot B-1-1 b Shot B-2-1 c Shot B-3-1
d Shot B-4-1

Fig. 11.25 Photographs of target damages in Test 4 a Shot B-1-2 b Shot B-2-2 c Shot B-3-2
d Shot B-4-2

Fig. 11.26 Photographs of target damages in Test 5 a Shot B-1-3 b Shot B-2-3 c Shot B-3-3
d Shot B-4-3

(2) Projectile abrasion and deformation

According to Test 1 and Test 2, we observe that the projectile A is not deformed
or damaged for the relatively low striking velocity, except for some minor
abrasions on the projectile surfaces, as shown in Fig. 11.27a. While in Test 3–
Test 5, the projectile B exhibits obvious nose-blunting or even bending or
fractures as the initial striking velocity increases, as shown in Fig. 11.27b–d.
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile Penetration 455

Fig. 11.27 Abrasions and damages of the projectiles after penetration tests a Test 1–Test 2 b Test
3 c Test 4 d Test 5

(3) Terminal ballistic trajectory

According to Test 1 and Test 2, we observe that the penetration paths of the
projectiles are all nearly straight. While for Test 3–Test 5, the terminal ballistic
trajectories of the projectiles are curved to some extent, and some projectiles
even exit the edges of the targets (such as Shot B-1-2 and Shot B-4-3).
Figure 11.28 illustrates some terminal ballistic trajectories in the tests obtained
by cutting the targets after test.

Fig. 11.28 Terminal ballistic trajectories obtained by cutting the targets a Shot B-3-1 b Shot
B-2-3 c Shot B-4-2 d Shot B-4-3
456 11 UHPCC Targets Under Projectile Impact

11.5.2 Rigid Projectile Penetration Analyses

In this section, aiming to theoretically predict the DOP of rigid projectiles pene-
trating into UHP-BASFRC targets, an improved dynamic cavity expansion model is
firstly proposed, in which the plastic behavior of UHP-BASFRC material is
described by a hyperbolic yield criterion and nonlinear EOS. An analytical
approach of rigid projectile penetration into UHP-BASFRC targets is then estab-
lished by combining the improved dynamic cavity expansion model with Newton’s
second law. Finally, the validation of the proposed approach is verified by com-
paring with the test data given in Table 11.6, the predicted results by the classical
empirical formula (Forrestal et al. 1994), and the Forrestal and Tzou (1997) model
that is based on the linear yield criterion and equation of state.

11.5.2.1 Improved Dynamic Cavity Expansion Model

The description of the cavity expansion theory (CET), elastic-cracked-plastic or

elastic-plastic response regions of concrete targets as well as the solving procedures
of CET are similar with Sect. 3.3. It is noted that the relationships of shear
strength-pressure and pressure-volumetric strain of UHPCC material show obvious
nonlinear characteristics (see Figs. 11.31 and 11.32). Consequently, in this section,
a hyperbolic yield criterion and nonlinear EOS in Holmquist–Johnson–Cook
(HJC) model are used to describe the plastic response in UHPCC targets.
The hyperbolic yield criterion is also used in Sect. 3.3.1 and rewritten as

P
rr  rh ¼ a0 þ ð11:10aÞ
a1 þ a 2 P

P ¼ ðrr þ rh þ ru Þ=3; rh ¼ ru ð11:10bÞ

where rr is the radial component of the stress, rh and ru are the hoop components
of the stresses, respectively. P is the hydrostatic pressure, which are measured
positive in compression. a0 , a1 , and a2 are the constants, which need to be deter-
mined from a suitable set of triaxial compression data.
The EOS in HJC model shown in Fig. 11.29 is suitable to describe the com-
pressive behavior of concrete materials and expressed as:
8
< Kl
ðllcrush ÞðPlock Pcrush Þ
P¼ ðlplock lcrush Þ þ Pcrush ð11:11Þ
:
K1 l þ K2 l þ K3 l3 ; l ¼ ðl
2
 llock Þ=ð1 þ llock Þ

where K is the bulk modulus of elasticity, l ¼ 1  q0 =q is the volumetric strain

with q0 and q being the densities of the undeformed and deformed target material,
respectively. As observed in Fig. 11.29 and Eq. (11.11), the EOS is piecewise
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile Penetration 457

Fig. 11.29 EOS in HJC

model (Holmquist et al. 1993) P

2 3
P = K 1 μ +K2μ + K3 μ

Plock

Pcruch Κ

μ crush μ lock μ plock

μ

defined. The first region is linear elastic from the zero to the elastic limit Pcrush,
beyond which the second transition region begins. During the transition region, air
void is gradually compressed out of the material and plastic volumetric strain
produces the compaction damage until (lplock , Plock). In the third region, all air void
is removed from material that is assumed to be completely nonlinear elastic, where
a modified volumetric strain l ¼ ðl  llock Þ=ð1 þ llock Þ is introduced. For the
UHP-BASFRC targets and the range of projectile velocity concerned in this study,
the pressure around the projectile–target interface is generally less than Plock.
Consequently, the relationship of pressure and volumetric strain for the plastic
region is described by the second segment of Eq. (11.11). In particular, when
Plock =lplock ¼ K, the above EOS reduces to a linear one.
Similar to Eq. (3.28), the relationship of the radial stress at the cavity surface
with the cavity-expansion velocity can be obtained numerically and described as:
!2
rr Vc Vc
¼ A þ B pffiffiffiffiffiffiffiffiffiffi þ C pffiffiffiffiffiffiffiffiffiffi ð11:12Þ
fc fc =q0 fc =q0

where A, B, and C are the non-dimensional constants.

UHP-BASFRC targets with the compressive strength of 95 MPa described in
Sect. 11.4 are taken as an example to gain some insights into the improved model.
The parameters of the yield criterion and EOS will be given in the following
section. Figure 11.30 presents the predictions of the interface velocities and radial
stresses at the cavity surface by the proposed model, where the velocity range is
pffiffiffiffiffiffiffiffiffiffi
0\Vc = fc =q0 \3:25 corresponding to 0\Vc \650 m/s. It is observed that the
pffiffiffiffiffiffiffiffiffiffi
cracked region disappears when Vc = fc =q0 [ 2:25.
458 11 UHPCC Targets Under Projectile Impact

Fig. 11.30 Predictions of (a) 12

95 MPa UHPCC targets
a interface velocities versus 10
cavity-expansion velocity
b cavity surface radial stresses
versus cavity-expansion 8
velocity
6

4
1/2
c/(fc/ρ0) , elastic-plastic
2 1/2
c1/(fc/ρ0)
1/2
elastic-cracked-plastic
c/(fc/ρ0)
0
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
1/2
Vc/( fc/ρ0 )

(b) 25
Elastic-cracked-plastic
Elastic-plastic
20
σr /fc

15

10

5
0.0 0.5 1.0 1.5 2.0 2.5 3.0 3.5
1/2
Vc/( fc/ρ0)

11.5.2.2 Model Parameters

The parameters of the hyperbolic yield criterion (a0, a1, and a2) and EOS (Pcrush,
lcrush , Plock, and llock ) are required in the improved dynamic cavity expansion
model. In this section, based on the triaxial compression test data given in
Sect. 11.4 as well as the existing pressure–volumetric strain test data, these
parameters are suggested for UHP-BASFRC material.
(1) Yield criterion
Based on the test data listed in Table 11.3, we normalized the pressure and
shear strength with the uniaxial compressive strength, respectively. As shown
in Fig. 11.31, it is observed that all the test data follows a general trend. The
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile Penetration 459

Fig. 11.31 Shear strength– 4.0

pressure data and yield Test data (129MPa)
3.5 Test data (95MPa)
criteria
Mohr-Coulomb
3.0
Eqs. (11.10) and (11.13)

(σr-σθ ) / fc
2.5

2.0

1.5

1.0

0.5

0.0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0 2.2 2.4
P / fc

following parameters for the hyperbolic yield criterion are proposed from the
data fitting

a0 ¼ 0:416 fc ; a1 ¼ 0:499; a2 ¼ 0:087=fc ð11:13Þ

(2) EOS
Since the first segment of EOS is linear, Pcrush = fc/3 and lcrush ¼ Pcrush =K can
be obtained. The test data of pressure–volumetric strain is needed in order to
determine the parameters in the second segment, which is, however, limited in
the literature. The isotropic compression data from Hanchak et al. (1992) for
140 MPa concrete, flyer-plate-impact test data from Yan (2001) for 80 MPa
concrete, and 105 MPa steel fiber-reinforced high-strength concrete as well as
the full-scale explosive detonation and flyer-plate-impact test data from
Gebbeken et al. (2006) for 51.2 MPa concrete are all plotted on Fig. 11.32.

Fig. 11.32 Pressure– 8

volumetric strain test data and Hanchack et al. (1992)
EOSs 7 Yan (2001)
Gebekeen et al. (2006)
6
Linear EOS
5 Eq. (11.11)
P (GPa)

3
(0.11, 2)
2

0
0.00 0.03 0.06 0.09 0.12 0.15 0.18 0.21 0.24
460 11 UHPCC Targets Under Projectile Impact

Using the EOS of HJC model with Plock = 2GPa and llock ¼ 0:11 gives a
satisfied agreement with the experimental data. It is also indicated that the
frequently used linear elastic EOS could not agree well with the experimental
data.

11.5.2.3 Proposed Model

For rigid projectile penetrations, neglecting the tangential friction stress and inte-
grating the normal compressive stresses on the projectile nose surface, the instan-
taneous axial resistance can be derived as the expression given in Eq. (3.29). As for
the present ogive-nosed projectile, the projectile nose-shaped coefficients N 0 , N 1 ,
and N 2 are formulated in Eq. (3.31). The projectile penetration into concrete-like
targets commonly induces a conical crater followed by a tunnel with the diameter
nearly equaling to the projectile shank diameter. Different from the works given in
Sect. 3.3.3, the crater depth Hcf = kd is accounted here and k = 2.57 is obtained by
averaging the experimental data.
For simplification, in the cratering region, the resistance is assumed to increase
linearly with the penetration depth. By using Newton’s second law, it has

MdV=dt ¼ chpen hpen  kd ð11:14Þ

where c is a coefficient. In the tunneling region, the motion equation is obtained

from Eq. (3.29), thus

pd 2  pffiffiffiffiffiffiffiffi 
MdV=dt ¼  N 0 Afc þ N 1 B q0 fc V þ N 2 Cq0 V 2 hpen [ kd ð11:15Þ
4

Applying the initial conditions hpen (t = 0) = 0 and V(t = 0) = V0, the solutions
of Eq. (11.14) are obtained as follows:

V0 dhpen dV
hpen ¼ sin xt; V ¼ ¼ V0 cos xt; ¼ xV0 sin xt;
x dt dt
c
x2 ¼ 0\hpen  kd ð11:16Þ
m

By using Eqs. (11.15) and (11.16), when hpen = kd, it derives the following:

pd 2  pffiffiffiffiffiffiffiffi 
MxV0 sin xt1 ¼ N 0 Afc þ N 1 B q0 fc V1 þ N 2 Cq0 V12 ð11:17aÞ
4
V0 cos xt1 ¼ V1 ð11:17bÞ
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile Penetration 461

V0
sin xt1 ¼ kd ð11:17cÞ
x

where subscript 1 denotes quantities at hpen = kd (transition from the cratering stage
to the tunneling stage). Substituting Eq. (11.17c) into Eq. (11.17a) gives the
following:

pd  pffiffiffiffiffiffiffiffi 
c¼ N 0 Afc þ N 1 B q0 fc V1 þ N 2 Cq0 V12 ð11:18Þ
4k

By using Eqs. (11.17b, c), it has


m V02  V12
c¼ ð11:19Þ
k2 d 2

where V1 can be obtained by equating Eq. (11.18) with Eq. (11.19). By using the
equation of dV/dt = VdV/dh and integrating Eq. (11.15) from V1 to zero, the
non-dimensional DOP hpen/Leff is derived finally as:
pffiffiffiffiffiffiffiffi ! pffiffiffiffiffiffiffiffi
hpen qp Afc N 0 þ N 1 B q0 fc V1 þ N 2 Cq0 V12 qp N 1 B q0 fc
¼ ln þ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 
Leff 2N 2 Cq0 Afc N 0 2
N 2 Cq0 q0 fc 4ACN 0 N 2  N 1 B2
2 0 1 0 13
pffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffi
6 B N 1 B q0 fc C B N 1 B q0 fc þ 2N 2 Cq0 V1 C7 kd
6 arctanBrffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiC B C7
4 @  A  arctan@rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
  A5 þ L
2 2 2 eff
q0 fc 4ACN 0 N 2  N 1 B q0 fc 4ACN 0 N 2  N 1 B2

ð11:20Þ

where Leff ¼ 4M= qp pd 2 is the effective length of the projectile.

11.5.2.4 Comparison with Test Data

In this section, based on the projectile penetration test on the UHP-BASFRC targets
in Sect. 11.5.1, the DOPs predicted by Eq. (11.20) with the improved dynamic
cavity expansion model, the empirical formula suggested by Forrestal et al.
(1994) with A = 82.6 (fc)−0.544 (Frew et al. 1998), B = 0 and C = 1 as well as the
classical dynamic cavity expansion model (Mohr–Coulomb yield criterion and
linear elastic EOS) proposed by Forrestal and Tzou (1997) are compared and shown
in Fig. 11.33.
It is observed that the empirical formula for concrete targets suggested by
Forrestal et al. (1994) greatly overestimates the DOP, which indicates that this
frequently used formula cannot be applied for UHP-BASFRC targets. Furthermore,
the improved dynamic cavity expansion model gives excellent predictions, while
the classical model underestimates the experimental data. The reason mainly lies in
462 11 UHPCC Targets Under Projectile Impact

(a) 7 Test data (Wu et al. (2015))

(b) 7 Test data (Wu et al. (2015))
6 Forrestal et al. (1994) model 6 Forrestal et al. (1994) model
Forrestal and Tzou (1997) model Forrestal and Tzou (1997) model
5 Proposed model 5 Proposed model

hpen/Leff
hpen/Leff

4 4

3 3

2 2

1 1

0 0
300 400 500 600 700 800 900 1000 1100 1200 300 400 500 600 700 800 900 1000 1100 1200
V0(m/s) V0(m/s)

(c) 7 (d) 7
Test data (Wu et al. (2015c)) Test data (Wu et al. (2015))
6 Forrestal et al. (1994) model 6 Forrestal et al. (1994) model
Forrestal and Tzou (1997) model Forrestal and Tzou (1997) model
5 Proposed model 5 Proposed model
hpen/Leff

4 4
hpen/Leff

3 3

2 2

1 1

0 0
300 400 500 600 700 800 900 1000 1100 1200 300 400 500 600 700 800 900 1000 1100 1200
V0(m/s) V0(m/s)

(e) 7
Test data (Wu et al. (2015))
6 Forrestal et al. (1994) model
Forrestal and Tzou (1997) model
5 Proposed model
hpen/Leff

0
300 400 500 600 700 800 900 1000 1100 1200
V0(m/s)

Fig. 11.33 Experimental and predicted non-dimensional DOP with targets strengths of
a 67.5 MPa b 87.3 MPa c 99.3 MPa d 114 MPa e 125.2 MPa

the inappropriate use of the linear yield criterion. As shown in Fig. 11.31, when P/fc
is less than 0.6 or larger than 1.5, the linear yield criterion overestimates the
experimental data. In addition, even though the EOS of HJC model is greater than
linear EOS shown in Fig. 11.32, the predicted radial stress at the cavity surface for
the improved cavity expansion model is less than the classical one. This indicates
that the contribution of EOS to radial stress at the cavity surface is less important
than the yield criterion, at least for the relatively low velocities.
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile Penetration 463

11.5.3 Mass Abrasive Projectile Penetration Analyses

In this section, as discussed in Chap. 4, the mass abrasive projectile penetration

Tests 3–5 within 1 km/s < V0  1.5 km/s are discussed. Based on the above
proposed model for the limit striking velocity of mass abrasive projectiles (Wu et al.
2014b), the structural destructions of the projectiles in above tests are analyzed.

11.5.3.1 Projectile Abrasion and Structural Destruction

From Fig. 11.27, as for Test 1 and Test 2 where the initial striking velocity
V0  1 km/s, nearly no obvious deformations of the projectiles occur and the mass
loss can be neglected (Fig. 11.27a). While for the initial striking velocity
1 km/s < V0  1.5 km/s in Tests 3–5, with the striking velocities increasing
gradually, the projectiles always experience nose abrasion, bending, and fracture
(Fig. 11.27b–d). The coarse aggregates in the NSC targets with the compressive
strength of 35 MPa are gravel with the maximum diameters up to 25–30 mm,
which are larger than the sizes of the basalt coarse aggregate (  10 mm) in the
UHP-BASFRC targets. The projectiles penetrating into the NSC targets are abraded
more obviously than 88 and 128 MPa targets. While for the 142 MPa targets, the
projectile nose is relatively less abraded because there is no coarse aggregate in the
targets. It indicates that the size and strength of the coarse aggregate both have
obvious effects on blunting the projectile nose and reducing the penetration effi-
ciency. Qualitatively speaking, the influential degree of coarse aggregate size is
more obviously than that of the strength of the coarse aggregate.
It should be pointed out that not only the axial resistance is responsible for the
bending or even fracture of the high-speed penetration projectiles, but also the
lateral resistance induced by the initial attacking/oblique angles as well as the
asymmetric projectile nose abrasions. In Sect. 4.3.5, we discuss the influential
degrees of the parameters, such as oblique angle, attacking angle, nose abrasions,
and striking velocity on the structural stability of the projectile (Wu et al. 2014b).

11.5.3.2 Terminal Ballistic Stability

From Fig. 11.28, it can be seen that for projectile high-speed penetration into
UHP-BASFRC targets, the projectiles are all deviated from the impact direction
(along the symmetric axis of the target) to some extent, and a J-shaped ballistic
trajectory is formed. However, no obvious curvilinear trajectory occurs in the
penetration tests on the 35 MPa target. Thus, we conclude that the UHP-BASFRC
targets tend to change the terminal ballistic trajectory more easily. Based on our
discussions in Wu et al. (2015b), the other influential factors leading to the devi-
ations of the terminal ballistic trajectory are the attacking/oblique angle, asymmetric
464 11 UHPCC Targets Under Projectile Impact

nose abrasion, location of the projectile mass center, and size and strength of coarse
aggregate, etc.

11.5.3.3 Limit Striking Velocity

As stated in Chap. 4, with the increase in impact velocity, the symmetrical or

asymmetrical nose abrasions occur inevitably. The symmetrical nose abrasion
induces only the axial resistance while the asymmetrical nose abrasion induces both
axial and transverse resistances. Comparing the envelope contours of the recovered
projectile noses in Tests 3–5 with the unfired projectile, we find that most of the
projectiles endure symmetrical nose abrasions.
In Sect. 4.3, we propose an approach for predicting the limit striking velocity
V0;lim , which is defined as the highest velocity at which no yielding failure occur
and the mass abrasive projectile still maintain its integral structural stability. In
other words, when the initial striking velocity exceeds V0;lim , the projectile may lose
its structural integrity and bend or even fracture. The prediction equations of V0;lim
for normal penetration without nose abrasion and with symmetrical nose abrasion
can be given again as follows:
vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
u 4Yp k1
u    c
Sf
V0;lim ¼t ð11:21aÞ
0 q0
 z0m N
N q 1 0 0 L

sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi !1=2
Yp Sfc N0
V0;lim ¼ 2 0   0  0 ð11:21bÞ
k q0 C1 1  zLm k q0 k

The parameter k 0 (unit in s2/km2) reflects the abrasion law of the projectile during
the high-speed penetration. In Sect. 4.2, the analytical expression of k0 is given as
follows:
 
gN a ðw þ 4ÞM0 1 pffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k0 ¼ p0ffiffiffi 0 ; N0a ¼ 4w20 cos1 1   ð2w  1Þ 4w  1
6 3jQSMn w20 2w
ð11:21cÞ

The definitions of the parameters in Eqs. (11.21a–c) are given in Chap. 4.

Figure 11.34 gives the test data of V0–fc from Tests 3–5 as well as the predicted
curves of V0,lim–fc based on Eqs. (11.21a, b), where the solid and hollow symbols
denote the unbroken and fractured projectiles, respectively. The related parameter
HM = 7 is chosen for the high-strength basalt coarse aggregate and
q0 ¼ 2300 kg=m3 , M0/Mn = 4.06. Ideally, the predicted curve for V0,lim should go
between the solid and hollow experimental data. It means that the projectiles cor-
responding to the experimental data located above the predicted curves of V0,lim
11.5 UHP-BASFRC Targets Under Ogive-Nosed Projectile Penetration 465

Fig. 11.34 Test data of V0–fc 2200

Unbroken projectile
and predicted V0,lim–fc curves
Fractured projectile
2000 Eq. (11.21a) Eq. (11.21b)

1800
V0,lim

V0 (m/s)
1600

1400

1200

1000
30 45 60 75 90 105 120 135 150
fc (MPa)

might be bended or even fractured, and vice versa, the projectile might maintain its
structural integrity. It can be seen that the predicted curve with the consideration of
the symmetrical nose abrasion (Eq. 11.21b) agrees much better than that without
considering the nose abrasion (Eq. 11.21a). It indicates again that the projectile
nose abrasion should be included in the structural destruction analyses under the
high-speed projectile penetration. The predicted results are little higher than those
of the test data of the unbroken projectiles. The reason is that we treat the tests as
the ideal normal penetrations in the above analyses, and the reducing influences of
initial small oblique and attacking angles as well as asymmetrical nose abrasions on
V0,lim are neglected.

11.6 UHP-BASFRC Targets Under Flat-Nosed Projectile

Penetration

Most of the existing projectile impact experiments on HSC targets were conducted
by using sharp-nosed (ogive, conic) projectiles except for Tai (2009), in which the
flat-ended projectile impact test on steel fiber-reinforced RPC (161.9–192.8 MPa)
slabs was conducted with the relatively low striking velocities (27–104.1 m/s). The
flat-nosed projectile high-speed penetration test on UHP-SFRC targets and the
corresponding calculation models are absent.
In this section, aiming to provide insight into the impact resistance of
UHP-BASFRC targets under the impact of various nose-shaped projectiles, the
supplemental test of flat-nosed projectile penetrating into UHP-BASFRC targets is
further conducted based on the previous work of the ogive-nosed projectile pene-
trating into UHP-BASFRC targets introduced in Sect. 11.5. The impact crater
dimension and volume are recorded and discussed. Both rigid and eroding pro-
jectile penetrations occur in the test, and the DOPs are discussed. By comparing
466 11 UHPCC Targets Under Projectile Impact

with the existing models (Li et al. 2005; Chen and Li 2002; Teland and Sjøl 2004;
Wen and Yang 2014; Peng et al. 2015a), the validity of our model introduced in
Sect. 3.4 (Peng et al. 2015a) for different nosed projectile and broad striking
velocity range is verified. Furthermore, based on the cavity expansion theory and
Alekseevskii-Tate equation, a model for eroding projectile penetration into concrete
targets is proposed and validated (Peng et al. 2016b).

11.6.1 Projectile Penetration Test

11.6.1.1 Flat-Nosed Projectiles

The flat-nosed cylindrical projectiles with a diameter of 15 mm, a length of 90 mm,

and a mass of 122.2 g are used. They are made of 45CrNiMoV (D6A) steel rods;
the yield strength and the density are 1420 MPa and 7800 kg/m3, respectively.
A 25.3-mm caliber smooth-bore powder gun is utilized to launch the projectiles.
The projectiles are fitted with a set of special designed aluminum sabots and
obturators. Figure 11.35a shows the separated projectile, obturator, and three pieces
of sabots as well as the package processing. The dimensions of the sabot and
obturator are given in Fig. 11.35b. After the projectile propelled out of the barrel,

(a)

(b) Obturator Sabot Projectile

25.3mm
20mm

15mm
16mm

5mm

25mm 10mm 5mm

32.8mm 90mm
127.8mm

Fig. 11.35 Projectile with sabots and obturator a photograph b dimensions

11.6 UHP-BASFRC Targets Under Flat-Nosed Projectile Penetration 467

Table 11.7 Mix proportions of UHP-BASFRC

Cement Water Fly Silica Ultra-fine Fine Basalt Fibers HRWR
ash fume slag sand
1.00 0.26 0.17 0.17 0.33 1.67 1.67 0.33 0.03

the sabot and obturator are stripped off the projectile aerodynamically prior to the
impact on the target.

11.6.1.2 UHP-BASFRC Targets

The mix proportion of UHP-BASFRC in the test is listed in Table 11.7. The
components are normalized to the cement weight, and the wet density of
UHP-BASFRC is 2530 kg/m3. The volumetric ratio of the mixing fiber is 2%, and
the straight brass-coated steel fiber is used.
To perform the penetration experiment at reasonable costs, the thickness of
UHP-BASFRC targets is designed differently according to striking velocities. For
high-speed shots (637–842 m/s), cylindrical targets with a steel culvert and the
diameter and thickness of both 500 mm are used. For low-to-mid striking veloci-
ties, the square targets with the two kinds of thicknesses (150 and 200 mm) and the
width of 400 mm  400 mm are used. All the targets are cured atmospherically for
160 days before test. The average 160d unconfined compressive strength of
UHP-BASFRC is 128.4 MPa.

11.6.1.3 Test Setup

Figure 11.36 illustrates the sketch of the projectile penetration test. The cylindrical
targets are placed with their top faces perpendicular to the barrel of the smooth-bore
powder gun, and the square panels are mounted on a stationary stiff steel frame.
The impact process of the projectiles is recorded by a high-speed camera system.
Figure 11.37 shows the typical photographs of the projectile penetration process,
which indicates that the projectile impacts on the target perpendicularly.

11.6.1.4 Test Results

Seven shots of flat-nosed projectiles penetration into the UHP-BASFRC targets are
conducted. The striking velocity V0 and DOP are listed in Table 11.8. Posttest
observations and high-speed video show that the aluminum sabot and obturator
bounce back and have little influence on the DOP and impact crater dimensions.
The target label in Table 11.8 denotes different shapes and thicknesses of the target,
e.g., Slab 150-1 means the first square slab with the thickness of 150 mm, and
Cylinder 500-3 represents the third cylindrical target with the thickness of 500 mm.
468 11 UHPCC Targets Under Projectile Impact

UHP-BASFRC
Projectile target

1000mm

Wires Nets for V0

Bullet-proof
glass Wall

0000 High-Speed
Camera
oscilloscope

Fig. 11.36 Sketch of projectile penetration test

Fig. 11.37 Photographs of impacting projectile from high-speed camera

The typical damage of the targets (e.g., Shot 1-2 and Shot 1-7) after penetration
is shown in Fig. 11.38, in which the labels F and R represent the front and rear
faces of the square slab. It can be seen that the rear face of the target is intact, and
the localized damage occurs on the impacted face due to the crack prohibiting
ability of the incorporated steel fibers.
Table 11.8 also lists the dimensions of the impact crater, including the height
Hcf, volume Vcf, equivalent diameter Dcf, and surface area Acf. In which, Acf is
measured by the pixel approach illustrated in Fig. 11.39 and Vcf is measured by the
sand filling approach. The crater depth Hcf is always considered as proportional to
the shank diameter d, e.g., Hcf = kd. For the present flat-nosed projectile penetration
test into the UHP-BASFRC target, the average coefficient k = 2.60 is obtained from
Table 11.8. For the ogive-nosed projectile penetration test into the UHP-BASFRC
11.6 UHP-BASFRC Targets Under Flat-Nosed Projectile Penetration 469

Table 11.8 Test data

Shot no. Target label V0 DOP (mm) Impact crater
(m/s) Dcf (cm) Hcf (cm) Acf (cm2) Vcf (cm3)
1-1 Slab 150-1 299 44 6.83 3.0 49.5 52.5
1-2 Slab 150-2 384 52 9.6 2.9 79.1 71
1-3 Slab 200-1 438 67 10.2 4.2 94.1 102
1-4 Slab 200-2 541 90 11.4 3.8 117.5 132
1-5 Cylinder 500-1 637 109 10.8 4.0 112.0 142
1-6 Cylinder 500-2 772 118 14.1 5.5 181.6 254
1-7 Cylinder 500-3 842 144 14.4 4.2 190.6 276

Fig. 11.38 Damage of the targets a Slab 150-2-F b Slab 150-2-R c Cylinder 500-3

target as introduced in Sect. 11.5, the corresponding averaged coefficient is

k = 2.57. Therefore, it is concluded that for the UHP-BASFRC target, the projectile
nose shape has no obvious influences on the crater depth. Whether this conclusion
is also valid for NSC targets, there is no sufficient test data available until now.
Figure 11.40 illustrates the photograph of the unfired projectile (far left one) and
recovered projectiles after penetration test. The diameter, length, and mass of the
unfired projectile (d0, L0, M0) as well as the residual diameter, length, and mass of
the recovered projectile (dr, Lr, Mr) are listed in Table 11.9. It should be pointed out
that the residual diameter dr is measured at the projectile nose, and slight bulge of
the projectile nose occurred for all shots in the test. It is found from Table 11.9 and
Fig. 11.40 that the projectiles with the initial velocity less than 637 m/s in the test
can be treated as rigid bodies approximately. However, when the impact velocity
exceeds 637 m/s, both the projectile nose and shank endure obvious erosions, and
the maximal relative mass loss and blunted length for Shots 1-7 reaches 12.3 and
10.8%, respectively.
470 11 UHPCC Targets Under Projectile Impact

(a)

(b)

Fig. 11.39 Pixel approach for measuring the area of impact crater surface a crater surface b target
face

Fig. 11.40 Photograph of

unfired and recovered
projectiles
11.6 UHP-BASFRC Targets Under Flat-Nosed Projectile Penetration 471

Table 11.9 Comparisons of mass and dimensions between the unfired and recovered projectiles
No. V0 DOP (mm) Projectile
(m/s) M0 (g) Mr (g) L0 (mm) Lr (mm) d0 (mm) dr (mm)
1-1 299 44 122.1 122.0 90 89.98 15 15.13
1-2 384 52 122.3 122.1 90 89.99 15 15.26
1-3 438 67 122.3 121.9 90 89.85 15 15.14
1-4 541 90 122.2 121.8 90 89.60 15 15.56
1-5 637 109 122.2 121.2 90 89.45 15 15.70
1-6 772 118 122.2 107.7 90 81.50 15 15.18
1-7 842 144 122.2 107.2 90 80.30 15 15.14

11.6.2 Analysis and Discussion on DOP

In this section, the DOP of the rigid and eroding flat-nosed projectile is discussed.
Firstly, comparing with the existing and present flat-nosed projectile penetration test
data, the validations of the representative formula and our proposed unified pro-
jectile penetration model in Sect. 3.4 are analyzed. Secondly, combining the cavity
expansion theory with Alekseevskii-Tate equation, a model for the DOP of eroding
projectile penetration into concrete targets is proposed and validated.

11.6.2.1 Rigid Projectile Penetration

(1) Discussion on existing models

For the flat-nosed projectile penetrating into concrete targets, there are mainly
three kinds of models to predict DOP in the existing works, namely empirical
formulae, Wen’s model (Wen and Yang 2014), and the cavity-expansion-based
models. Generally, the empirical formulae (Hughes, modified NDRC, etc. listed in
Sect. 2.2) are derived from the penetration tests. Thus, they are only applied for the
projectile nose shape as well as the striking velocity ranges where the tests covered,
and the definitions of the nose shape factors in the empirical formulae are
ambiguous. Among the existing empirical formulae, the modified NDRC formula
agrees well with the test data for impact velocities up to 300 m/s (Li et al. 2005). In
Wen’s model, a flat-nosed projectile is treated as a conical-nosed projectile with its
nose length equaling to projectile diameter. By integrating the resistance on the
projectile nose surface, the cavity-expansion-based models could be suitable for
arbitrary nose shape. Descriptions of the nose shape effect in the above three kinds
of models are different, and their rationality should be discussed and clarified.
As for the cavity-expansion based models, Chen and Li (2002) and Teland and
Sjøl (2004) proposed the formulae to predict DOP of the projectiles having arbitrary
and truncated nose shape, respectively. For flat-nosed projectiles, the only differ-
ence between the above two models is the axial resistance on the projectile during
472 11 UHPCC Targets Under Projectile Impact

the initial cratering phase, while the drag force on projectile in the tunneling region
are totally identical. In detail, Chen and Li (2002) assumed that the resistance acted
on the projectile is proportional to the penetration depth in the cratering phase for
flat-nosed projectiles, while Teland and Sjøl (2004) applied the resistance obtained
by the cavity-expansion theory on the flat nose for both cratering and tunneling
phases. Actually, when the projectile impacts on the frontal face of the target, the
instantaneous resistance acting on the flat projectile is far more than zero; thus, the
formula proposed by Teland and Sjøl (2004) is more reasonable. In Sect. 3.4, we
propose a unified model (Peng et al. 2015a) to predict DOP based on the
cavity-expansion theory and the mean resistance approach. Our predictions are
consistent with the results calculated by Teland and Sjøl (2004), but our model is
more concise in expression. Therefore, Peng’s model is adopted here to represent
the cavity-expansion-based models.
Sjøl et al. (1998, 2000) conducted the 12-mm-diameter flat-nosed projectile
penetration test into the NSC targets with the compressive strength of 35 MPa, in
which the 4340 steel and tungsten cylindrical projectiles with the masses of 20.5,
65.8, and 122.8 g were used, respectively. Figure 11.41 illustrates the test data of
Sjøl et al. (1998, 2000) as well as the predicted curves by the above-mentioned
three kind models. It indicates that except for the NDRC formula, although the
definitions of the nose shape factor are different, Wen’s model and Peng’s model
both give good predictions. However, it should be noted that owing to the scattered
test data in Fig. 11.41a as well as the lower striking velocity range in Fig. 11.41c,
more penetration tests with a broad velocity range are needed.
To further assess the influences of projectile nose shape on DOP predicted by the
existing models, Fig. 11.42 shows the predicted DOP curves by Wen’s model and
Peng’s model for different nose-shaped projectiles, in which w is CRH of the
ogive-nosed projectile. It indicates that for relatively low striking velocities, e.g.,
V0 < 400 m/s, the influences of the projectile nose shape on the predicted DOP are
not obvious. Thus, the existing flat-nosed projectile penetration test data (Sjøl et al.
1998, 2000) failed to verify the validity of the existing models sufficiently. The
flat-nosed projectile penetration test with much higher striking velocity is needed;
this is also one of the motivations to perform this test.
(2) Validation
Figure 11.43 illustrates the test data as well as the predictions by the
above-mentioned models for DOP. It indicates that for the broad striking velocity
ranging from 299 to 842 m/s in this test, a good agreement is obtained for the rigid
penetration depths between the test data and the predicted results by Peng’s model,
while Wen’s model and NDRC formula overpredicts and underestimates the DOP,
respectively.
To further confirm the validity of Peng’s model for an ogive-nosed projectile
penetrating into UHP-BASFRC targets, Fig. 11.44 shows a comparison between
the test data and the predicted curves by Peng’s model, in which the ogive-nosed
11.6 UHP-BASFRC Targets Under Flat-Nosed Projectile Penetration 473

Fig. 11.41 Test data and (a) 0.07

predicted curves of hpen–V0
for projectile with mass of 0.06
a 20.5 g b 65.8 g c 122.8 g
0.05

hpen (m)
0.04

0.03
Test data (M=20.5g)
NDRC (Li et al. (2005))
0.02 ,
Wen s model (Wen and Yang (2014))
,
Peng s model (Peng et al. (2015a))
0.01
300 400 500 600 700 800
V0 (m/s)

(b) 0.30
Test data (M=65.8g)
NDRC (Li et al. (2005))
0.25 ,
Wen s model (Wen and Yang (2014))
,
Peng s model (Peng et al. (2015a))
0.20
hpen (m)

0.15

0.10

0.05

0.00
200 300 400 500 600 700 800
V0 (m/s)

(c) 0.40
Test data (M=122.8g)
0.35 NDRC (Li et al. (2005))
,
Wen s model (Wen and Yang (2014))
0.30 ,
Peng s model (Peng et al. (2015a))
hpen (m)

0.25

0.20

0.15

0.10

0.05

0.00
200 300 400 500 600 700 800
V0(m/s)

projectile penetration test data on UHP-BASFRC are given in Sect. 11.5. It can be
seen that the predicted curves are consistent with the test data, and the validity of
Peng’s model for predicting the DOP of both ogive-nosed and flat-nosed projectiles
474 11 UHPCC Targets Under Projectile Impact

Fig. 11.42 DOP of (a) 1.0

projectiles with different nose Test data (M=122.8g)
,
shapes predicted by a Wen’s Predictions by Wen s model (Wen and Yang (2014))
model b Peng’s model 0.8 Ogive nose ( =3)
Hemispherical nose ( =0.5)
Flat nose
0.6

hpen (m)
0.4

0.2

0.0
200 300 400 500 600 700 800 900
V0(m/s)

(b) 1.0
Test data (M=122.8g)
,
Predictions by Peng s model (Peng et al. (2015a))
0.8 Ogive nose ( =3)
Conical nose ( =0.5)
Flat nose
0.6
hpen (m)

0.4

0.2

0.0
200 300 400 500 600 700 800 900
V0(m/s)

Fig. 11.43 Present test data 0.24

Test data (rigid penetration)
and model predictions for
Test data (eroding penetration)
DOP versus striking velocity 0.20 ,
Peng s model (Peng et al. (2015))
,
Wen s model (Wen and Yang (2014))
0.16 NDRC (Li et al. (2005))
hpen (m)

0.12

0.08

0.04

200 300 400 500 600 700 800 900 1000

V0 (m/s)
11.6 UHP-BASFRC Targets Under Flat-Nosed Projectile Penetration 475

Fig. 11.44 Test data and 22

predicted curves of 20 Ogive nose projectile (Sec.11.5)
hpen/d–I0 18 Flat nose projectile
16
14

hpen/d
12
10
8
6
Eroding penetration
4
2
0
0 5 10 15 20 25 30 35 40 45
I0

penetrating into UHP-BASFRC targets is validated. It should be pointed out that the
friction at the ogive-nosed projectile/target interface is considered.

11.6.2.2 Eroding Projectile Penetration

For a long-rod projectile high-speed penetrating into thick metallic targets, the
projectile begins to be eroding when striking velocity is large enough. The pene-
tration process is commonly described by the 1D modified hydrodynamic models
such as Alekseevskii-Tate (A-T) equation

1 1
qp ðv  uÞ2 þ Yp ¼ q0 u2 þ Rt ð11:22Þ
2 2

where v is the speed of the rigid portion of the projectile

 andu is the penetration

velocity. Yp ¼ ryp ð1  mÞ=ð1  2mÞ and Rt ¼ ryt 2=3 þ ln 0:57E=ryt (Tate
1967, 1986); m is the Poisson’s ratio; E is the Young’s modulus of target; and ryp
and ryt are the dynamic yield strengths of the projectile and target, respectively.
However, it is difficult to determine the value of Rt for concrete targets. In Sect. 5.1,
to make the A-T equation predictable for eroding projectile penetration into con-
crete targets, we combine the spherical cavity expansion theory with the A-T
equation to describe pressure balance at the projectile/target interface. While in this
section, we present the combination of A-T equation and cavity expansion theory in
another concise way based on the experimental results. According to the spherical
cavity expansion theory and neglecting the monomial term in Eq. (3.36), the stress
at the cavity surface during the projectile penetration process can be written as:
476 11 UHPCC Targets Under Projectile Impact

rr ¼ Afc þ Bq0 Vc2 ð11:23Þ

where Vc is the cavity expansion velocity, A and B are dimensionless coefficients.

In this test, as shown in Fig. 11.40, the residual projectiles after eroding pene-
tration still kept almost flat nose shapes. Thus, we regard that the cavity expansion
velocity Vc is equal to the penetration velocity u. The pressure balance equation at
the projectile/target interface can be derived as follows based on Eqs. (11.22) and
(11.23).

1
q ðv  uÞ2 þ Yp ¼ Afc þ Bq0 u2 ð11:24Þ
2 p

Equation (11.24), together with the following equations, provides a complete

solution of the eroding projectile penetration into concrete targets.

dl
¼  ð v  uÞ ð11:25Þ
dt
dv Yp
¼ ð11:26Þ
dt qp l

where Eq. (11.25) is the reducing rate of instantaneous length l of the projectile,
and Eq. (11.26) is the projectile deceleration equation.
For the 128.4 MPa concrete target, coefficients A and B in Eq. (11.24) can be
obtained as A = 7.03 and B = 1.83 based on the existed works (Forrestal and Tzou
1997; Peng et al. 2015b). During the calculation, the concrete target is assumed to
be a compressible elastic-plastic solid. The tested uniaxial compressive strength fc is
adopted, and Young’s modulus Ec is calculated through Ec ¼ 3375fc0:5 (Iravani
1996), while the remaining parameters are identical with Forrestal and Tzou (1997).
For the projectile, Yp = 2414 MPa in Eq. (11.27) is derived with m = 0.3 for the
D6A steel. Thus, the eroding penetration depth hpen,e can be solved through
Eqs. (11.24–11.26).
When the speed of the rigid portion of the projectile v drops to the value
Vcrit = [(Yp−Afc)/Bq0]0.5, v equals u. This implies that the residual projectile stops to
erode and then penetrates rigidly into the target with the initial striking velocity of
Vcrit. The rigid penetration depth is denoted as hpen,r. Thus, the total DOP hpen is the
sum of the eroding penetration depth hpen,e and the residual rigid penetration depth
hpen,r derived from Peng’s model. The mass of the residual projectile Mr ¼
pqp d 2 Lr =4 can be obtained by solving the residual length Lr through Eqs. (11.24–
11.26). Particularly, when the initial striking velocity V0 is lower than Vcrit, the
projectile will keep rigid during the whole penetration process and the eroding
penetration will not occur.
11.6 UHP-BASFRC Targets Under Flat-Nosed Projectile Penetration 477

Fig. 11.45 Comparisons (a) 0.095

between the test data and
model predictions for 0.090
a residual projectile length
b DOP 0.085

Lr (m)
0.080

Test data
0.075 Vcrit=571.3m/s
Prediction

0.070

0.065
200 300 400 500 600 700 800 900 1000 1100
V0 (m/s)

(b) 0.32

0.28
Test data (rigid penetration)
Test data (erosive penetration)
0.24 Erosive penetration model
Rigid penetration model
0.20
hpen (m)

0.16

0.12

0.08

0.04

0.00
0 200 400 600 800 1000 1200 1400 1600
V0 (m/s)

Figure 11.45 shows the comparisons between the test data and the predicted
curves for the final DOP and residual projectile length by Eqs. (11.24–11.26). It
indicates that the predictions by the proposed model agree well with the test data.

11.7 UHP-CASFRC Targets Under Ogive-Nosed

Projectile Penetration

In this section, aiming to find a more superior anti-strike concrete material for
protective structures, corundum coarse aggregate, which has the main chemical
component of Al2O3 and compressive strength of >2000 MPa, Young’s modulus of
about 400 GPa, density of 4000 kg/m3, Moh’s hardness of 9.5, are added to prepare
the corundum-aggregated UHP-SFRC (UHP-CASFRC, Wu et al. 2015d). A series
of static and projectile impact tests are firstly conducted. By comparing with the
478 11 UHPCC Targets Under Projectile Impact

previous projectile penetration test on UHP-BASFRC targets introduced in

Sect. 11.5, the influences of the strength (hardness), size, and volumetric ratio of
coarse aggregate, incident velocity of the projectile as well as the repeated strikes
on the impact resistance of UHP-CASFRC targets are assessed. The outstanding
resistance of UHP-CASFRC targets against the projectile penetration is verified. An
optimal size of corundum aggregate in UHP-CASFRC targets is proposed after
carefully examining the structural integrity of the projectiles. Finally, the experi-
mental mass loss and the DOP of the abrasive penetration is compared with the
existing models.

11.7.1 Static Tests of UHP-SFRC

Static tests are aimed to investigate the influences of the strength (hardness) and
volumetric ratio of the coarse aggregate on the static strength of the UHP-SFRC.
There are totally five kinds of specimens with the aggregate of size 5–15 mm in the
test and their mixing ratios of ingredients are listed in Table 11.10, where UHP-NA,
UHP-BA, and UHP-CA stand for UHP-SFRC without coarse aggregate, with basalt
aggregate and corundum aggregate, respectively. Also two volumetric fractions of
basalt or corundum aggregate (15 and 30%, e.g., UHP-BA15, UHP-BA30,
UHP-CA15, UHP-CA30) are considered. Wca (kg/m3) and Vca (%) are the weight
and the volumetric fractions of the coarse aggregate. Each kind has three speci-
mens, and their average values are listed in Table 11.10.

11.7.1.1 Uniaxial Compression Test

Uniaxial compression test is conducted on the 100 mm  100 mm  100 mm

cubic specimens after 7d and 28d, respectively. The workability indexes and
compressive strengths are given in Tables 11.11. It can be seen that the average
compressive strengths of the specimens increase with the curing time as well as the
strength (hardness) and the volumetric fractions of the coarse aggregate. For
example, the compressive strength of UHP-CA30 after 28d is 15 and 28% larger
than that of UHP-BA30 and UHP-NA, respectively.

Table 11.10 Ingredients mixing ratios

Specimens Binder Sand Water Wca Vca Vf HRWR
(kg/m3) (kg/m3) (kg/m3) (kg/m3) (%) (%) (%)
UHP-NA 1100 1100 176 0 0 1.5 4.1
UHP-BA15 935 935 150 420 15 1.5 4.1
UHP-BA30 770 770 123 840 30 1.5 4.1
UHP-CA15 935 935 150 591 15 1.5 4.1
UHP-CA30 770 770 123 1192 30 1.5 4.1
11.7 UHP-CASFRC Targets Under Ogive-Nosed Projectile Penetration 479

Table 11.11 Workability index and compressive strength

Specimens Slump Spread Air content Density 7d strength 28d strength
(mm) (mm) (%) (kg/m−3) (MPa) (MPa)
UHP-NA 280 750 3.8 2420 95.6 107.1
UHP-BA15 255 530 3.2 2490 100 115.5
UHP-BA30 180 400 3.9 2520 98.7 118.9
UHP-CA15 280 700 – – 108.7 123.0
UHP-CA30 230 540 2.4 2890 115.1 137.1

Fig. 11.46 Schematic diagram of direct tensile test a anchor b initial specimen c loading scheme
d broken specimen

11.7.1.2 Direct Tensile Test

Direct tensile test on UHP-NA, UHP-BA30, and UHP-CA30 specimens with the
dimension of 100 mm  100 mm  500 mm is conducted at the loading rate of
3.3  10−7 m/s monotonically on a MTS machine (Fig. 11.46). Figure 11.47
illustrates the direct tensile stress–strain curves. It is observed that the specimens
have high ductility and the direct tensile strengths of UHP-CA30 specimens

Fig. 11.47 Direct tensile 7

stress–strain curves
UHP-NA
6 UHP-BA30
UHP-CA30
5
Stress (MPa)

0
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Strain
480 11 UHPCC Targets Under Projectile Impact

(6.75 MPa) are 12.3 and 18.2% larger than those of the UHP-BA30 (6.01 MPa) and
UHP-NA (5.71 MPa), respectively.

11.7.1.3 Four Points Bending Test

As shown in Fig. 11.48a, four-point bending test is conducted on UHP-NA,

UHP-BA30, and UHP-CA30 specimens with the dimension of 100  100  400
mm at a loading rate of 3.3  10−7 m/s. The loading versus mid-span deflection
curves are shown in Fig. 11.48b, and the corresponding flexure strengths and
fracture energies are given in Table 11.12. It is found that the flexure strength and
the fracture energy of UHP-CA30 specimens are 7.7 and 9.0% higher than those of
UHP-BA30 specimens, as well as 10.9 and 9.1% larger than that of UHP-NA
specimens, respectively.

11.7.1.4 Discussion

The influences of the strength (hardness), size, and volumetric ratio of the coarse
aggregate on the static strength of UHP-SFRC are discussed in this section.

(a) (b) 50
45 UHP-NA
UHP-BA30
40 UHP-CA30
35
F (KN)

30
25
20
15
10
5
0
0 1 2 3 4 5 6
Mid-span deflection (mm)

Fig. 11.48 Four-point bending test a loading scheme b loading-mid-span deflection curves

Table 11.12 Four-point bending test results

Specimens Flexure strength (MPa) Fracture energy (J/m2)
UHP-NA 17.88 9813.2
UHP-BA30 18.41 9816.8
UHP-CA30 19.83 10,705.8
11.7 UHP-CASFRC Targets Under Ogive-Nosed Projectile Penetration 481

(i) The 28d compressive strength, direct tensile strength, as well as the
four-point flexure strength of the UHP-CASFRC are all larger than those of
the UHP-BASFRC with the identical coarse aggregate size and the volu-
metric ratio based on the results in Sects. 11.7.1.1–11.7.1.3. It indicates that
the static strengths of UHP-SFRC increases with the rise of the strength
(hardness) of the coarse aggregate. However, it should be pointed out that
such increasing degrees are not obvious and less than 15% are obtained in
the test. Zhang (2010) drawn the same conclusion through the comparison
between the basalt- and granite (0–10 mm)-aggregated UHP-SFRC.
(ii) Static mechanical tests are conducted before the following penetration tests.
According to the test results of UHP-CASFRC in Shots 1-1, 1-2, 2-1, 2-2,
3-1, 3-2, 5-1, and 5-2 under the identical coarse aggregate and volumetric
ratio as shown in Table 11.14, it is found that with the size of coarse
aggregate rising from 5–20 mm to 35–45 mm and 65–75 mm, the corre-
sponding average 28d compressive strength of the UHP-CASFRC also
increases from 106.6 to 125.6 MPa and 129.2 MPa, respectively.
(iii) From the compression test results given in Table 11.11, it indicates that the
increase in the volumetric ratio of the coarse aggregate from 15 to 30% leads
to 11.5 and 2.9% increase in the compressive strength for UHP-CASFRC
and UHP-BASFRC, respectively. Also the corresponding 16% increase can
be found in Table 11.14 by comparisons of UHP-CASFRC targets between
Shots 1-1, 1-2, and Shots 4-1, 4-2 (Vf increases from 30 to 45%).

11.7.2 Projectile Penetration Test

In the projectile penetration test, totally 16 UHP-CASFRC targets and 2 compar-

ative HSC targets with basalt coarse aggregate are prepared, and the compositions
of the targets are listed in Table 11.13. As shown in Fig. 11.49, the diameter of
cylindrical targets is 750 mm. The unconfined cylindrical compressive strength of
the target fc (MPa), the target thickness H (cm) in each shot as well as the corre-
sponding weight Wca (kg/m3) and the volumetric fractions Vca (%) of corundum
aggregate are given in Table 11.14. The steel fiber volumetric ratios Vf (%) are 2%.
Three different sizes of corundum da (5–20, 35–45 and 65–75 mm, as shown in
Fig. 11.50) and two volumetric percentages Vca (30 and 45%), are considered.
In the test, projectiles normally penetrate into UHP-CASFRC targets. The pro-
jectiles are identical with the projectile A as introduced in Sect. 11.5, which have
the diameter of 25.3 mm and the average mass of 341.2 g. Besides, aiming to
assess the performance of UHP-CASFRC targets under multiple impacts, repeated
shots are conducted on Targets 6-1 and 6-2, denoted by Shots 6-1-1, 6-1-2, and
6-2-1, 6-2-2 in Table 11.14, respectively.
In the test, there are no visible cracks emerged on the rear faces of all targets, and
the damage views of the front (impacted) faces for UHP-CASFRC and HSC targets
482 11 UHPCC Targets Under Projectile Impact

Table 11.13 Compositions of UHP-CASFRC and HSC targets (kg/m3)

Cement Silica Fly Slag Sand Wca w/cm HRWR Vf
fume ash ratio
UHP-CASFRC 377– 63–80 63– 126– 629– 800– 0.26 1.5–16 160
480 80 160 800 1700
HSC 266 – 38 76 720 1090 0.58 1.5 0

Fig. 11.49 UHP-CASFRC

targets

are shown in Fig. 11.51. Table 11.14 lists the DOP and the dimensions of craters,
including the equivalent diameter Dcf, area of the crater surface Acf, and the crater
volume Vcf. Pixel approach and sand filling approach introduced in Sect. 11.6.1 are
adopted to measure the area of the crater surface Acf and the crater volume Vcf,
respectively. To facilitate comparisons, the results of the previous projectile pen-
etration test on UHP-BASFRC targets in Sect. 11.5 are also given, and “—” denotes
that the data is not obtained.
Figure 11.52a shows the photographs of the unfired projectile (far left one) and
recovered projectiles (the projectile in Shot 6-1-2 missed for the deep penetration
into the target) in the test. Table 11.15 lists the mass and length of the unfired
projectile (M0 and L0), as well as the residual mass and length of the recovered
projectile (Mr and Lr). It shows that the maximal relative mass loss and blunted
length of the projectiles are 8.4 and 9.9%, respectively. Figure 11.52b also shows
the typical unfired and recovered projectile A in the penetration test on
UHP-BASFRC targets given in Sect. 11.5, in which the projectiles are not
deformed or damaged, except for some minor abrasions on the surface of nose and
shank due to the high-speed frictions.
From Fig. 11.52 and Tables 11.14 and 11.15, it indicates that:
(i) Under the nearly identical striking velocity (*510 m/s), volumetric fractions
of aggregate (30%) and compressive strength (*129.3 MPa), such as Shots
1-1, 1-2, 2-1, 2-2, 3-1, and 3-2 into the UHP-CASFRC targets, the corundum
aggregate with larger size tends to break the projectile more easily, especially
when the size is larger than 1.5 times of the projectile shank diameter d.
(ii) Comparing the test results between Shots 1-1 and 1-2 into the
UHP-CASFRC targets with and Shots 4-1 and 4-3 into the UHP-BASFRC
Table 11.14 Projectile penetration test data
11.7

Test Shot V0 fc H (cm) Vf da Vca Wca DOP Crater

no. (m/s) (MPa) (%) (mm) (%) (kg/m3) (mm) Dcf Hcf Area Volume / (°)
(mm) (mm) (cm2) (cm3)
UHP-BAFRC 1-1 510 67.5 65 1.5 5–15 36 978 148 189 50 242.1 – 58.6
1-3 65 1.5 5–15 36 978 168 185 51 223.6 – 57.4
2-1 87.3 55 0 5–15 39 1042 115 430 50 1260.5 – 76.1
2-3 55 0 5–15 39 1042 132 485 76 1780.1 – 71.7
3-1 99.3 55 1.5 5–15 32 870 133 335 65 780.3 – 67.2
3-3 55 1.5 5–15 32 870 116 200 71 370.1 – 50.9
4-1 125.2 50 2 5–15 30 800 124 172.5 42 250.2 – 60.3
4-3 50 2 5–15 30 800 120 152.5 56 178.7 – 48.6
5-1 114.0 50 3 0 0 0 134 153 52 149.8 – 50.8
5-3 50 3 0 0 0 133 130 37 114.9 – 54.7
1-2 850 67.5 100 1.5 5–15 36 978 345 216 62 369.9 – 57.0
1-4 100 1.5 5–15 36 978 346 250 69 392.1 – 58.4
2-2 87.3 65 0 5–15 39 1042 260 490 110 1789.8 – 64.7
2-4 65 0 5–15 39 1042 278 486 91 1887.3 – 68.4
3-2 99.3 65 1.5 5–15 32 870 255 314.5 58 525.4 – 68.1
3-4 65 1.5 5–15 32 870 260 220.5 56 363.6 – 60.2
UHP-CASFRC Targets Under Ogive-Nosed Projectile Penetration

4-2 125.2 60 2 5–15 30 800 252 152.5 59 199.7 – 47.2

4-4 60 2 5–15 30 800 261 212.5 60 296.7 – 57.3
5-2 114.0 60 3 0 0 0 288 144 48 131.1 – 51.0
5-4 60 3 0 0 0 268 189 43 266.0 – 62.3
(continued)
483
Table 11.14 (continued)
484

Test Shot V0 fc H (cm) Vf da Vca Wca DOP Crater

no. (m/s) (MPa) (%) (mm) (%) (kg/m3) (mm) Dcf Hcf Area Volume / (°)
(mm) (mm) (cm2) (cm3)
UHP-CAFRC 1-1 508 110.7 50 2 5–20 30 1126 98 174.5 54 237.4 387 54.1
1-2 508 50 2 5–20 30 1126 101 218.8 55 372.5 676 60.4
2-1 503 125.6 50 2 35–45 30 1126 21 138.8 21 135.8 126 73.2
2-2 515 50 2 35–45 30 1126 68 191 68 302.9 548 54.5
3-1 510 129.2 50 2 65–75 30 1126 59 161.5 59 197.8 214 53.8
3-2 517 50 2 65–75 30 1126 37 132 37 136.1 121 60.7
4-1 505 128.8 50 2 5–20 45 1700 145 161.5 54 212.0 299 51.6
4-2 507 50 2 5–20 45 1700 106 181.8 75 278.7 479 46.2
5-1 524 102.5 50 2 5–20 30 1126 106 179.3 56 253.4 453 54.0
5-2 511 50 2 5–20 30 1126 88 178.5 51 258.3 512 56.3
6-1-1 708 110.7 60 2 5–20 30 1126 171 255.5 84 549.3 995 53.9
6-1-2 710 60 2 5–20 30 1126 310 279.5 84 645.4 1169 56.5
6-2-1 722 110.7 60 2 5–20 30 1126 176 217.5 86 368.4 728 48.2
6-2-2 720 60 2 5–20 30 1126 323 242.8 95 435.9 943 48.9
7-1 832 110.7 60 2 5–20 30 1126 185 179 82 244.5 409 43.1
7-2 846 60 2 5–20 30 1126 191 206.5 84 352.4 596 47.2
HSC 1-1 510 61.8 60 0 5–20 40 1090 151 310.8 93 818.5 1843 56.9
1-2 516 60 0 5–20 40 1090 143 376.5 85 1204.7 2664 64.2
11 UHPCC Targets Under Projectile Impact
11.7 UHP-CASFRC Targets Under Ogive-Nosed Projectile Penetration 485

Fig. 11.50 Corundum

aggregate with various sizes

Fig. 11.51 Impact damages of UHP-CASFRC targets in shots a 1-1 b 1-2 c 2-1 d 2-2 e 3-1 f 3-2
g 4-1 h 4-2 i 5-1 j 5-2 k 6-1-1 l 6-1-2 m 6-2-1 n 6-2-2 o 7-1 p 7-2 q HSC1-1 r HSC1-2
486 11 UHPCC Targets Under Projectile Impact

Fig. 11.52 Unfired and recovered projectiles from the test on a UHP-CASFRC b UHP-BASFRC

targets, the projectiles penetrating into the UHP-CASFRC targets endure

much more severe abrasions than those of the UHP-BASFRC targets under
nearly the same aggregate size, compressive strength, and striking velocity.
(iii) Comparing the test results between Shots 1-1 and 1-2 into the HSC targets
and Shots 2-1 and 2-3 into the UHP-BASFRC targets, in which the basalt
aggregate size and striking velocity are nearly the same, the mass abrasions
of the projectiles have no obvious difference although the compressive
strength increases about 35%.
(iv) According to the test results of Shots 1-1 and 1-2, 6-1-1 and 6-2-1, 7-1 and
7-2 into the UHP-CASFRC targets, in which the size of the corundum
aggregate (5–20 mm) and compressive strength (129.3 MPa) are nearly the
same, the mass loss and blunted length increase gradually with the rising of
the striking velocities.

11.7.3 Discussion on DOP

In this section, based on Table 11.14, the influence of strength (hardness), size, and
the volumetric ratio of the coarse aggregate as well as the projectile striking velocity
on the DOP are discussed. Besides, the repeated strikes on the impact resistance of
UHP-CASFRC targets are assessed.

11.7.3.1 Influences of Aggregate Strength (Hardness)

Figure 11.53 illustrates the DOPs of the UHP-BASFRC and UHP-CASFRC targets
when the projectile striking velocities are nearly 510 and 850 m/s, in which the size
and volumetric ratio of the coarse aggregate are almost identical. It indicates that for
both two striking velocities, although the compressive strengths of the
UHP-BASFRC targets are higher than those of the UHP-CASFRC targets, the DOP
 11.7

Table 11.15 Comparisons of mass and length of unfired and recovered projectiles
Target Shot no. Projectile mass Projectile length Projectile
M0 (g) Mr (g) Mass loss (%) L0 (mm) Lr (mm) Length loss (%)
UHP-CASFRC 1-1 341.2 329.2 3.5 150.7 140.5 6.8 Bounced and moderately abraded
1-2 341.2 329.5 3.4 150.7 141.6 6.0 Bounced and moderately abraded
2-1 341.2 Broken – 150.9 Broken – Bounced and broken
2-2 341.2 Broken – 151.9 Broken – Bounced and broken
3-1 341.3 Broken – 150.8 Broken – Bounced and broken
3-2 341.2 Broken – 151.1 Broken – Bounced and broken
4-1 341.2 332.0 2.7 150.9 143.6 4.8 Embed and little abraded
4-2 341.2 328.4 3.8 151.9 139.2 8.4 Bounced and moderately abraded
5-1 341.2 327.7 4.0 151.2 143.1 5.4 Bounced and moderately abraded
5-2 340.7 329.3 3.3 150.8 142 5.8 Bounced and moderately abraded
6-1-1 341.3 318.6 6.7 151.1 140.2 7.2 Bounced and severely abraded
6-1-2 341.3 – – 150.7 – – Embed
6-2-1 341.2 323.5 5.2 150.4 139.7 7.1 Bounced and severely abraded
6-2-2 341.2 319.9 6.2 150.8 135.8 9.9 Bounced and bended
UHP-CASFRC Targets Under Ogive-Nosed Projectile Penetration

7-1 341.2 312.7 8.4 151.0 139.2 7.8 Embed and severely abraded
7-2 341.2 315.2 7.6 150.9 136.4 9.6 Embed and severely abraded
HSC 1-1 341.2 336.1 1.5 151.2 149.2 1.3 Bounced and nearly intact
1-2 341.2 336.3 1.4 150.8 150.3 0.3 Bounced and nearly intact
487
488 11 UHPCC Targets Under Projectile Impact

(a) (b)
180 360
UHP-BASFRC UHP-BASFRC
160 UHP-CASFRC 320 UHP-CASFRC
125.2 MPa 125.2 MPa
140 280 (252) (261)
(124)
DOP (mm)

DOP (mm)
(120) 110.7 MPa 102.5 MPa
120 (106) 240 110.7 MPa
(98) (101)
100 (191)
(88) 200 (185)

80 160
60 120
40 80
20 40
0 0
4-1 4-3 1-1 1-2 5-1 5-2 4-2 4-4 7-1 7-2

Fig. 11.53 Influences of the coarse aggregate strength (hardness) on DOP a V0 = 510 m/s
b V0 = 850 m/s

of the UHP-BASFRC targets is larger than that of the UHP-CASFRC targets

inversely. Therefore, we can derive that the decreasing degree of DOP by corundum
aggregate is more influential than that of compressive strength, although the
compressive strength helps to decrease the DOP, which agrees with the theoretical
conclusions of Wu et al. (2014a). Actually, as will be discussed in the following,
the harder coarse aggregate can reduce the DOP by aggravating the mass abrasion
of the projectile.

11.7.3.2 Influences of Aggregate Size

For the UHP-CASFRC targets, Fig. 11.54a illustrates the DOPs of the eight targets
with the nearly identical projectile striking velocities as listed in Table 11.14. It can
be easily obtained that with the average coarse aggregate size increasing from
nearly 0.5d, to 1.5d and 2.5d, the DOPs decrease gradually. For example, the

(a) (b)
160
5-20mm
140 35-45mm
102.5 MPa 65-75mm
120 110.7 MPa
DOP (mm)

(106)
100 (98) (101)
(88)
125.6 MPa
80
(68) 129.2 MPa
60 (59)

40 (37)
(21)
20

0
1-1 1-2 5-1 5-2 2-1 2-2 3-1 3-2

Fig. 11.54 Influences of the coarse aggregate size on DOP a DOP b damage of Target 2-1
11.7 UHP-CASFRC Targets Under Ogive-Nosed Projectile Penetration 489

average DOP of the Targets 3-1 and 3-2 (48 mm) is 51% lower than that of the
Targets 1-1, 1-2, 5-1, and 5-2 (98.3 mm). Note that the DOP in Shot 2-1 is
unexpectedly low, which may be caused by the projectiles’ direct hit to the
corundum aggregate and is broken into pieces, as shown in Figs. 11.52a and
11.54b. With the enlarging of the aggregate size, the probability that the projectile
hit the coarse aggregate also increases. Therefore, the impact resistance of the target
enhances considerably if the (hardness) coarse aggregate with high strength is used.

11.7.3.3 Influences of Volumetric Ratio of Corundum Aggregate

Figure 11.55a illustrates the DOPs of the six targets with nearly identical projectile
striking velocities listed in Table 11.14. It indicates that the DOPs of the targets
with higher strength and coarse aggregate volumetric ratio (Shots 4-1 and 4-2) are
inversely deeper than the targets with lower strength and coarse aggregate volu-
metric ratio (Shots 1-1 and 1-2, 5-1 and 5-2), which is opposite to our common
sense. As shown in Fig. 11.55b, many voids formed on the impacted surfaces, the
actual strength of the targets is probably lower than its nominal strength tested on
the individual cubic specimen. This may be the reason why the unexpected
experimental results occur.

11.7.3.4 Comparisons Between UHP-CASFRC and HSC Targets

Figure 11.56 shows the DOPs of the UHP-CASFRC and HSC targets of the present
test listed in Table 11.14 under nearly the same projectile penetrating velocities. It
indicates that the average DOPs of the UHP-CASFRC targets (98.3 mm) are 33%
lower than that of HSC targets (147 mm).

(a) (b)
180
Volumetric ratio 30% 128.8 MPa
160
Volumetric ratio 45% (145)
140
102.5 MPa
DOP (mm)

120 110.7 MPa

(106) (106)
(98) (101)
100
(88)
80

60

40

20

0
1-1 1-2 5-1 5-2 4-1 4-2

Fig. 11.55 Influences of the coarse aggregate volumetric ratio on DOP a DOP b detail view of
Target 4-1
490 11 UHPCC Targets Under Projectile Impact

Fig. 11.56 Comparisons of 200

DOPs between the UHP-CASFRC
180 HSC
UHP-CASFRC and HSC 61.8 MPa
targets 160 (151)
(143)
140 102.5 MPa

DOP (mm)
120 110.7 MPa
(106)
(98) (101)
100 (88)
80
60
40
20
0
1-1 1-2 5-1 5-2 HSC1-1 HSC1-2

11.7.3.5 Projectile Striking Velocity

Figure 11.57 shows the DOPs of the UHP-CASFRC targets with the projectiles
striking velocities are about 510, 700, and 850 m/s, in which the targets have nearly
the same compressive strength and aggregate size. It indicates that the DOP
increases with the rising of the projectile striking velocity, while the increasing
magnitude is weakened gradually. The reason lies in that the mass abrasions of the
projectiles become more and more serious with the increase in the striking velocity,
as seen in Fig. 11.52. This coincides with the conclusions drawn by Zhao et al.
(2010a) and Wu et al. (2015c) based on the theoretically analyses of the DOP of the
mass abrasive projectile.

Fig. 11.57 DOPs of 240

UHP-CASFRC targets versus 510m/s
210 110.7 MPa
projectile striking velocity 700m/s 110.7 MPa
850m/s (185) (191)
180 (171) (176)
DOP (mm)

150
102.5 MPa
110.7 MPa
120
(98) (101) (106)
(88)
90

60

30

0
1-1 1-2 5-1 5-2 6-1-1 6-2-1 7-1 7-2
11.7 UHP-CASFRC Targets Under Ogive-Nosed Projectile Penetration 491

11.7.3.6 Repeated Impact

Since the projectiles in Shots 6-1-1 and 6-2-1 are bounced back during the pene-
trations, the repeated penetration Shots 6-1-2 and 6-2-2 are conducted successively.
Figure 11.58 shows the comparisons of the impact crater areas and volumes of
Targets 6-1 and 6-2 under repeated impacts, respectively. It indicates that under
nearly the same striking velocities, the impact craters areas increases about 17.5 and
18.3% for Targets 6-1 and 6-2, respectively. Correspondingly, the impact craters
volumes increases about 17.5 and 29.5%, respectively. It is pointed out that the
second projectile almost penetrated along the terminal ballistic trajectories of the
first projectile, based on the target damages shown in Fig. 11.51.
Figure 11.59 shows the comparisons of the DOPs of Targets 6-1 and 6-2 under
the repeated impacts with the velocities of about 700 m/s. It indicates that under the
identical compressive concrete strength, aggregate size, and volumetric ratio, as
well as almost the same striking velocity, the average DOP of the first and second
impacts are 173.5 and 316.5 mm, respectively. In addition, it is found that the
average DOP induced by the second impact (143 mm) is smaller than that by the
first impact, which is consistent with the test results of Gomeza and Shukla (2001).

Fig. 11.58 Influences of Crater area (cm2)

1400
repeated impacts on craters Crater volume (cm3)
areas and volumes 1200 (1169)
Crater area/volume

(995)
1000 (943)

800 (728)
(645.4)
600 (549.3)
(435.9)
400 (368.4)

200

0
6-1-1 6-1-2 6-1-1 6-1-2 6-2-1 6-2-2 6-2-1 6-2-2

Fig. 11.59 DOP of the 500

repeated impacts on
UHP-CASFRC targets 110.7MPa
400

(323)
(310)
DOP (mm)

300

110.7 MPa
200
(171) (176)

100

0
6-1-1 6-2-1 6-1-2 6-2-2
492 11 UHPCC Targets Under Projectile Impact

The reason lies in that the projectiles may contact with the formed borehole surface
during the second penetration since the ballistic trajectories deviate to some extent.

11.7.4 Theoretical Predictions of Mass Loss and DOP

As shown in Fig. 11.52, the projectiles endure obvious mass abrasion and nose
blunting during the penetrations into the UHP-CASFRC targets. In Sect. 4.2, an
engineering model for the mass loss of the ogive-nosed projectile high-speed
penetrating into concrete targets is established as:

DM M0  Mr 1 0:8gN a
d¼ ¼ ¼ CV02 ; C ¼ pffiffiffi 0 ð11:27Þ
M0 M0 2 3jQS

As given in Eq. (4.53), for the DOP of the mass abrasive projectile, Zhao et al.
(2010a) proposed that
8
 9
> q0 ðN0 þ k 0 V02 Þ q0 ðN0 k0 V02 Þ >
M0 < arctan h  arctan h =
v v
hpen ¼  0    
2pk0 d 2 q0 v >
:  8k q þ 2q C N  k0 V 2  Cv ln 1 þ q0 N0 V02 >
;
0 0 0 0 Sfc

ð11:28Þ

Forrestal et al. (1996) and Frew et al. (1998) conducted six sets of ogive-nosed
projectiles penetration tests on concrete targets, in which the coarse aggregate of
four sets of tests are quartz (Moh’s hardness HM = 7) and the other two sets are
limestone (Moh’s hardness HM = 3). Figure 11.60a experimentally shows the rel-
ative mass loss of the projectile in the above six sets of tests and the present test
(Shots 1-1, 1-2, 5-1, 5-2, 6-1-1, 6-2-1, 6-2-2, 7-1, and 7-2 with identical size of
corundum aggregate). Obviously, the relative mass losses from the above test data
are scattered.
If we normalize this test data based on the Moh’s hardness of the aggregate, e.g.,
take the maximum HM = 9.5, for example, the normalized relative mass loss can be
derived as illustrated in Fig. 11.60b. It is found that the normalized relative mass
losses nearly increase linearly with the rising of 0:5V02 when V0  1 km/s, although
the projectile material (4340, AerMet100, DT300), concrete strength (13.5–
110.7 MPa), and aggregate hardness (3–9.5) are different. It is verified that the most
influential parameter on the mass abrasion of the projectile is the Moh’s hardness of
aggregate, which agrees with the theoretical conclusion of Wu et al. (2014a).
Figures 11.60a, b show the predictions of the relative mass loss based on
Eq. (11.27), where the reference value of Moh’s hardness of corundum
aggregate is taken as 9.5, and the mass loss coefficient C for the present test on
the UHP-CASFRC targets is obtained as 0.24 s2/km2. Neglecting the influences
of concrete strength, the corresponding mass loss coefficients for quartz
11.7 UHP-CASFRC Targets Under Ogive-Nosed Projectile Penetration 493

Fig. 11.60 Comparisons of (a) 12

Forrestal et al. (1996)
the test data of relative mass Frew et al. (1998)
loss and DOP with the 10 Present test
predicted curves a relative Quartz aggregate HM=7

Relative mass loss (%)

Corundum aggregate HM=9.5
mass loss b normalized
8
relative mass loss c DOP

2
Limestone aggregate HM=3
V0=1 km/s
0
0.0 0.2 0.4 0.6 0.8 1.0
V02/2 (km/s)2

(b) 18
Forrestal et al. (1996)
Normalized relative mass loss (%)

16 Frew et al. (1998)

Present test
14 Eq. (11.27)
12
C=0.24
10

2
V0=1 km/s
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
V02/2 (km/s)2

(c) 400
Test data on UHP-CASFRC targets
Eqs. (11.27)~(11.28)
Forrestal et al. (1996)
300
DOP (mm)

200

100

0
400 500 600 700 800 900 1000
V0 (m/s)
494 11 UHPCC Targets Under Projectile Impact

(0.176 s2/km2) and limestone (0.076 s2/km2) aggregated concrete are derived by
a reduction factor of 7/9.5 and 3/9.5, respectively. It indicates that the pre-
dictions by Eq. (11.27) have good consistency with the test data. Therefore, the
validation of the mass loss models (Eq. 11.27) is further verified. In Fig. 11.60a, b,
the predicted lines are slightly higher than those of the experimental data, espe-
cially when the striking velocity is larger. The possible reason is due to the
assumption of the friction work is totally transformed into the melting heat, and the
above friction work is calculated based on the rigid projectile penetrations, which
generally tends to magnify the theoretical predictions.
Figure 11.60c illustrates the experimental DOP of the projectiles penetrating into
the above-mentioned eight UHP-CASFRC targets, as well as the predicted curve by
the existing models. In which the black dashed line is predicted by Forrestal for-
mula (Forrestal et al. 1996) where the projectiles are assumed as rigid bodies, and
the solid blue line is obtained by Eq. (11.28), with the mass loss and nose-blunting
coefficients C = 0.24 s2/km2 and k 0 ¼ 0:16 s2 =km2 , respectively. It indicates that
for the UHP-CASFRC targets with the corundum size less than d, Eqs. (11.27) and
(11.28) give better predictions of DOP than those by Forrestal formula. Therefore,
the mass abrasion of the projectile must be considered in the analysis of the pen-
etrations into UHP-CASFRC targets.

11.8 Summary

In this chapter, the existing studies on projectile impact resistance of HSC targets
and the static and dynamic properties of UHP-SFRC material are firstly reviewed.
The resistance of UHP-BASFRC and UHP-CASFRC targets against high-speed
ogive-nosed or flat-nosed projectile penetrations is investigated systematically.
(1) In Sect. 11.4, the UHP-BASFRC with higher impact resistance, good work-
ability, low costs, and energy saving is optimally prepared under the ordinary
procedures without the curing conditions of high temperature and pressure.
The triaxial compressive tests on two batches of UHP-BASFRC specimens
(95 and 129 MPa) are conducted, and the deviatoric stress–strain curves with
high confining pressure up to 100 MPa are obtained. It indicates that the
triaxial compressive strength and ductility of UHP-BASFRC material increase
with the rising of confining pressure. Besides, the Willam–Warnke and
power-law failure criteria show excellent predictions of the triaxial compres-
sive strength envelope of high-strength concrete, and the influences of
strength, size, and compositions of the material on the failure criteria are not
pronounced.
(2) In Sect. 11.5, the high-speed ogival-nosed projectile penetration tests on
UHP-BASFRC targets with the striking velocities from 510 to 1320 m/s are
carried out. The rigid and mass abrasive projectile penetrations are observed
when the striking velocity V0  1 km/s and 1 km/s < V0  1.5 km/s,
11.8 Summary 495

respectively. The proposed models for rigid projectile penetration into

UHP-BASFRC targets based on the improved dynamic cavity expansion
model as well as the limit striking velocity causing the structural destruction of
the mass abrasive projectile proposed in Sect. 4.3 are both well validated.
(3) In Sect. 11.6, to further study the impact resistance of UHP-BASFRC targets
against projectile impact, seven shots of flat-nosed projectiles penetration test
on UHP-BASFRC targets with the broad striking velocities from 299 to
842 m/s are carried out. For rigid projectile penetrations, the previously pro-
posed model in Sect. 3.4 is further validated for different projectile nose
shapes and broad striking velocity range. Combining the cavity expansion
theory with A-T equation, a simple model for predicting DOP of an eroding
projectile impacting on concrete targets is proposed and validated.
(4) In Sect. 11.7, the static test and projectile penetration test on UHP-CASFRC
targets as well as HSC targets that are mainly used for comparison are con-
ducted. The superior impact resistance of UHP-CASFRC targets against the
projectile penetration is validated. It indicates that (i) the static strengths of
UHP-SFRC are improved by increasing of the strength (hardness), size, and
volumetric ratio of coarse aggregate, and the increasing degrees are less than
20%; (ii) larger sized coarse aggregate can help to reduce the DOP. Besides,
the harder coarse aggregate is beneficial to decrease the DOP by aggravating
the mass abrasion of the projectile; (iii) the corundum with larger size tends to
break the projectile more easily, especially when the size is larger than 1.5
times of the projectile shank diameter. The mass loss and DOP of the pro-
jectile penetrating into UHP-CASFRC targets are discussed.
Chapter 12
Concrete Structures Under
Projectile Impact

Aiming to protect persons and equipment in the protective structures, four types of
concrete structures are presented in this chapter: UHP-BASFRC/fabric composite
panels, monolithic and segmented reinforced concrete (RC) panels with a rear steel
liner, SFRHSC/steel/sandy soil-layered targets, and the rock-rubble overlays. The
impact resistances of the above composite structures against small-caliber arm or
reduce-scaled earth penetration weapons are studied experimentally and/or
numerically.

12.1 UHP-BASFRC/Fabric Composite Slabs Against

Small-Caliber Arms

12.1.1 Introduction

Persons and valuable equipment in the protective structures are threatened by the
perforated small-caliber arms and rear ejecting fragments of the concrete wall
during attacks.
Most of the existing high-speed impact tests on plain or steel fiber (bar) rein-
forced HSC were conducted on reduced scale projectile (12.6–37 mm) of earth
penetration weapons (EPW), and the detailed review can be referred in Sect. 11.1.
However, few studies focused on the impact resistance of concrete targets against
small-caliber arms. Almansa and Cánovas (1999) fired three types of small arms
(5.56, 7.62, and 12.7 mm in diameter) impacting on normal and steel fiber rein-
forced concrete panels (compressive strength of about 40 MPa) with the thickness
of 40–200 mm. A model was proposed to predict the minimal thickness to avoid

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 497
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0_12
498 12 Concrete Structures Under Projectile Impact

the perforation or scabbing, and the parameters of the model were determined by
the impact test data. Bludau et al. (2006) experimentally studied the influences of
coarse aggregate type (basalt, corundum, quartzite, recycled, and boron glass), the
maximum aggregate diameter (5–16 mm), and the grading curve of the coarse
aggregate as well as binder content on the impact resistances of the HSC panels
(90–130 MPa) impacted by the 7.62 mm hardcore bullet. Sovják et al. (2013a,
2015) and Máca et al. (2014) conducted the deformable (Pb core) and
non-deformable (steel core) 7.92-mm bullets impacting tests on the UHP-SFRC
(148–164 MPa) without coarse aggregate, HSC (69–130 MPa), FRC, (37–
38 MPa), ultra-high-performance concrete (UHPC, 110–141.9 MPa), and NSC
(43 MPa) slabs with the striking velocities of 691–720 m/s. The excellent impact
resistance of UHP-SFRC was validated, and the influences of steel fiber volu-
metric fractions (1–3%) on the DOP, the impact crater size, and the debris frag-
ment mass were studied, respectively.
To further reduce the damage of the concrete slab and block the high-speed
ejecting fragment from the rear face of the slab, pasting fabric on the rear face of the
concrete panel may be an effective approach. Vossoughi et al. (2007) conducted
12.7 mm conical-nosed projectile impact test on NSC (30–43 MPa) panels with the
front and/or rear Polypropylene or Zylon fabric liners. It was found that the rear
scabbing of the fabric-protected concrete panel was considerably reduced and the
debris was caught by the fabric. Almusallam et al. (2015) presented the responses of
rear CFRP-strengthened RC panels (49 MPa, 90 mm thick) impacted by the
hemispherical nosed projectile (40 mm in diameter, striking velocity at 92–
158 m/s). It indicated that the rear-strengthened CFRP fabric could increase the
ballistic limit by about 18%, as well as reduce the front crater damage and block the
fragments from the rear face of the panel.
In this section, the armor-piercing incendiary (API) bullet impact test on the bare
or rear fabric-strengthened UHP-BASFRC panels with the thickness of 40–130 mm
is conducted. In addition, a practical approach to predict the terminal ballistic
parameters of the bullet impacting on the UHP-SFRC without coarse aggregate and
UHP-BASFRC slabs is suggested.

12.1.2 Bullet Perforation Test

12.1.2.1 Bullets

As shown in Fig. 12.1, the in-service API bullet with 7.62 mm in diameter and
37.88 mm in length is used in the test. The weight of the bullet is 10.4 g. It consists
of a brass jacket, a lead spacer, incendiary agent filler, and a 5.15 g, ogive-nosed
hard steel core. The hard steel core (6.12 mm in diameter and 27.5 mm in length) is
made of T12A steel with a hardness of HRC 62–67 and the yield strength of
980 MPa.
12.1 UHP-BASFRC/Fabric Composite Slabs … 499

Fig. 12.1 7.62-mm API bullet a photograph (with cartridge case) and b dimensions

12.1.2.2 UHP-BASFRC Panels and Strengthening Fabrics

The mix proportion of UHP-BASFRC is listed in Table 12.1. The size range of the
basalt aggregate and the volumetric ratio of the mixing straight steel fiber are 5–
10 mm and 2%, respectively.
There are totally nineteen square UHP-BASFRC panels with dimensions of
400 mm
400 mm and nominal thicknesses of 40–130 mm in the test. The
average 28d unconfined cylindrical compressive strength and the density of
UHP-BASFRC are 106.2 MPa and 2530 kg/m3, respectively.
Carbon fiber reinforced polymer (CFRP) and ultra-high molecular weight
polyethylene fiber (UHMWPE) fabrics have been widely used to strengthen/retrofit
the structural members subjected to blast and impact loads (Wu et al. 2009;
Rodriguez-Nikl 2012; Mutalib et al. 2011; Jena et al. 2009; Ong et al. 2003).
Currently, in order to reduce the ejected fragments from the rear face of the concrete
panels, as shown in Fig. 12.2, an integral piece of squared CFRP or UHMWPE
fabric (single-layered and bidirectional) with the dimensions of 400 mm
400
mm is adhered onto the rear faces of the 10 UHP-BASFRC panels (Table 12.2).
Thickness of the CRFP and UHMWPE fabrics is 0.35 and 0.12 mm, respectively.

Table 12.1 Mix proportions of UHP-BASFRC (kg/m3)

Cement Water Fly Silica Ultra-fine Fine Basalt Fibers HRWR
ash fume slag sand
450 115 75 75 150 750 750 150 15
500 12 Concrete Structures Under Projectile Impact

(a) (b) (c)

Bullet CFRP
UHP-BASFRC or
panel UHMWPE

CFRP UHMWPE
Side view

Fig. 12.2 Strengthened panels a schematic b with rear CFRP and c with rear UHMWPE

12.1.2.3 Test Setup

Figure 12.3 illustrates the sketch of the impact test, in which the UHP-BASFRC
panel is mounted perpendicularly to the rifle by a special steel frame. In the test, the
average impact velocity of the bullets is 810 m/s. A piece of aluminum plate with
the thickness of 2 mm is placed 0.6 m behind the rear face of the UHP-BASFRC
panel to evaluate the residual impact energy. The perforation of 2-mm aluminum
plate is assumed as a fatal hit of the human body, and it is referred from Bludau
et al. (2006), in which they considered that the perforation of 1.5-mm steel sheets is
equivalent to a fatal hit of the human body.

12.1.2.4 Test Results

A total of 19 panels are divided into three groups, as listed in Table 12.2. Groups 1
to 3 denote the bare UHP-BASFRC, and UHMWPE- and CFRP-strengthened
UHP-BASFRC panels, respectively. The first and second numbers of the Shot
No. in Table 12.2 denote the group and nominal thickness of the slab in cm, e.g., 2–
8 means UHP-BASFRC panel with nominal thickness of 8 cm and strengthened by
UHMWPE fabric in Group 2.
The failure modes of the panels are listed in Table 12.2, and they are classified
as perforated the concrete target (P), perforation limit (PL), and non-perforated
(UP), and perforated both the concrete target and the rear aluminum plate (Pp). For
the unperforated panels, DOP is obtained by measuring the distance of the projectile
penetrating trajectory, where the DOP* denotes the penetrations deviated from the
normal direction to some degree.
Figure 12.4 shows the local damage of the rear face of the target in Shot 1-8, in
which the hard steel core of the bullet is still with 6.5-mm-length body embedded
inside. However, the composite panels in Shots 2-8 and 3-8 with nominal thickness
of 80 mm prevent the perforation of the bullet, and only partial delamination of the
UHMWPE fabric and little bulge of the CFRP fabric are observed, as shown in
 12.1

Table 12.2 Bullets’ impact test data

Group No. Shot No. H (mm) Response DOP (mm) Front crater Rear crater
Dcf (mm) Hcf (mm) Vcf (cm3) Dcr (mm) Hcr (mm) Vcr (cm3) / (°)
1 (Bare panels) 1-4 41.5 P – 63.5 18 18 79.0 23 40 57.7
1-5 47.5 Pp – 68.5 19 24 86.0 27 31 55.9
1-6 60.9 Pp – 57.0 26 20 66.2 22 31 53.8
1-7 67.9 Pp – 69.5 24 25 94.5 40 70 47.8
1-8 77.2 PL – 70.2 27 33 72.7 19 18 60.3
1-9 90.7 UP 43* 91.2 29 43 – – – –
1-10 95.9 UP 65.5* 71.2 28 25 – – – –
1-11 107.0 UP 33 81.7 31 43 – – – –
UHP-BASFRC/Fabric Composite Slabs …

1-13 128.2 UP 78.5* 50.5 33 18 – – – –

2 (UHMWPE protected) 2-4 43.2 Pp – 54.5 21 12 61.7 32 25 40.2
2-7 68.7 P – 82.0 29 24 30.7 21 4 28.8
2-8 78.7 UP 70.5 69.2 31 32 – – – –
2-9 83.6 UP 70 59.5 27 25 – – – –
2-10 98.5 UP 79.5* 79.0 32 25 – – – –
3 (CFRP protected) 3-4 42.4 Pp – 50.0 24 15 – – – –
3-7 67.2 Pp – 71.2 45 28 – – – –
3-8 77.7 UP 58.5* 79.2 31 33 – – – –
3-9 84.5 UP 55.5 57.7 28 16 – – – –
3-10 97.5 UP 66.5 64.5 25 25 – – – –
501
502 12 Concrete Structures Under Projectile Impact

Fig. 12.3 Sketch of the test setup

Bullet

Fig. 12.4 Critical perforation in Shot 1-8

Figs. 12.5 and 12.6. Therefore, the perforation limit of the present bare
UHP-BASFRC panel is experimentally determined as 77.2 mm, which is the actual
thickness of Slab 1-8 as shown in Table 12.2.
Figures 12.5, 12.6, and 12.7 show the after-shot damage of the panels, including
the front and rear faces of the panels. The symbols F and R written on the targets
represent the front and rear faces of the panel, and R0 denotes the rear face with the
fabric removed.
Comparing the damage of targets in Shots 1-8, 2-7, 2-8, 3-7, and 3-8, it is shown
that the perforation limits of the UHP-BASFRC/UHMWPE and
UHP-BASFRC/CFRP panels are 68.7–77.2 and 67.2–77.2 mm, respectively. It can
be derived that rear fabrics help to decrease the perforation limit of the
UHP-BASFRC panels impacted by the small-caliber arms, and the maximal
reducing magnitudes are 11 and 13% for the UHMWPE and CFRP strengthening,
respectively.
Comparing the damage at the rear face of the bare, CFRP-, and UHMWPE
fabric-strengthened slabs with the identical concrete slab thickness in Figs. 12.5, 12.6,
12.1 UHP-BASFRC/Fabric Composite Slabs … 503

Fig. 12.5 Damage of the bare UHP-BASFRC panels a 1-4F, b 1-4R, c 1-7F, d 1-7R, e 1-8F,
f 1-8R, g 1-9F, h 1-9R, i 1-10F, j 1-11F, and k 1-13F

and 12.7, it can be derived that (i) the fabric reduces the crater sizes obviously (e.g.,
Shots 1-7, 2-7, and 3-7) and (ii) the delamination area at the rear face strengthened by
the CFRP fabric is considerably less than which by the UHMWPE fabric (e.g., Shots
2-4 and 3-4).
Figure 12.8 shows the recovered hard steel core of the API bullets. It indicates
that the brass jackets of the API bullets are almost all stripped from the hard steel
core except for one steel core still clung with a half jacket in Shot 2-4. All the steel
cores endure mass abrasions to some degree, and some of them even fractured.
Figure 12.9 shows the post-test damage of the rear aluminum plate. It indicates
that the bullets perforate the plates, while the ejected concrete fragments only
induce slight damage. In this study, the persons behind the UHP-BASFRC panel
with nominal thickness of 8 cm are believed to be safe since the bullets in Shots
1-8, 2-8, and 3-8 are still embedded in the concrete panel. However, they may suffer
a fatal hit if they are protected by the panels with nominal thickness of 7 cm as the
aluminum plates in Shots 1-7 and 3-7 are perforated as shown in Fig. 12.9.
504 12 Concrete Structures Under Projectile Impact

(a) (b) (c)

Perforation & Partly

delamination

(d) (e) (f)

Perforation & Partly

delamination

(g) (h) (i)

Partly
delamination

(j) (k) (l)

Partly
delamination

(m) (n) (o)

No response

Fig. 12.6 Damage of the panels protected by UHMWPE fabric a 2-4F, b 2-4R, c 2-4R0, d 2-7F,
e 2-7R, f 2-7R0, g 2-8F, h 2-8R, i 2-8R0, j 2-9F, k 2-9R, l 2-9R0, m 2-10F, n 2-10R, and
o 2-10R0
12.1 UHP-BASFRC/Fabric Composite Slabs … 505

(a) (b) (c) (d)

(e) (f) (g) (h)

Little bulge
Little bulge

(i) (j)

No response

Fig. 12.7 Damage of the panels protected by CFRP fabric a 3-4F, b 3-4R, c 3-7F, d 3-7R, e 3-8F,
f 3-8R, g 3-9F, h 3-9R, i 3-10F, and j 3-10R

Fig. 12.8 Recovered bullets

506 12 Concrete Structures Under Projectile Impact

Fig. 12.9 Damage of the aluminum plate

12.1.3 Analysis of Terminal Ballistic Parameters

Forrestal et al. (2010) pointed out that the brass jacket and the filler of bullets had a
relatively small effect on the perforation process, and impact resistance of the hard
steel core dominated the bullet perforation. Thus, only the hard core of the API
bullet is discussed in this section. There are no proper models to predict the terminal
ballistic parameters of small-caliber arms impacting on the UHP-SFRC without
coarse aggregate and the present UHP-BASFRC target. Therefore, we try to esti-
mate the terminal ballistic parameters of the bullet by the following four existing
models.
The first two models are the modified NDRC formula (Li et al. 2005; NDRC
1946) and Kar model (Kar 1978), as introduced in Chap. 2. The other two models
are listed as follows.
Chen et al. (2008b) proposed the formula to predict the perforation limit:
8 qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
< hpen ¼ ð4k=pÞð1 þ kp=4NÞ for
hpen
k
d
h ð1=I0 þ 1=NÞ i d
ð12:1aÞ
: hpen ¼ 2 ln ð1 þ I0 =NÞ þ k for
hpen
[k
d p 1 þ kp=4N d

MV02 M
I0 ¼ ; N¼ ð12:1bÞ
Sfc d 3 N1 q0 d 3
12.1 UHP-BASFRC/Fabric Composite Slabs … 507

8 pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
pffiffi
< hper ¼ hpen þ 1 þ 3Sðhpen =kdÞ tan /1 for hpen
k
2 tan /
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
d d
pffiffi
d
ð12:1cÞ
: hper hpen 1 þ 3S tan /1 hpen
d ¼ d þ 2 tan / for d [k

where k = 0.707 + l0/d and l0 is the length of the hard core nose, and N1 is the nose
shape factor which can be referred in Chen et al. (2008b) for different nose-shaped
projectiles. / ¼ 55 is determined according to the average value of the test data
listed in Table 12.2. For the relatively thick target in the present test, the ballistic
limit VBL and residual velocity Vr can be obtained reversely from Eqs. (12.1a–c) as
follows:
rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  
Sfc d 3 IBL kp 2Np ðHdHdcr kÞ
VBL ¼ ; IBL ¼ Nþ e N ð12:1dÞ
M 4

2 0:5
Vr ¼ ðV02  VBL Þ ð12:1eÞ

where H is the thickness of the given slab and Hcr is the depth of rear crater.
We proposed an empirical approach for predicting the terminal ballistic
parameters of projectile perforating finite thickness concrete panel (Wu et al. 2012a,
2015a). The residual velocity Vr is expressed as follows:
rffiffiffiffiffiffiffiffiffiffiffiffiffi
Sfc d 3 Ir
Vr ¼ ; Ir ¼ I0  KðI0 ÞH=d;
M

KðI0 Þ ¼ 1:99  1:52 expðI0 =15:11Þ; S ¼ 127:7ðfc =106 Þ0:675 ð12:2Þ

In addition, the perforation limit hper could be solved through Eq. (12.2) by
setting Ir = 0. Equations (12.1a–e) can be applied for diversified projectile nosed
shapes, including truncated-ogive nose, hemispherical nose, conical nose, and flat
nose. It should be pointed that Eq. (12.2) is only suitable for ogive-nosed
projectiles.
The available small-caliber projectile impacting tests on UHP-SFRC targets are
limited. Sovják et al. (2015) conducted a set of truncated-ogive-nosed bullet
impacting test on 45-mm-thick UHP-SFRC panels without coarse aggregate, and
the experimental and predicted residual velocities are listed in Table 12.3.

Table 12.3 Residual velocities of the bullet in Sovják et al. (2015)

fc (MPa) V0 (m/s) Vr (m/s) Predicted Vr (m/s)
Chen et al. (2008b) Wu et al. (2015a)
132 691 260 223.5 (−14.0%) 300.8 (15.7%)
154 705 226 201.1 (−11.0%) 303.1 (34.1%)
149 703 228 209.7 (−8.0%) 304.1 (33.4%)
508 12 Concrete Structures Under Projectile Impact

Máca et al. (2014) also obtained the perforation limit of the bullet impacting on
UHP-SFRC panels without coarse aggregate under the striking velocity of 710 m/s,
and the experimental and predicted perforation limits for the experiments in Máca
et al. (2014) and present test are listed in the first two rows of Table 12.4.
For predicting the bullet impact resistance of UHP-BASFRC targets, compared
with the UHP-SFRC target without coarse aggregate, it is difficult to study the
enhancing degree induced by adding coarse aggregate experimentally. With the
consideration of the effect of coarse aggregate, Whiffen (1943) proposed a formula
for dimensionless DOP
   0:1  97:51
hpen 2:61 M d V0 fc0:25
¼ pffiffiffiffi ð12:3Þ
d fc d3 da;max 533:4

where da,max is the maximum aggregate size and the application ranges are given in
Sect. 2.2. At follows, the influence of coarse aggregate on the impact resistance is
discussed according to Eq. (12.3). It could be derived from Eq. (12.3) that with the
other identical parameters of both projectiles and targets, the projectile penetration
depth of UHP-BASFRC targets with basalt coarse aggregate (da,max = 10 mm) is
nearly 77.7% of that of UHP-SFRC targets with only fine sand (da,max = 0.8 mm).
Considering the above reduction, the corresponding predicted perforation limits of
the present test are listed in the 3rd row of Table 12.4, in which the superscript * is
added to distinguish from the data of UHP-SFRC targets without coarse aggregate
in the 2nd row. It should be noted that two assumptions are made during the above
calculations: (i) The enhancing degree of adding coarse aggregate given by Eq. (12.
3) for NSC targets is also suitable for UHP-SFRC targets and (ii) the enhancing
effect by adding coarse aggregate is identical to DOP and perforation limit.
Besides, aiming to further validate the distinguished impact resistance of the
present UHP-BASFRC target, another impact scenario is assumed as shown in the
last row of Table 12.4, in which the UHP-SFRC (151.7 MPa) panels without coarse
aggregate in Máca et al. (2014) were struck by the 7.62-mm API bullet in this test.
From Tables 12.3 and 12.4, it is derived that:
(i) As for UHP-SFRC targets without coarse aggregate in Sovják et al.
(2015) and Máca et al. (2014), Chen’s model agrees well with the test data,
while Wu’s model gives over predictions. The reason lies in that Eq. (12.2)
was obtained by fitting the ogive-nosed projectile perforation test data, which
may overestimate the perforation capacity of the truncated-ogive-nosed
projectile; the modified NDRC and Kar formulae both show poor prediction
accuracies.
(ii) As for UHP-BASFRC targets, the existing four models all give poor pre-
dictions. When the enhancing effect induced by adding coarse aggregate is
considered, Chen’s model also shows good predictions (the 3rd row in
Table 12.4). Thus, Chen’s model given in Eqs. (12.1a–e) with a reduction
coefficient of 0.777 can be used to predict the terminal ballistic parameters of
the bullet impacting on UHP-BASFRC targets.
 12.1

Table 12.4 Perforation limits of the UHP-SFRC and UHP-BASFRC panels

Targets fc (MPa) M (g) V0 (m/s) Nose shape Perforation limit (mm)
Exp. Chen et al. (2008b) Wu et al. (2015a) NDRC (1946) Kar (1978)
UHP-SFRC Máca et al. (2014) 151.7 3.65 710 Truncated-ogive 50 47.6 (−4.9%) 55.8 (11.6%) 33.4 (−33.2%) 33.8 (−32.4%)
UHP-BASFRC/Fabric Composite Slabs …

UHP-BASFRC 106.2 5.15 810 Ogive 77.2 99.7 (29%) 90.9 (17.7%) 60.2 (−22%) 65.2 (−15.5%)
UHP-BASFRC 106.2 5.15 810 Ogive 77.2 77.5* (0.4%) 70.6* (−8.5%) 46.8* (−39%) 50.7* (−34%)
UHP-SFRC 151.7 5.15 810 Ogive – 90.41 – – –
509
510 12 Concrete Structures Under Projectile Impact

(iii) In the last row of Table 12.4, the predicted perforation limit of 90.41 mm is
obtained by Chen’s model. By comparing the predicted results in the second
row and fourth row, for the same bullet and strike velocity, the perforation
limit of the present UHP-BASFRC (77.2 mm) panel is nearly 14.6% lower
than which of the UHP-SFRC (90.41 mm) panel, even though the com-
pressive strength (106.2 MPa) of the former panel is only 70% of the latter
panel (151.7 MPa).

12.2 Monolithic and Segmented RC Slabs

with a Rear Steel Liner

12.2.1 Introduction

For both ground and underground protective structures subjected to projectile

impact, a thin metal (e.g., steel) plate is commonly attached onto the rear face of the
concrete shield. The function of the frontal RC slabs is to resist the projectile
penetration, while the rear steel liner is to catch the scabbing fragment to protect the
persons and equipment inside the structures from injury or damage.
In the existing experimental studies on the projectile perforating the plain, steel
bar reinforced, and fiber reinforced concrete panels, Dancygier (1997, 1998), Beppu
et al. (2008), Mohamed and Ahmed (2014), and Almusallam et al. (2013) experi-
mentally evaluated the impact resistances of concrete slabs by examining the visible
damages (e.g., crater, penetration depth, and perforation) of the targets. Besides,
many researchers quantitatively evaluated the impact resistance of concrete slabs by
analyzing the terminal ballistic parameters of the projectile such as residual velocity
and deceleration, as introduced in Sects. 3.5 and 3.8.
For the RC slab with a rear steel liner (denoted as RC/steel targets), the projectile
perforation tests conducted by Kojima (1991), Mohamed and Ahmed (2014),
Tsubota et al. (1993), Tetsuo et al. (1997), and Hashimoto et al. (2005) indicated
that the rear steel liner could restrain the damage at the rear face and improve the
perforation resistance of the composite target efficiently. By conducting the com-
parative projectile impact tests on the monolithic RC slabs with and without the rear
steel liner as well as the spaced double-layered RC slabs, Kojima (1991) showed
that the impact resistance of the double-layered RC target was inferior to which of
the monolithic RC slab with the same overall thickness. Tsubota et al. (1993) fur-
ther proposed that the double-layered spaced RC target with the thicker slab at the
rear could obtain more impact resistance.
Assuming the rear steel liner as an additional reinforcement to the target, Barr
(1990) proposed a formula to predict the ballistic limit of the projectile perforating
the RC/steel composite target. Grisaro and Dancygier (2014b) proposed an
approach to assess the perforation resistance of the RC/steel composite target
12.2 Monolithic and Segmented RC Slabs … 511

impacted by non-deforming projectiles by converting the rear steel liner to an extra

certain depth of concrete slabs with the same ballistic limit.
Investigations on the impact resistances of the multilayered RC/steel composite
targets are helpful in the arrangement and optimization of the anti-strike shield of
protective structures. However, most of the existing tests on the terminal ballistic
performances of the projectile were mainly concentrated on the single-layered bare
reinforced or plain concrete slabs without a rear steel liner. The related quantitative
experimental studies on the impact resistances of the segmented RC/steel composite
targets, as well as the influences of the target arrangement (stacked or spaced,
orders, etc.) on the impact resistances, are very limited.
In this section, twenty-five shots of hard projectile impacting tests on monolithic
and segmented RC/steel composite targets with five different arrangements under
the same total thickness are conducted. The impact resistances of single- and
multilayered bare RC and RC/steel composite targets are discussed in detail. An
explicit dimensionless expression for predicting the terminal ballistic parameters of
the projectile perforating the RC slab is proposed and validated by the present and
existing test data. In addition, the impact resistances of RC/steel composite targets
are analyzed by the models proposed by Grisaro and Dancygier (2014b).

12.2.2 Test

12.2.2.1 Projectile

The reduce-scaled projectiles used in the experiments are made of 45CrNiMoV

steel rods with the hardness of HRC45 and the yield strength of 1420 MPa, shown
in Fig. 12.10a. The mass of the projectile is 386 g, and the geometry and dimension
of the projectile are given in Fig. 12.10b, where the shank diameter is 25.3 mm,
projectile cartridge thickness is 3.65 mm, and the caliber-radius-head (CRH) of the
ogival nose is 3.
A small-caliber acceleration data recorder, shown in Fig. 12.10a, is mounted
inside the projectile to record the acceleration-time history of the projectile. The
recorder is 18 mm in diameter, 68.5 mm in length, and 42 g in weigh. The sam-
pling amplitude and frequency of the recorder are 2.5
105g and 1.05
105 Hz,
respectively. During each shot, the mass and length of the unfired projectile (M0 and
L0), as well as the residual mass and length of the recovered projectile (Mr and Lr),
are measured and listed in Table 12.5, in which the mass of the recorder is included.

12.2.2.2 Targets

There are fifty RC slabs with four different thicknesses (100, 150, 200 and 300 mm)
in the test. Figure 12.11a–c shows the geometry and the reinforcement mesh of the
RC slabs, in which the diameter of the steel bars is 6 mm and the reinforcement
512 12 Concrete Structures Under Projectile Impact

Fig. 12.10 Projectile and accelerometer a photograph and b geometry and dimensions

ratio in each way is about 0.5%. The impact point of the projectile is plotted
by “
”.
As shown in Fig. 12.11d, a steel plate with the thickness of 1 mm is welded by
four stud bolts onto the rear face of the last RC slab in the segmented target. The
unconfined cylinder compressive strength of concrete is 41 MPa, and the maximum
diameter of the gravel coarse aggregate is less than 10 mm.

12.2.2.3 Test Setup

As listed in Table 12.5, there are five arrangements of monolithic and segmented
RC/steel composite targets with the equal total thickness of 300 mm (the last RC
panel attached with a rear steel liner) in this test. For example, the expression
“100 + 100 + 100(1)” refers to three RC slabs with the same thickness of 100 mm
and an attached steel liner with the thickness of 1 mm welded on the rear face of the
last RC slab. The air spaces between each slab range from 200 to 300 mm for the
spaced segmented targets. Besides, in order to find the difference of the impact
resistance between the stacked and spaced RC slabs under the same projectile
impact velocity, the test on the stacked segmented targets is also conducted, where
the shot number is denoted by a superscript s (e.g., 2–4s) as shown in Table 12.5.
Figure 12.12 illustrates the sketch diagram of Test 5. The projectile is launched
by a powder gun and impacts normally on the composite targets. Twenty-five shots
are carried out with the initial striking velocity of the projectile ranged from 536 to
12.2 Monolithic and Segmented RC Slabs … 513

Table 12.5 Perforation test data

Test Shot Target M0 Mr L0 Lr V0 V1r V2r V3r
No. No. arrangement (g) (g) (mm) (mm) (m/s) (m/s) (m/s) (m/s)
1 1-1 300(1) 429 419.8 152 147 641.5 272 – –
1-2 428.3 420.9 152 147 540 0 – –
1-3 427.1 419.4 152 147 601 180 – –
1-4 428.4 417.7 152 147 679 329 – –
1-5 426.9 418.1 152 148 737 422 – –
2 2-1 100 + 200(1) 428.5 423 152 148 539 436 137 –
2-2 428.1 420.8 152 148 601 482 207 –
2-3 428.4 420.8 152 148 634 544 304 –
2-4s 429.4 420.7 152 147 643 – 269 –
2-5 428.1 420.7 152 147 729 651 451 –
3 3-1 200 + 100(1) 427.7 421.9 152 147 540 292 159 –
3-2 429.9 423.1 152 148 597 404 270 –
3-3 427.6 421.5 152 147 641 438 295 –
3-4s 428.7 420.9 152 147 638 – 286 –
3-5 427 418.5 152 147 731 527 484 –
4 4-1 150 + 150(1) 429.7 423.4 152 148 536 379 161 –
4-2 430.7 424.1 152 148 603 448 246 –
4-3 426.9 423.9 152 148 643 511 333 –
4-4s 428.8 424 152 148.2 644 – 317 –
4-5 428.9 424 152 148.3 721 620 444 –
5 5-1 100 + 100 + 100 428.9 423.6 152 148 536 423 307 182
5-2 (1) 428.8 422.9 152 147.5 607 520 465 309
5-3 427.9 420.9 152 147 640 549 486 354
5-4s 427.3 421.2 152 147.4 639 – – 311
5-5 428.3 421.3 152 147 729 665 567 482

737 m/s. The striking velocity of the projectile V0 is measured by the electric probes
shown in Fig. 12.12a. The exterior ballistic flight process and pose of the projectile
are captured by the high-speed camera system, which is trigged by the electric
probe in front of the target, as shown in Fig. 12.12a. V1r, V2r, and V3r denote the
residual velocities of the projectile after perforating the first, second, and third RC
slabs, respectively. Additionally, the acceleration-time history of the projectile in
the whole ballistic range is recorded by the onboard accelerometer.

12.2.2.4 Test Results

Figure 12.13 shows three photographs of typical instants in Shot 2-1. It indicates
that the projectile impacts and perforates the target perpendicularly and straightly,
and the ejected fragments from the rear face of the 2nd slab are much smaller than
514 12 Concrete Structures Under Projectile Impact

(a) (b)
675mm 100mm 150mm 675mm 200mm
675mm

96 .5mm

675mm

75mm
96 .5mm 64mm 57mm
75mm 82mm

(c) (d)
675mm
675mm 300mm
67 .5mm

675mm

4 stud bolts Stud

675mm

Welding
10mm 10mm

Steel
88mm 88mm 88mm
RC slab liner
67 .5mm
Front view Side view Front view Side view

Fig. 12.11 Schematic diagrams of RC slabs a 100 mm and 150 mm, b 200 mm, c 300 mm, and
d welding of the rear steel liner

that of 1st slab as the prevention of the steel liner. Table 12.5 lists the striking and
residual velocities of the projectile.
Figure 12.14 illustrates the photographs of the unfired projectile (far left one)
and several recovered projectiles after the impact test. From Table 12.5, it can be
found that the blunted lengths and mass losses of the projectiles are all less than 3.3
and 2.5%, respectively. Therefore, the projectiles in this test can be treated as rigid
bodies approximately.
Table 12.6 lists the dimensions of the front and rear craters of the targets in the
test, where Dcf, Hcf, Dcr and Hcr are the equivalent diameter and depth of the front
and rear crater. Figures 12.15, 12.16, 12.17, 12.18, 12.19, 12.20, 12.21, 12.22, and
12.23 show the after-shot damage of the targets in each configuration where the
initial striking velocities are all nearly 640 m/s. The labels 2-3-1F and 2-3-1R
represent the front and rear faces of the 1st slab in Test 2-3, respectively, and the
labels 2-3-2R(1) and 2-3-2R(0) represent the rear face of the 2nd slab in Test 2-3
with the rear steel liner and with the rear steel liner removed, respectively.
The deceleration-time histories of the projectile in Shots 1-1, 1-2, 1-3, 2-1, 3-1,
and 3-2 are recorded within the whole ballistic range, including the projectile
accelerated in the barrel, flight in the air, decelerated in the monolithic or segmented
12.2 Monolithic and Segmented RC Slabs … 515

(a) Recover device

Steel frame
V3r V2r V1r Camera trigger V0

Projectile Powder gun

Electric probes
Steel liner RC slab

Bullet-proof glass

Thickened brick wall

High-speed camera

(b)

Fig. 12.12 Sketch of test a setup and b constraints of the target

target, and after-perforation flight, as shown in Fig. 12.24. The detailed discussions
on the accuracy of the acceleration data are given in Sect. 3.8.

12.2.3 Analysis and Discussion

12.2.3.1 Influence of the Slab Arrangement on Impact Resistance

Figure 12.25 illustrates the residual velocities of the projectile after perforating the
above stacked and spaced segmented targets with the same total thickness and
projectile impact velocity. It indicates that the residual velocities of the projectile
after perforating the stacked targets are lower than that of the spaced targets, which
means that for the same overall thickness, the stacked segmented targets have
higher impact resistance than the spaced segmented targets.
516 12 Concrete Structures Under Projectile Impact

(a)
Steel frame 2nd RC slab 1st RC slab

Projectile

(b)
Steel frame 2nd RC slab 1st RC slab

Projectile

(c)

Steel frame 2nd RC slab 1st RC slab

Projectile

Fig. 12.13 Photographs of typical instants in Shot 2-1 a projectile impacting the 1st slab,
b projectile impacting the 2nd slab, and c projectile perforating the 2nd slab

Furthermore, according to Figs. 12.16, 12.17, 12.18, 12.19, 12.20, 12.21, 12.22,
and 12.23, the stacked segmented targets exhibit less damage occurred at the
contact faces compared with the spaced segmented target. Only a hole with a
diameter of nearly the same as the projectile shank diameter is observed, e.g.,
2-4-1R and 2-4-2F.
Figure 12.26 illustrates the relationship between striking velocity V0 and residual
velocity Vr of the projectile perforating the single and spaced segmented targets
with the same total thickness. Within the range of the experimental parameters, it
indicated that (i) the impact resistance of the monolithic concrete slab is better than
that of the segmented target; (ii) the more the layers of the segmented target, the
12.2 Monolithic and Segmented RC Slabs … 517

more the inferiority of the segmented target has; (iii) when the number of the layers
is fixed, e.g., two layers in the present test, the impact resistance of the segmented
target is dependent on placing the order of the slabs, which is enhanced by putting
the thicker slab at the rear position. However, Ben-Dor et al. (2009) theoretically
found that (i) the ballistic limit of the multilayered concrete shield does not depend
on the placing order of the slabs in the shield; (ii) the largest decrease in the ballistic
limit occurs when a shield is divided into a number of slabs with the same
thickness.

12.2.3.2 Experimental Approach to Determine the Ballistic Limit

From Table 12.5, we can also obtain the striking and residual velocities of pro-
jectile perforating monolithic bare RC and RC/steel composite targets, illustrated in
Fig. 12.27. It indicates that the improvement in the rear steel liner on enhancing the
impact resistance of the RC/steel composite targets is not obvious within the dis-
cussed parametric ranges in this experiment. Meanwhile, the almost linear depen-
dence of the Vr–V0 relationship tends to be hyperbolic with the thickness of the RC
slab increasing gradually, no matter the rear steel liner exists or not. Thus, the
ballistic limit of the projectile is difficult to be determined directly from the test data
of Vr–V0.
Figure 12.28 shows the relationships between the residual impact functions Ir
and the initial impact functions I0, where I0 ¼ M0 V02 =Sfc d 3 and Ir ¼ M0 Vr2 =Sfc d 3 .
The values of the related parameters are fc = 41 MPa, M0 = 0.428 kg,
d = 25.3 mm, and S ¼ 127:7
fc0:675 (fc unit in MPa) (Wu et al. 2012a). It can be
seen obviously that Ir–I0 relationships are almost linear for both monolithic bare RC
and RC/steel composite targets, which does not depend on the thickness of the
panel. Thus, the ballistic limit VBL can be predicted straightforward by fitting the
test data linearly (e.g., IBL = 19.1 m/s and VBL = 555 m/s for the
300 mm-depth-RC slab, where IBL ¼ M0 VBL 2
=Sfc d 3 is the intersection of the fitted
straight line and the horizontal axis, as shown in Fig. 12.28a).

Fig. 12.14 Photographs of unfired and recovered projectiles

518 12 Concrete Structures Under Projectile Impact

Table 12.6 Dimensions of the front and rear craters in the test
Test No. Shot No. Front crater (mm) Rear crater (mm)
dv dh d1 d2 Dcf Hcf dv dh d1 d2 Dcr Hcr
1 1-1 360 300 400 320 345 55 370 380 440 450 410 55
1-2 330 290 360 410 347.5 55 Penetration depth 254 mm
1-3 360 380 440 360 385 60 450 380 550 540 480 70
1-4 290 300 290 300 295 60 360 480 440 480 440 65
1-5 410 400 390 470 417.5 57 400 400 440 500 435 60
2 2-1 1st slab 210 220 230 260 230 35 190 210 210 240 212.5 40
2nd slab 300 215 270 360 286.2 50 350 220 380 410 340 70
2-2 1st slab 240 230 260 190 230 45 210 170 180 220 195 55
2nd slab 215 260 280 280 258.8 45 250 340 360 400 337.5 65
2-3 1st slab 210 240 230 230 227.5 50 230 230 210 240 227.5 50
2nd slab 220 270 210 260 240 50 400 270 350 410 357.5 75
2-4 1st slab 200 200 230 200 207.5 65 Contact
2nd slab Contact 400 330 480 420 407.5 70
2-5 1st slab 220 180 250 230 220 45 260 200 230 200 222.5 45
2nd slab 250 250 280 270 262.5 55 400 360 320 310 347.5 60
3 3-1 1st slab 310 340 400 390 360 50 340 400 310 360 352.5 65
2nd slab 130 130 130 160 137.5 45 210 170 180 210 192.5 50
3-2 1st slab 360 290 360 440 362.5 55 260 340 290 370 315 70
2nd slab 120 150 150 160 145 50 150 170 210 230 190 50
3-3 1st slab 290 240 310 260 275 60 210 240 250 270 242.5 65
2nd slab 200 170 160 180 177.5 50 230 230 250 230 235 50
3-4 1st slab 220 220 210 240 222.5 50 Contact
2nd slab Contact 320 270 300 330 305 80
3-5 1st slab 350 260 350 320 320 60 240 300 310 290 285 60
2nd slab 220 250 230 260 240 50 210 150 200 200 190 50
4 4-1 1st slab 250 290 250 280 267.5 50 250 260 300 320 282.5 65
2nd slab 170 190 180 210 187.5 45 330 240 220 300 272.5 70
4-2 1st slab 270 300 370 210 287.5 50 310 300 300 300 302.5 60
2nd slab 200 250 220 240 227.5 55 270 280 330 270 287.5 60
4-3 1st slab 230 200 200 230 215 50 140 230 180 220 192.5 50
2nd slab 230 230 240 220 230 50 310 260 320 370 315 60
4-4 1st slab 200 260 270 220 237.5 50 Contact
2nd slab Contact 300 400 300 350 337.5 60
4-5 1st slab 220 190 270 200 220 50 190 310 260 230 247.5 55
2nd slab 200 190 180 200 192.5 50 240 320 280 300 285 55
(continued)
12.2 Monolithic and Segmented RC Slabs … 519

Table 12.6 (continued)

Test No. Shot No. Front crater (mm) Rear crater (mm)
dv dh d1 d2 Dcf Hcf dv dh d1 d2 Dcr Hcr
5 5-1 1st slab 190 150 200 210 187.5 50 210 210 240 240 225 50
2nd slab 150 140 150 160 150 50 280 270 300 290 285 50
3rd slab 130 140 130 140 135 50 260 200 270 270 250 50
5-2 1st slab 230 200 250 250 232.5 50 140 180 170 180 167.5 50
2nd slab 190 150 170 170 170 50 230 190 280 270 242.5 50
3rd slab 200 200 200 200 200 50 190 130 220 220 190 50
5-3 1st slab 190 200 230 170 197.5 50 280 200 190 200 217.5 50
2nd slab 180 180 190 170 180 50 200 250 240 200 222.5 50
3rd slab 160 150 160 150 155 50 220 200 250 230 225 50
5-4 1st slab 220 210 250 210 222.5 40 Contact
2nd slab Contact Contact
3rd slab Contact 340 300 420 400 365 60
5-5 1st slab 190 210 190 220 202.5 50 190 240 200 200 207.5 50
2nd slab 130 150 150 170 150 50 260 210 280 220 242.5 50
3rd slab 180 160 170 180 172.5 50 220 210 230 240 225 50

Fig. 12.15 Damages of the composite targets in Shot 1-1 a target configuration, b 1-1F, c 1-1R
(1), and d 1-1R(0)

12.2.3.3 Proposed Approach for Terminal Ballistic Parameters

In this section, we will propose an approach to predict the terminal ballistic

parameters of the projectile perforating the RC target. According to Table 12.5,
Fig. 12.29a, b shows the relationships between the residual velocity and the target
thickness Vr–H as well as its dimensionless form of the residual impact function and
dimensionless target thickness Ir–H/d, respectively, in which the data for the RC
slab with the thickness of 100 mm are the average value of Tests 2 and 5.
Figure 12.29b shows that the relationship of Ir–H/d is almost linear for a given
I0. Assuming that the slopes of the fitted straight lines are only dependent on the
initial impact function I0; thus, the residual impact function Ir can be expressed as
follows:
520 12 Concrete Structures Under Projectile Impact

Fig. 12.16 Damages of the composite targets in Shot 2-3 a target configuration, b 2-3-1F,
c 2-3-1R, d 2-3-2F, e 2-3-2R(1), and f 2-3-2R(0)

Fig. 12.17 Damages of the composite targets in Shot 2-4s a target configuration, b 2-4-1F,
c 2-4-1R, d 2-4-2F, e 2-4-2R(1), and f 2-4-2R(0)
12.2 Monolithic and Segmented RC Slabs … 521

Fig. 12.18 Damages of the composite targets in Shot 3-3 a target configuration, b 3-3-1F,
c 3-3-1R, d 3-3-2F, e 3-3-2R(1), and f 3-3-2R(0)

Fig. 12.19 Damages of the composite targets in Shot 3-4s a target configuration, b 3-4-1F,
c 3-4-1R, d 3-4-2F, e 3-4-2R(1), and f 3-4-2R(0)
522 12 Concrete Structures Under Projectile Impact

Fig. 12.20 Damages of the composite targets in Shot 4-3 a target configuration, b 4-3-1F,
c 4-3-1R, d 4-3-2F, e 4-3-2R(1), and f 4-3-2R(0)

Fig. 12.21 Damages of the composite targets in Shot 4-4s a target configuration, b 4-4-1F,
c 4-4-1R, d 4-4-2F, e 4-4-2R(1), and f 4-4-2R(0)
12.2 Monolithic and Segmented RC Slabs … 523

Fig. 12.22 Damages of the composite targets in Shot 5-3 a target configuration, b 5-3-1F,
c 5-3-1R, d 5-3-2F, e 5-3-2R, f 5-3-3F, g 5-3-3R(1), and h 5-3-3R(0)

Fig. 12.23 Damages of the composite targets in Shot 5-4s a target configuration, b 5-4-1F,
c 5-4-1R, d 5-4-2F, e 5-4-2R, f 5-4-3F, g 5-4-3R(1), and h 5-4-3R(0)

Ir ¼ I0  KðI0 ÞH=d ð12:4aÞ

The expression of KðI0 Þ can be obtained by data fitting of the existing and
present projectile perforation tests on monolithic RC panels. Figure 12.29c shows
the test data of K(I0)–I0 and the corresponding fitted curve, where the expression of
K(I0) is given in Eq. (12.4b).
524 12 Concrete Structures Under Projectile Impact

(a) 20 (b) 15.0

12.5
15
10.0
Deceleration (10 g)

Deceleration (104g)
10
4

7.5

5 5.0

2.5
0
0.0

-5 -2.5

-5.0
-10
-7.5
0 5 10 15 20 25 30 0 5 10 15 20 25 30 35 40
Time (ms) Time (ms)

(c) 20 (d) 17.5

15.0
15
12.5

Deceleration (10 g)
4
Deceleration (104g)

10.0
10
7.5
5 5.0
2.5
0
0.0
-2.5
-5
-5.0
-10 -7.5
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
Time (ms) Time (ms)

(e) 25 (f) 20
20
15
Deceleration (10 g)

15
4

Deceleration (104g)

10
10

5 5

0
0
-5
-5
-10

-15 -10
0 5 10 15 20 25 30 0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
Time (ms) Time (ms)

Fig. 12.24 Deceleration-time histories of projectiles in Shots a 1-1, b 1-2, c 1-3, d 2-1, e 3-1, and
f 3-2

KðI0 Þ ¼ 1:52 exp ðI0 = 15:11Þ þ 1:99 ð12:4bÞ

The comparisons between the existing and present test data, as well as the
predicted results by Eqs. (12.4a, b) and the model proposed by Grisaro and
Dancygier (2014a) are illustrated in Fig. 12.30. It indicates that the predicted
residual velocities by Eqs. (12.4a, b) agree better with the test data, especially for
the thicker target. It should be pointed out that relatively large deviations occur in
Fig. 12.30b which are predicted by Grisaro and Dancygier (2014a) compared with
Fig. 5 in Grisaro and Dancygier (2014a). The reason is that the ballistic limit VBL
12.2 Monolithic and Segmented RC Slabs … 525

Fig. 12.25 Residual 360

5-3
velocities of stacked and V0=640m/s
350
spaced segmented targets Stacked segmented targets
340 Spaced segmented targets 4-3
330
s
4-4

Vr (m/s)
320
s
5-4
310 2-3
300 3-3
s
290 3-4
280 s
2-4
270
260
1 2 3 4 5 6
Test No.

used in the present calculations is obtained by NDRC (1946) formula and by data
fittings in Grisaro and Dancygier (2014a), respectively.

12.2.3.4 Impact Resistance of RC/Steel Targets

Based on the existing perforation models for concrete slabs (Li and Chen 2003;
NDRC 1946) and steel slabs [BRL (Alco Products Inc. 1955; Rosenberg and Dekel
2010)], Grisaro and Dancygier (2014b) proposed an approach for predicting the
ballistic limit of the projectile perforating the monolithic RC slab with a steel liner.
In this approach, the rear steel liner is firstly equivalent to a certain thickness of the
concrete slab with the same ballistic limit. Then, the ballistic limit of the RC/steel
composite target can be predicted from a relatively thicker uniform RC slab, with
the thickness equaling to the sum of the original RC slab thickness and the above
equivalent thicknesses of rear steel liner.
Figure 12.31a shows the present test data of V0–H for the monolithic RC slab
with a steel liner, and the predicted curves for the ballistic limit VBL by the four
combinations of the existing models in Grisaro and Dancygier (2014b). It should be
noted that the solid parts of the lines correspond to the validated ranges of the
models and the dashed parts of the lines are the extrapolations beyond their vali-
dated ranges. Correspondingly, based on the ballistic limits VBL given in
Fig. 12.31a and the conversation of the kinetic energy of the projectile, the
dependence curves of Vr–V0 for different depths of targets can be obtained by
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Vr ¼ V02  VBL 2 and illustrated in Fig. 12.31b–e.

It can be obviously seen that the predicted ballistic limit and the residual
velocities by “NDRC + BRL” and “NDRC + Rosenberg” models agree better with
the test data. Also, the predicted values by the above four combinations tend to be
closer with the target thickness decreasing gradually. It should be pointed out that
the perforation limit in the above two combinations “Li + BRL” and
526 12 Concrete Structures Under Projectile Impact

Fig. 12.26 Vr–V0 (a) 500

relationship between With a steel liner
450 200mm
projectile perforating single
100mm+100mm
and multilayered RC targets 400
a total thickness 200 mm with
a steel liner, b total thickness 350

Vr (m/s)
200 mm without a steel liner, 300
and c total thickness 300 mm
with a steel liner 250

200

150

100
400 450 500 550 600 650 700
V0 (m/s)

(b) 600
Without a steel liner
550 200mm
100mm+100mm
500
Vr (m/s)

450

400

350

300

250
500 550 600 650 700 750
V0 (m/s)

(c) 500

400
Vr (m/s)

300

200
300mm
100mm+200mm
100 150mm+150mm
Unperforated 200mm+100mm
100mm+100mm+100mm
0
525 550 575 600 625 650 675 700 725 750
V0 (m/s)
12.2 Monolithic and Segmented RC Slabs … 527

(a) 500 (b) 550 200mm-depth RC slab with steel liner

300mm-depth RC slab with steel liner 500 200mm-depth RC slab without steel liner
400
450

400
300
Vr (m/s)

Vr (m/s)
350

300
200
250

100 200

Unperforated 150

0 100
500 550 600 650 700 750 400 450 500 550 600 650 700 750
V0 (m/s) V0 (m/s)

(c) 650
150mm-depth RC slab with steel liner
(d) 700 100mm-depth RC slab with steel liner
600 650
150mm-depth RC slab without steel liner 100mm-depth RC slab without steel liner
550 600

500 550
500
Vr (m/s)
Vr (m/s)

450
450
400
400
350
350
300
300
250 250
200 200
150 150
350 400 450 500 550 600 650 700 750 250 300 350 400 450 500 550 600 650 700 750 800
V0 (m/s) V0 (m/s)

Fig. 12.27 Vr–V0 relationship between projectile perforating monolithic bare RC and RC/steel
composite targets with the depth of a 300 mm, b 200 mm, c 150 mm, and d 100 mm

“Li + Rosenberg” is also obtained from NDRC formulae, and only the penetration
depths of the projectile are predicted by Li and Chen (2003) formulae.

12.3 Hard Projectile Impact on Layered SFRHSC

Composite Targets

12.3.1 Introduction

The protective shields against projectile impact are commonly multilayered struc-
tures, e.g., concrete/steel liner/sandy soil composite targets. The function of frontal
concrete layer is to resist the striking projectile, while the steel plate layer is to
block the scabbing fragments, and the function of the sandy soil layer is to disperse
the intensive stress wave induced by the projectile (Wu et al. 2015e). However, the
related experimental studies on the impact resistance of the above-mentioned
composite structures are limited. In this section, a series of hard projectile
high-speed (500–700 m/s) impact tests on the composite concrete target “SFRHSC
panel + rear steel liner + backfilled sandy soil” are conducted. The impact resis-
tance of the layered composite concrete target and the parametric influences, such
528 12 Concrete Structures Under Projectile Impact

(a) 15 (b) 20 200mm-depth RC slab with steel liner

300mm-depth RC slab with steel liner 200mm-depth RC slab without steel liner
Fitted straight line Fitted straight line
12
15

Ir
10
Ir

5
3
Unperforated
IBL=19.1, VBL= 555m/s IBL=10.6, VBL= 414m/s

0 0
16 20 24 28 32 36 10 15 20 25 30 35
I0 I0

(c) 25 150mm-depth RC slab with steel liner

(d) 30 100mm-depth RC slab with steel liner
150mm-depth RC slab without steel liner 100mm-depth RC slab without steel liner
25 Fitted straight line
20 Fitted straight line

20
15
Ir

15
Ir

10
10

5
5
IBL=8.39, VBL= 368m/s IBL=4.64, VBL= 274m/s
0 0
5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
I0 I0

Fig. 12.28 Ir–I0 relationship between projectile perforating monolithic bare RC or RC/steel
composite targets with the depth of a 300 mm, b 200 mm, c 150 mm, and d 100 mm

as the compressive strength of the concrete (47–100.8 MPa), incorporating steel

fiber volume fraction (0 and 3%), steel bar reinforced or plain concrete, and the rear
layers (steel liner and the sandy soil), are assessed and discussed.

12.3.2 Projectile Impact Test

12.3.2.1 Projectile

As shown in Fig. 12.32a, the solid semi-armor-piercing projectiles used in the

penetration test are made of 35GrMnSi steel rods with the yield strength and
hardness of 1620 MPa and HB241, respectively. The scheme and dimensions of the
projectile’s longitudinal section are given in Fig. 12.32b. The projectile
length-to-diameter ratio is 3.5, and the average mass of the projectile is 693 g.
During the test, the projectiles are launched by a 37-mm-diameter smooth-bore
powder gun at the striking velocities of 500–700 m/s.
12.3 Hard Projectile Impact on Layered SFRHSC Composite Targets 529

Fig. 12.29 Test data of a Vr– (a) 800

H, b Ir–H/d, and c K(I0)–I0 V0=538m/s
700 V0=601m/s
600 V0=640m/s
V0=729m/s
500

Vr (m/s)
400

300
V0
200

100

0
0 50 100 150 200 250 300
H (mm)

36
(b) V0=538m/s
32 V0=601m/s
28 V0=640m/s
V0=729m/s
24
Ir=I0-1.48 H/d
20 Ir=I0-1.61 H/d
Ir

16 Ir=I0-1.65 H/d
12 Ir=I0-1.74 H/d

8
I0
4

0
0 2 4 6 8 10 12 14 16 18
H /d

(c) 2.50
2.25

2.00

1.75
K(I0 )

1.50
Hanchak et al. (1992) (48MPa)
1.25 Hanchak et al. (1992) (140MPa)
Cargile et al. (1993)
1.00 Li et al. (2013)
200mm-depth slab
0.75 150mm-depth slab
100mm-depth slab
0.50
0 10 20 30 40 50 60 70 80 90
I0
530 12 Concrete Structures Under Projectile Impact

(a) 1200 Hanchak et al. (1992) (48MPa)

(b) 1000 Hanchak et al. (1992) (48MPa)
1000 Eq. (12.6) Eq. (12.6)
Grisaro and Dancygier (2014) 800 Grisaro and Dancygier (2014)

800
Vr (m/s)

Vr (m/s)
600
600
400
400

Unperforated 200 Unperforated

200

0 0
200 300 400 500 600 700 800 900 1000 1100 200 300 400 500 600 700 800 900 1000 1100
V0 (m/s) V0 (m/s)

(c) 300 Cargile et al. (1993) (d) 300 Li et al. (2013)

250 Eq. (12.6) Eq. (12.6)
250
Grisaro and Dancygier (2014a) Grisaro and Dancygier (2014a)
200
200
Vr (m/s)

Vr (m/s)
150
150

100
100
50 Unperforated
50
0
0
0.10 0.15 0.20 0.25 0.30 0.2 0.3 0.4 0.5 0.6 0.7 0.8
H(m) H(m)

(e) 800 100mm-depth RC slab with steel liner (f) 700

150mm-depth RC slab with steel liner
700 100mm-depth RC slab without steel liner 150mm-depth RC slab without steel liner
Eq. (12.6) 600 Eq. (12.6)
Grisaro and Dancygier (2014a) Grisaro and Dancygier (2014)
600
500
Vr (m/s)

Vr (m/s)

500
400
400
300
300
200
200

100 100
350 400 450 500 550 600 650 700 750
200 300 400 500 600 700 800
V0 (m/s) V0 (m/s)

(g) 700 200mm-depth RC slab with steel liner (h) 600 300mm-depth RC slab with steel liner
600 200mm-depth RC slab without steel liner Eq. (12.6)
500
Eq. (12.6) Grisaro and Dancygier (2014a)
500 Grisaro and Dancygier (2014a)
400
Vr (m/s)
Vr (m/s)

400
300
300
200
200

100 100 Unperforated

0 0
400 450 500 550 600 650 700 750 450 500 550 600 650 700 750 800
V0 (m/s)
V0 (m/s)

Fig. 12.30 Comparisons of predicted results with test data of a Hanchak et al. (1992) (48 MPa),
b Hanchak et al. (1992) (140 MPa) c Cargile et al. (1993), d Li et al. (2013) as well as the present
test with target thickness of e 100 mm, f 150 mm, g 200 mm, and h 300 mm
12.3 Hard Projectile Impact on Layered SFRHSC Composite Targets 531

(a) 1000 NDRC+BRL (b) 600 300mm-depth RC slab with steel liner
900
500
800 Li et al.+BRL
Li et al.+Rosenberg et al.
700 400
V0(m/s)

Vr (m/s)
600 Li et al.+BRL
NDRC+Rosenberg et al. 300
500

400 200 NDRC+BRL

Li et al.+Rosenberg et al.
300
100
200 Perforated NDRC+Rosenberg et al.
Unperforated
100 0
0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 450 500 550 600 650 700 750 800
H(m) V0 (m/s)

650
(c) 200mm-depth RC slab with steel liner (d) 700 150mm-depth RC slab with steel liner
600 650
200mm-depth RC slab without steel liner 150mm-depth RC slab without steel liner
550 600
550 Li et al.+Rosenberg et al.
500
Li et al.+Rosenberg et al. 500
450
Vr (m/s)

Vr (m/s)
450
400 Li et al.+BRL Li et al.+BRL
400
350
NDRC+BRL 350 NDRC+BRL
300
300
250 250
200 200
150 NDRC+Rosenberg et al. 150 NDRC+Rosenberg et al.
100 100
350 400 450 500 550 600 650 700 750 300 350 400 450 500 550 600 650 700 750
V0 (m/s) V0 (m/s)

800
(e) 100mm-depth RC slab with steel liner
700 100mm-depth RC slab without steel liner

600
Li et al.+Rosenberg et al.
Vr (m/s)

500
NDRC+BRL
400
Li et al.+BRL
300
NDRC+Rosenberg et al.
200

100
200 300 400 500 600 700 800
V0 (m/s)

Fig. 12.31 Comparisons of test data and predicted curves for monolithic RC/steel composite
targets a V0–H and prediction curves of VBL; Vr–V0 dependence with the depth of b 300 mm,
c 200 mm, d 150 mm, and e 100 mm

12.3.2.2 Targets

The cylindrical SFRHSC targets have the diameter of 1 m and the average density
of 2465 kg/m3. The weight ratios of SFRHSC compositions are listed in
Table 12.7.
Figure 12.33 shows the reinforcing details of the SFRHSC panel. Table 12.8
gives the configurations of each composite target as well as the basic mechanical
properties of the SFRHSC slabs. H is the thickness of the SFRHSC panel, fc is the
unconfined cylinder compressive strength of the SFRHSC, fsplit is the splitting
532 12 Concrete Structures Under Projectile Impact

Fig. 12.32 Unfired projectile (a) (b)

a photograph and 18mm
b dimensions

30mm 20mm
22mm

33mm

130mm
37mm

63mm
17mm
tensile strength of the SFRHSC, fbend is the bending strength of SFRHSC, and Vf is
the steel fiber volumetric ratio. The symbols Y and N denote the certain layer exists
or not in the composite targets. The unit weight, specific gravity, and porosity of the
sandy soil are 15.40 kN/m3, 2.67, and 0.76, respectively.

12.3.2.3 Test Setup

The schematic diagram of the test is depicted in Fig. 12.34. During the test, the
cylindrical targets are placed with the top face perpendicular to the longitudinal axis
of the barrel. The sandy soil is backfilled at the rear of the concrete/steel panels by
the piled sand bags walls.

Table 12.7 Weight ratios of SFRHSC compositions

Cement Silica fume Water Fly ash Sand Aggregate Super-plasticizer Steel fibers
1 0.05 0.33–0.5 0.22–0.24 1.19–1.6 1.6–2.04 0.015 0–0.6

Fig. 12.33 Reinforcing

details of the SFRHSC panels
12.3 Hard Projectile Impact on Layered SFRHSC Composite Targets 533

Table 12.8 Configurations of composite targets

Shots H (mm) fc fsplit fbend Vf Rear steel Rear Reinforcing
No. (MPa) (MPa) (MPa) (%) liner sandy soil bar
11 200 83.7 12 11 3 Y Y Y
12 200 89 11 – 3 Y N Y
13 200 85.5 14 15 3 Y Y Y
14 200 100.8 10 – 3 Y N Y
15 200 89 13 – 3 Y N Y
16 350 63 4 8 0 Y Y Y
17 200 98 12 14 3 Y N Y
18 200 88 13 17 3 N Y Y
19 200 92.7 12 15 3 N Y Y
20 350 46 4 – 0 Y Y Y
21 200 81 8 11 3 N Y Y
22 350 47 4 7 0 Y Y Y
23 250 56.7 8 – 3 Y Y Y
24 350 47 5 8 0 Y Y Y
25 250 53 9 – 3 Y Y Y
26 250 54 9 11 3 Y Y N
27 250 54.3 8 – 3 Y Y N
28 250 49.5 8 – 3 Y Y N
29 250 59.4 8 12 3 Y Y Y
30 250 57.6 9 – 3 Y Y Y

Sandy soil
Steel liner
Concrete panel

Sand
Projectile Powder gun
bags
walls

Fig. 12.34 Schematic diagram of the projectile impact test

12.3.2.4 Test Results

Figure 12.35 illustrates the photographs of several recovered projectiles after the
impact test, and the undeformed or undamaged projectiles are observed. Therefore,
the projectiles can be treated as rigid bodies approximately in this test.
Figures 12.36 and 12.37 show the damages of the concrete/steel targets at the
front and rear faces after each shot, respectively. All the projectiles penetrate
534 12 Concrete Structures Under Projectile Impact

Fig. 12.35 Recovered

projectiles

through the center of the steel mesh. The steel rebars nearby the center are bent
slightly by the extrusion of high-speed projectiles, but no rebars are ruptured. The
rear steel plates are not detached from the concrete slabs at the welding points after
perforation.
Table 12.9 lists the test results, which are classified into six groups according to
the target thickness and configurations. The numbers in the DOP column denote the
penetration depth of the projectile, P denotes that the target is totally perforated by
the projectile, and PL represents the perforation limit.

12.3.3 Analysis and Discussion

12.3.3.1 Influence of Target Configurations

According to Table 12.9, the influences of the steel liner and the soil layer at the
rear on the impact resistance of the composite structure are discussed qualitatively.
(i) Comparing Shot 11 with Shot 14, for the nearly same striking velocities, the
projectile just perforated the 83.7 MPa concrete target in Group 4 and totally
perforated the 100.8 MPa concrete target in Group 5, which indicates that the
sandy soil layer at the rear enhances the impact resistance of the RC/steel
composite target. The same conclusions can be drawn by the comparison
between Shot 13 and Shot 17.
(ii) Comparing Shot 13 with Shot 18, for nearly the same striking velocities and
concrete strength, the DOP of Shots 13 (130 mm) is 17% smaller than that of
Shot 18 (155 mm), which indicates that the steel liner at the rear enhances
the impact resistance of the RC/sandy layer composite target.
(iii) Comparing Shot 12 with Shot 18, for nearly the same concrete strength, the
DOP of projectile with the relatively lower striking velocity in Shot 12 is
12.3 Hard Projectile Impact on Layered SFRHSC Composite Targets 535

Fig. 12.36 Damages of the concrete/steel targets at the front face after Shots a 11, b 12, c 13,
d 14, e 15, f 16, g 17, h 18, i 19, j 20, k 21, l 22, m 23, n 24, o 25, p 26, q 27, r 28, s 29, and t 30

24% larger than that of Shot 18, which indicates that enhancing effect of the
sandy soil layer at the rear on the impact resistance of the bare RC target is
more influential than which of the steel liner at the rear. The same conclu-
sions can be drawn by the comparison between Shot 17 and Shot 21.
(iv) For the effect of reinforcing steel bars, the tested ballistic limits of Group 2
and Group 3 are, respectively, located in 593–661 and 595–669 m/s, by the
comparison between Group 2 and Group 3 with nearly the same compressive
strength. Since the projectile does not hit any embedded steel bars in the test,
536 12 Concrete Structures Under Projectile Impact

Fig. 12.37 Damages of the concrete/steel targets at the rear face after Shots a 11, b 12, c 13, d 14,
e 15, f 16, g 17, h 18, i 19, j 20, k 21, l 22, m 23, n 24, o 25, p 26, q 27, r 28, s 29, and t 30

the influences of reinforcing steel bars on the impact resistances of the

composite target can be neglected, which is consistent with the conclusion
drawn by Hanchak et al. (1992).
(v) Figure 12.36 shows that severe damages of the impact craters occurred in
Shots 16, 20, 22, and 24, as no fibers are incorporated in the concrete panels,
which further validated that mixing steel fibers can efficiently restrain the
cracking and reduce the damages of the concrete target.
12.3 Hard Projectile Impact on Layered SFRHSC Composite Targets 537

Table 12.9 Test results

Group Configurations H (mm) fc Shot V0 DOP
No. (MPa) No. (m/s) (mm)
1 Reinforced HSC slab 350 46 20 583 200
Steel liner at the rear 47 22 705 P
Sandy soil at the rear 47 24 679 265
2 Plain SFRHSC slab 250 54 26 661 P
Steel liner at the rear 54.3 27 580 210
Sandy soil at the rear 49.5 28 593 220
3 Reinforced SFRHSC 250 56.7 23 678 P
slab
Steel liner at the rear 53 25 680 P
Sandy soil at the rear 59.4 29 669 P
57.6 30 595 215
4 Reinforced SFRHSC 200 83.7 11 568 PL
slab
Steel liner at the rear 85.5 13 514 130
Sandy soil at the rear
5 Reinforced SFRHSC 200 89 12 488 192
slab
Steel liner at the rear 100.8 14 576 P
Sandy soil at the rear 98 17 502 170
6 Reinforced SFRHSC 200 88 18 524 155
slab
Steel liner at the rear 92.7 19 528 168
Sandy soil at the rear 81 21 558 166

12.3.3.2 Ballistic Limit

Actually, it is difficult to obtain the critical perforation scenario experimentally. In

this section, the ballistic limit of the layered composite target is analyzed. In
Sect. 12.2, an explicit dimensionless expression in Eqs. (12.4a, b) is proposed to
predict the terminal ballistic parameters of the projectile perforating the bare con-
crete slab. From Eqs. (12.4a, b), we can obtain the ballistic limit VBL by assuming
that Vr = 0.
It should be pointed out that the thickness of the steel liner at the rear is 1 mm in
Sect. 12.2, and it is found that the improvements in the steel liner on enhancing the
impact resistance of the RC/steel composite targets are not obvious. However, the
influence of the 3-mm-thick rear steel liner on the terminal ballistic performance of
the projectiles in this test could not be neglected. Based on the equivalent approach
proposed by Grisaro and Dancygier (2014b), the predictions of ballistic limit of the
RC/steel composite targets can be divided into following four steps.
538 12 Concrete Structures Under Projectile Impact

Step 1: Ballistic limit of the steel liner at the rear

The ballistic limit VBL (unit in m/s) of the steel liner with the thickness of H (unit in
m) can be obtained by BRL perforation model (Alco Products Inc. 1955) given as
follows:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2:88  109  H 1:5  d 1:5
VBL ¼ ð12:5Þ
M0

Step 2: Conversion of the steel liner into an equivalent concrete slab

Under the equal ballistic limit, substituting VBL obtained from Eq. (12.5) into the
concrete perforation equations that are the reverse form of Eqs. (12.4a, b) and
NDRC (1946) perforation model, the equivalent thickness of the concrete panel heq
of the steel liner could be derived.
Step 3: Equivalent thickness of the RC/steel composite targets
The total equivalent thickness of the RC/steel composite targets Heq is the sum of
original thickness of the concrete slab H and the equivalent thickness of the rear
steel liner heq, that is, Heq = H + heq.
Step 4: Ballistic limit of the RC/steel composite targets
The ballistic limit of the RC/steel composite targets with equivalent thickness Heq
can be obtained by Eqs. (12.4a, b) or NDRC (1946) perforation model. Thus, the
following three prediction models can be proposed based on the different combi-
nations of the Eqs. (12.4a, b, 12.5) and NDRC (1946) perforation model:
(i) BRL + Eqs. (12.4a, b) + Eqs. (12.4a, b); (ii) BRL + NDRC + Eqs. (12.4a, b);
and (iii) BRL + NDRC + NDRC.
Figure 12.38 shows the test data of V0–fc, as well as the predicted curves of
ballistic limit VBL for the bare RC slabs by Eqs. (12.4a, b) as well as the RC/steel
composite targets by the above three models, respectively. The solid symbols
denote the RC/steel targets are perforated, hollow symbol with a cross denotes the
perforation limit, and the hollow symbols denote the penetrations.
Since the above procedure is only suitable for the RC/steel composite target, the
predicted curve for VBL should go across between the solid and hollow experi-
mental data in Group 5. In addition, based on the above parametric influential
analyses, for the RC/sandy soil composite target in Group 6 as well as the
RC/steel/sandy soil composite target in Groups 1-4, the corresponding tested bal-
listic limits should be located above the ballistic limit curves for the RC/steel
composite target.
It can be found from Fig. 12.38 that (i) the predictions by the proposed models
“BRL + Eqs. (12.4a, b) + Eqs. (12.4a, b)” and “BRL + NDRC + Eqs. (12.4a, b)”
agree with the test data much more better than that by the model
“BRL + NDRC + NDRC”; (ii) for predicting the ballistic limit of RC/steel com-
posite targets, the model “BRL + NDRC + Eqs. (12.4a, b)” is recommended;
12.3 Hard Projectile Impact on Layered SFRHSC Composite Targets 539

Fig. 12.38 Comparisons (a) 950

between the test data of V0–fc Perforation (Group 1)
and the predicted VBL–fc for 900 Penetration (Group 1)
Eq. (12.4)
a Group 1, b Groups 2-3, and 850 BRL+Eq. (12.4)+Eq. (12.4)
c Groups 4-6 BRL+NDRC+Eq. (12.4)
800

V0 (m/s)
BRL+NDRC+NDRC

750

700

650

600

550
30 35 40 45 50 55 60 65 70
fc (MPa)
(b) 850
Perforation (Group 2)
800
Penetration (Group 2)
Perforation (Group 3)
750
Penetration (Group 3)
Eq. (12.4)
700
V0 (m/s)

BRL+Eq. (12.4)+Eq. (12.4)

BRL+NDRC+Eq. (12.4)
650
BRL+NDRC+NDRC
600

550

500

450
40 45 50 55 60 65 70 75 80
fc (MPa)
(c) 850
Penetration (Group 4)
800 Perforation limit (Group 4)
Penetration (Group 5)
750 Perforation (Group 5)
Penetration (Group 6)
700
Eq. (12.4)
V0 (m/s)

650 BRL+Eq. (12.4)+Eq. (12.4)

BRL+NDRC+Eq. (12.4)
600 BRL+NDRC+NDRC

550

500

450

400
40 50 60 70 80 90 100 110 120
fc (MPa)

(iii) for the RC/steel/sandy soil and RC/sandy soil composite targets, based on the
above analyses in Sect. 12.3.3.1 and Fig. 12.38, we can multiply the results of
model “BRL + Eqs. (12.4a, b) + Eqs. (12.4a, b)” by a coefficient of 1.2 to consider
540 12 Concrete Structures Under Projectile Impact

the strengthening effect of the rear sandy soil layer. It should be stated that the
above conclusions are only suitable for the relatively thick sandy soil layer.

12.4 Numerical Modeling of Rock-Rubble Overlays

Subjected to Projectile Penetration

12.4.1 Introduction

The rock-rubble overlay is composed of randomly distributed rock rubble, and

cement mortar is sometimes poured into it. The rock-rubble overlay has been
proved to be effective in protecting shelters from projectile penetration experi-
mentally (Austim et al. 1982).
Rohani (1987), Gelmanet al. (1987), and Underwood (1995) carried out series of
field tests to study the influence of the configurations of bursting layers on projectile
penetration. It was revealed that the protective potential of rock-rubble overlays
depends significantly on the characteristics of projectiles. The test results implied
that the penetration depth was influenced by the configurations of the bursting
layers. Langheim et al. (1993) presented subscale penetration tests at the
Ernst-Mach-Institut to study the protection function of the grouted rock boulder
layer, considering the armor-piercing (AP) bomb and the hard target penetrator
(HTP). It was found that the results were affected greatly by the impact conditions
of the penetrator. These factors include impact oblique angle and velocity,
mechanical properties of rock rubble and grouted concrete, parameters of projectile,
e.g., length-to-diameter ratio (L/d), sectional pressure (M/A), and diameter (d),
where A is the cross-sectional area of projectile. However, the fundamental
understanding of the defensive mechanism of rock-rubble overlays is still clear as
the limited test data.
As is known to all, field tests are very expensive and time-consuming and cannot
provide insights into the detailed interaction between the projectile and target.
Available analytical approach was used extensively to predict the penetration depth
only for targets with homogeneous material. To the best of our knowledge, there is
not an appropriate empirical formula to predict the projectile penetration into the
rock-rubble overlays as it is a heterogeneous structure. In fact, the available ana-
lytical approaches cannot take into account the influence of the randomly dis-
tributed rock rubble in the target realistically. Therefore, an appropriate numerical
approach may be the best choice to meet this challenge.
Gebara et al. (1993) employed the finite block method to study projectile pen-
etration into rock-rubble overlays. However, this method is based on 2D model, and
how to simulate the random distribution of rock rubble in the overlay was not
12.4 Numerical Modeling of Rock-Rubble Overlays Subjected … 541

presented. In order to get more realistic numerical results, it is necessary to take into
account the real distribution of rock rubble in the target. Actually, the rock bursting
layer is piled up with rock rubble randomly. Hence, the realistic and reliable
numerical modeling for projectile penetration into rock-rubble overlays must con-
sider the randomness of rock-rubble distribution in the target, in order to get
insights into the protective mechanism of rock-rubble overlays. However, the
modeling of the randomness of the shape and size of rock rubble and its spatial
distribution in the overlay is a tough problem.
In this section, we develop a 3D finite element model for the rock-rubble
overlays with grouted concrete, taking into account of the randomness of the shape
and size of rock rubble and its spatial distribution in the overlay. In order to model
the real distribution of rock rubble in a bursting layer, a generation algorithm for
rock rubble with random size and shape is firstly proposed. The taking and drop-
ping as well as compacting algorithms are then developed. Finally, the mapping
algorithm for the generation of finite element grid is proposed, which could be
easily used by commercial finite element software such as LS-DYNA to perform
numerical simulations of rock-rubble overlays subjected to projectile penetration.
The numerical results are compared with available test data, and the validity of the
proposed numerical approach is verified. Besides, parametric studies are also pre-
sented to analyze the effects of the size, strength, and volume percentage of rock
rubble and the strength of grouted concrete on the penetration depth.
Figure 12.39 shows the flowchart of the numerical modeling.

Fig. 12.39 Flowchart of the Quadrilateral

numerical modeling
3D rock-rubble Octahedron

Rock-rubble with
random size and shape
Dropping
Rock-rubble overlay
Compacting
Mapping
Finite element Model
algorithm
Hydrocodes:
Analysis approach
LS-DYNA
Comparison with the
Validation
test data
Effects of the impact
conditions and target
Analysis and discussion
configurations on
penetration
542 12 Concrete Structures Under Projectile Impact

12.4.2 3D Finite Element Model

Constructing a 3D finite element model of rock-rubble overlays with the consid-

eration of the randomness of the shape and size of rock rubble and its spatial
distribution consists of the following three parts: (i) proposing an algorithm for the
generation of all the necessary 3D rock rubble with randomness in size and shape;
(ii) proposing an algorithm for the generation of random distribution of rock rubble
in a bursting layer by the so-called take-and-place method as well as the compacting
method to meet the requirement of volume percentage of rock rubble in the overlay;
and (iii) proposing a mapping algorithm to generate the finite element grid. The
detail of the above three steps are given as follows.

12.4.2.1 Generation of 3D Rock Rubble with Random Size and Shape

Firstly, we should adopt an algorithm for the generation of random number. In this
section, we employ the mixed congruent algorithm to generate the pseudorandom
number list. The general form of the mixed congruent algorithm can be written as
follows (Greenberger 1961): If Xi is assumed the initial value, the recursive function
is Xn+1 = N  Xn + C (mod M), in which N and C are constants, and M is modulus
operator.
The proposed generation algorithm for 3D rock rubble with random size and
shape can be divided into two steps: The first step is to generate an octahedron, and
the second step is to generate a polyhedron. They all have random sizes and shapes.
The first step consists of the following two substeps:
Substep 1: Generate a quadrilateral in a circle.
According to the size of rock rubble that can be approximately represented by a
diameter of Drk, generate a circle with the radius of Drk/2 within the XOY coordinate
plane. Randomly generate four points A, B, C, D on the circle counterclockwise,
and form a quadrilateral connecting the four points. L1, L2, L3, L4 and K1, K2, K3, K4
denote the four sides and angles of the quadrilateral, respectively. The condition of
Li > Drk/4 should be satisfied to avoid sharp corners in the quadrilateral.
Substep 2: Generate a convex octahedral.
Firstly, randomly generate two vertices Z+ and Z− along the z-direction on the
surface of the sphere with a diameter of Drk, and then, connect the four points to the
two vertices to generate a convex octahedral. The intersection point of the line Z+Z−
with the plane ABCD is Di.
This step is shown in Fig. 12.40a.
The second step consists of the following three substeps:
Substep 1: Find the longest line CiCj in the octahedral. Generate a point Ck in the
line CiCj. |CkCj| and |CkCi| are the length from point Ck to point Cj
12.4 Numerical Modeling of Rock-Rubble Overlays Subjected … 543

and point Ci, respectively. Generate a point from Ck along the direction
! ! ! !
of V ij ; where V ij ¼ V i þ V j . Vi and Vj are the unit normal vector of the
plane i and plane j attaching to the line of CiCj. This substep is shown in
Fig. 12.40b.
Substep 2: Generate a new polyhedron by connecting points CnCi, CnCj, CnCi1,
and CnCj1 in turn. Judge the convexity of the polyhedron using the
!
vector method. V r is the unit normal vector of the face CaCnCj (a = i1
or j1, as shown in Fig. 12.40b), given as follows:

2 3
xr
!
V r ¼ 4 yr 5 ð12:6Þ
zr

Vai is the unit vector from point Ca to2Ci, given3 as follows:

x  xa
! 1 4 i
Vai ¼ yi  ya 5 ð12:7Þ
Lia
zi  za

where Lia is the length of the line CiCa.

! !
If the generated polyhedron is convex, it has Vr  Vai \0. Otherwise, it has
! !
Vr  Vai [ 0.
Substep 3: Repeat the substeps 1 and 2 until the total faces of the polygon meet the
requirements.
The generated rock rubble is shown in Fig. 12.40c, in which the face number of
the rock rubble (N) is 40 and 60, respectively.

12.4.2.2 Generation of Random Distribution of Rock Rubble

in a Bursting Layer

This section presents the generation of random distribution of rock rubble in a

bursting layer, which is composed of two steps: taking and dropping, and com-
pacting. The taking and dropping is conducted by picking up and dropping all
particles of the previously generated rock rubble one by one into the bursting layer
according to the previously generated a random taking and dropping list of all
particles. The compacting is to make all particles of the dropped rock-rubble contact
with each other by gravity and/or forced vibration at the bottom of the bursting
layer. At the same time, the volume ratio of rock rubble in the bursting layer is
computed constantly. When the volume ratio reaches the requirement, the com-
pacting is stopped and the locations of all particles of the rock rubble in the bursting
layer are recorded.
544 12 Concrete Structures Under Projectile Impact

(a)

(b)

(c)

N=40 N=60

Fig. 12.40 Generation of 3D rock-rubble a octahedral, b convex polyhedron, and c 3D

rock-rubble

The taking and dropping step consists of two substeps and described in detail as
follows:
Substep 1: Generate a randomly taking and dropping order list of all particles of
the rock rubble generated in Sect. 12.4.2.1.
Substep 2: Put the all particles of the rock rubble one by one according to the
taking and dropping order list into the bursting layer. Check the
boundary conditions to ensure that the rock rubble is located inside the
bursting layer. If the boundary condition is not satisfied, replace this
particle again until the particle locates inside the bursting layer. Check
the overlapping with the previously placed particles. If not satisfied,
replace this particle again until meeting the requirements. Check the
volume percentage of the particle whether meets the requirement. Stop
taking and dropping if satisfied, and record the positions of all the
particles in the bursting layer.
12.4 Numerical Modeling of Rock-Rubble Overlays Subjected … 545

Figure 12.41a shows the result of the taking and dropping, in which all particles
of the rock rubble are randomly distributed in the bursting layer.
The particles of rock rubble can be compacted by the gravity and/or mechanical
vibration at the bottom of the overlay. The compacting step is listed in detail as
follows:
Substep 1: Firstly, a finite element model can be established since the final posi-
tions of all the particles in the bursting layer have been recorded. Then,
all the particles in the overlay can be compacted by their gravity,
assuming that the particles are rigid bodies. The effects of contact and
slip between the particles should be considered to avoid overlapping
during the compacting process. The above numerical analysis can be
easily completed by finite element programs such as LS-DYNA.
Substep 2: The particles in the overlay can be further compacted by mechanical
vibration such as placing an artificial sinusoidal forced vibration at the
bottom of the overlay, if the volume percentage of the particle does not
meet the requirement. Again, the effects of contact and slip between the
particles should be considered to avoid overlapping during the com-
pacting process, and the volume percentage of the rock rubble is
computed constantly. Stop compacting if the volume percentage meets
the requirement. Finally, the positions of all particles of the rock rubble
are recorded.
Figure 12.41b shows the result of the compacting, in which all particles of the
rock rubble are randomly distributed and contact with each other in the bursting
layer.

12.4.2.3 3D Finite Element Analysis Model

In this section, a mapping algorithm is presented to generate the finite element grid.
The rock-rubble overlay is usually composed of rock-rubble particles and grouted
concrete. Therefore, the key step of the mapping algorithm is to identify all the
elements’ material property in the rock-rubble overlay. The detailed steps are as
follows.
Step 1: Input the position data of all the particles of the rock rubble in the
overlay according to the final result of the compacting process.
Step 2: Generate a homogenous grid in the bursting layer, and the grid should be
fine enough in order to mesh the smallest space between all particles of
the rock rubble; normally, the largest dimension of the hexahedral grid
should be no more than one-fourth of the smallest space.
Step 3: Identify the material property of each 8-node element in the grid
according to its location by comparing the final positions of all particles
of the rock rubble in the bursting layer. The material property of this
546 12 Concrete Structures Under Projectile Impact

element is assumed to be rock rubble when more than four nodes of the
element are located in the rock rubble. Otherwise, the material property
of this element is assumed to be concrete. Repeat the above process until
the material property of all elements in the bursting layer is identified.
Finally, the finite element model for the rock-rubble overlay with
grouted concrete is established, as shown in Fig. 12.42.

Fig. 12.41 Results of a taking and dropping and b compacting

Fig. 12.42 Rock-rubble overlay with grouted concrete a random distribution of rock rubble and
b finite element grid
12.4 Numerical Modeling of Rock-Rubble Overlays Subjected … 547

12.4.3 Numerical Analysis Approach

The hydrocode such as LS-DYNA and AUTODYN has been used extensively in
modeling intense dynamic events, such as blast, impact, and penetration. In a
hydrocode, material models that properly describe material behaviors under high
strain rate and high pressure are necessary. The material model is usually composed
of two parts. The first part is the strength criterion that controls the yield strength
according to stress invariants. The second part is the equation of state (EOS) that
determines the hydropressure in terms of density and energy. In this section, the
hydrocode LS-DYNA is employed to simulate the projectile penetration into
rock-rubble overlays, based on the proposed finite element model in Sect. 12.4.2.
Firstly, the previously proposed 3D finite element model of the rock-rubble
overlay is mapped into LS-DYNA using FORTRAN and APDL (ANSYS
Parametric Design Language). Then, the material models such as the Johnson–
Cook (JC) model, Holmquist–Johnson–Cook (HJC) model, and the equation of
state (EOS) are used to simulate the different mechanical properties of penetrators
and targets. The details of the above models can be referred to Chap. 7 and the
related references. The contact and friction effects among the particles of rock
rubble, grouted concrete, and projectile are taken into account the simulation using
the Contact Eroding Surface To Surface in LS-DYNA. And the friction coefficient
is defined as follows (Hallquist 2007):

lc ¼ lk þ ðls  lk ÞeDCjvrel j ð12:8Þ

where ls is the static friction coefficient, lk is the dynamic friction coefficient, vrel is
the relative velocity of the surfaces in contact, and DC is an exponential decay
coefficient.
In the modeling of large deformation, the erosion technique is often used to
remove elements with excessive distortions to avoid computational overflow. Here,
the erosion technique is employed to model the failure of rock rubble or concrete. It
removes elements automatically when stress or strain in an element satisfies the
predefined erosion criteria. There are two frequently used erosion criteria. The first
is for concrete material depending on the maximum principal strain of 0.15 and
shear strain of 0.8. The second is for rock-rubble depending on the maximum
principal strain of 0.24. And these two erosion criteria have been successfully used
by other researchers to simulate concrete and rock subjected to explosion, pene-
tration, and perforation (Xu and Lu 2006; Unosson 2000; Shi et al. 2010).
The computational parameters are shown in Tables 12.10, 12.11, and 12.12
according to Hallquist (2007) and Holmquist et al. (1993) for the penetrator,
concrete, and rock rubble. The Gruneisen equation of state (EOS) for the penetrator
is employed. The details of the Gruneisen EOS can be referred to Chap. 5.
Table 12.13 shows the parameters of Gruneisen EOS for the penetrator.
 548

Table 12.10 Parameters of the JC material model for the penetrators

Projectile q ðkg=m3 Þ G (GPa) A (MPa) B n C m Tm (K) Tr (K) Cp (J/kg K) e_ 0
AP bomb 7830 98 7.92
10−3 0.0051 0.26 0.014 1.03 1793 293 4.77
10−6 1.0
10−6
12

HTP 17,700 160 1197 0.014 0.05 0.025 1.9 1723 293 15
10−6 1.0
10−6
SDB 3862.5 77 7.92
10−3 0.0051 0.26 0.014 1.03 1793 293 4.77
10−6 1.0
10−6
Concrete Structures Under Projectile Impact
12.4 Numerical Modeling of Rock-Rubble Overlays Subjected … 549

Table 12.11 Parameters of the HJC material model for concrete (unit in cm g ls K)
Concrete C30 C50 C80 Concrete C30 C50 C80
q 2.44 2.44 2.44 Smax 7.0 7.0 7.0
G 0.123 0.141 0.167 Pc 0.8
10−4 1.37
10−4 2.333
10−4
A0 0.79 0.79 0.79 PL 1.05
10−2 1.05
10−2 1.05
10−2
B0 1.60 1.60 1.60 D1 0.04 0.04 0.04
C0 0.007 0.007 0.007 fc 2.4
10−4 4.1
10−4 7.0
10−4
N 0.61 0.61 0.61 T 2.7
10−5 3.6
10−5 4.65
10−5
Ul 0.1 0.1 0.1 e_ 0 1
10−6 1
10−6 1
10−6

12.4.4 Validation

In this section, we conduct numerical simulations of the test by Langheim et al.

(1993) to verify the reliability of the proposed numerical modeling approach.
The geometry of the armor-piercing (AP) bomb and the grouted concrete rock
boulder layers used in the analysis are the same as the test, as shown in Fig. 12.43.
The rock boulder layer is 150 cm in width and length, and 30 and 60 cm in thickness
for different impact velocities of the AP bomb. The bottom of the rock boulder layer
is the concrete layer with the dimension of 150 cm
150 cm
30 cm. The con-
figuration of the AP bomb is 38.5 cm in length, 6.9 cm in diameter, and 1.5 in CRH.
The static and dynamic friction coefficients between the projectile and target are
taken as 0.05, while the static and dynamic friction coefficients among the target
material are set to be 0.572 according to the researches like Chen (1989), Stronge
(1994, 2000, 2013), and Jones and William (2000, 2008). The target’s bottom is
fully constrained according to the test. The 3D 8-noded solid element is utilized to
discretize the projectile and target in the numerical simulation.
The JC material model for the projectile is shown in Table 12.10 according to
Langheim et al. (1993). The HJC material model’s parameters for the grouted
concrete of C50 and rock rubble of 250 MPa are shown in Tables 12.11 and 12.12,
respectively, according to LS-DYNA (Hallquist 2007) and Holmquist et al. (1993).
Table 12.14 gives the key parameters of the computational cases, which are the
same as the test (Langheim et al. 1993), in which H and B are the thickness of
rock-rubble layer and the concrete layer at the bottom of the target, respectively,
and a is the oblique angle of the projectile.
Table 12.15 gives the comparison of the penetration depth hpen between the
numerical results and test data for each case and good agreements are observed.
 550

Table 12.12 Parameters of the HJC material model for rock rubble (unit in cm g ls K)
Rock rubble 80 MPa 100 MPa 120 MPa 250 MPa Rock rubble 80 MPa 100 MPa 120 MPa 250 MPa
q 2.66 2.66 2.66 2.66 Smax 7.0 7.0 7.0 7.0
G 0.167 0.185 0.201 0.406 Pc 2.33
10−4 3.0
10−4 3.67
10−4 4.21
10−4
A0 0.79 0.79 0.79 0.79 PL 1.2
10−2 1.2
10−2 1.2
10−2 1.2
10−2
12

B0 1.6 1.6 1.6 1.6 D1 0.04 0.04 0.04 0.04

C0 0.007 0.007 0.007 0.007 fc 7.0
10−4 9.0
10−4 1.1
10−3 2.4
10−3
N 0.61 0.61 0.61 0.61 T 4.65
10−5 5.3
10−5 5.47
10−5 5.91
10−5
UL 0.012 0.012 0.012 0.012 e_ 0 1.0
10−6 1.0
10−6 1.0
10−6 1.0
10−6
Concrete Structures Under Projectile Impact
12.4 Numerical Modeling of Rock-Rubble Overlays Subjected … 551

Table 12.13 Parameters of Gruneisen EOS for the penetrator

Parameters Value Parameters Value
Intercept C (m/s) 4029 Coefficient s3 0
Coefficient s1 1.237 Gruneisen parameter c0 2.17
Coefficient s2 0 Constant a 0.46

Fig. 12.43 Configurations of a projectile, b rock boulder layer, and c concrete bottom layer

Table 12.14 Key parameters of the penetrator and target

Case V0 (m/s) a (°) H (cm) B (cm) Projectile Rock volume Rock Size (cm)
fraction Vrk (%)
1-1 398 2 30 15 AP bomb 44.3 18
1-2 504 5 60 15 AP bomb 44.3 18
1-3 405 6 30 15 AP bomb 44.3 18

Table 12.15 Comparison of Case 1-1 1-2 1-3

hpen (cm)
Test 20 40 18
Simulation 21.8 41.5 19.1

12.4.5 Parametric Analysis and Discussion

The above validated analysis approach is employed to analyze the Small Diameter
Bomb (SDB) penetration into rock-rubble overlays, focusing on the effects of
impact position at the target and oblique angle of the projectile, grouted concrete
strength as well as size, strength, and volume percentage of rock rubble on the
penetration depth.
The projectile (SDB), as shown in Fig. 12.44, has a CRH of 3, weight of 113.4 kg,
diameter of 15.24 cm, and length of 182.88 cm. The density is 3862.5 kg/m3 based
on the equivalent to the total mass. The JC material model is used for SDB, and
corresponding computational parameters are shown in Table 12.10.
The finite element model of the SDB has 8000 solid elements, as shown in
Fig. 12.45. The target is divided into two layers: The upper is the rock-rubble layer
with the dimension of 150 cm
150 cm
250 cm, and the bottom is the concrete
552 12 Concrete Structures Under Projectile Impact

Fig. 12.44 Geometry and dimension of a projectile (SDB) and b target

layer with the dimension of 150 cm
150 cm
30 cm, as shown in Fig. 12.43c.
The transmitting boundaries are set at the bottom and four sides of the target to
absorb expansion and shear waves. The single-point integration and hourglass
control are used to predict the nonlinear responses of large deformation and material
failure effectively. In order to increase the computational efficiency and accuracy
simultaneously, the grid size in the center of the target within 2.5 times the pro-
jectile diameter is taken as 10 mm, while the other is set to be 20 mm. The total
number of solid elements in the target is 1,089,000.
The total eighteen computational cases, shown in Table 12.16, are designed to
analyze the effects of the impact position at the target and oblique angle of the
projectile, grouted concrete strength as well as size, strength, and volume per-
centage of rock rubble on the penetration depth.

Fig. 12.45 Finite element

model of a projectile
(SDB) and b target
12.4 Numerical Modeling of Rock-Rubble Overlays Subjected … 553

Table 12.16 Key parameters of 18 computational cases (V0 = 300 m/s)

Case Impact Target configurations a (°)
No. position Drk/ fc (MPa)/ fc Vrk
d rock (MPa)/concrete (%)
1 Center 1.5 100 C30 83 0
2 Left 0.5d 1.5 100 C30 83 0
3 Right 0.5d 1.5 100 C30 83 0
4 Up 0.5d 1.5 100 C30 83 0
5 Down 0.5d 1.5 100 C30 83 0
6 Center 1.5 100 C30 83 5
7 Center 1.5 100 C30 83 10
8 Center 1.5 100 C30 83 20
9 Center 1.0 100 C30 83 0
10 Center 2.0 100 C30 83 0
11 Center 3.0 100 C30 83 0
12 Center 1.5 80 C30 83 0
13 Center 1.5 120 C30 83 0
14 Center 1.5 100 C30 29.1 0
15 Center 1.5 100 C30 49.5 0
16 Center 1.5 100 – 83 0
17 Center 1.5 100 C50 83 0
18 Center 1.5 100 C80 83 0

12.4.5.1 Influence of Impact Position

In order to analyze the effect of different impact positions on the penetration, we

present five different impact positions as shown in Fig. 12.46, which corresponds to
Cases 1–5 in Table 12.16. The simulating results are given in Table 12.17. The
maximum deviation in the predicted DOPs is less than 5.7% when compared with
the striking at the center. Figure 12.47 shows that the numerical predictions of the
deformation and trajectory of the projectile. It indicates that the variations in the
impact positions have few effects on the penetration depth.

12.4.5.2 Influence of Impact Obliquity

Three different impact obliquity angles, i.e., 5°, 10°, and 20°, are considered to
analyze the effect of impact obliquity on the penetration, which corresponds to
Cases 6–8 in Table 12.16. The numerical results are shown in Fig. 12.48. It is
indicated that the impact oblique angle has great effect on the DOP as well as on the
deformation and trajectory of the projectile.
554 12 Concrete Structures Under Projectile Impact

Fig. 12.46 Impact positions

of the projectile

Table 12.17 Effect of impact Case No. 1 2 3 4 5

positions on penetration
hpen (m) 1.93 1.85 2.04 1.96 1.83

Fig. 12.47 Deformation and trajectory of the projectile at the different striking positions

12.4.5.3 Influence of Size of Rock Rubble

Cases 9–11 in Table 12.16 simulate the projectile penetration into the targets with
different sizes of the rock rubble. The numerical results of the penetration depth
hpen are shown in Fig. 12.49. It is shown that hpen decreases with the increase in
Drk/d. However, the tendency of the above decrease reduces greatly when
Drk/d  2. The size of rock rubble has significant effect on DOP. The target with
12.4 Numerical Modeling of Rock-Rubble Overlays Subjected … 555

2.00
(a) (b)
1.75

1.50
hpen (m)

1.25

1.00

0.75

0.50
0 5 10 15 20
α (°)

Fig. 12.48 Effect of impact obliquity on a hpen and b deformation and trajectory of the projectile

Fig. 12.49 Effect of Drk/d on 2.5

DOP

2.0
hpen (m)

1.5

1.0
0.5 1.0 1.5 2.0 2.5 3.0 3.5
Drk/d

bigger rock rubble increases the chance for projectile to strike the rock rubble,
which may consume more kinetic energy of the projectile.

12.4.5.4 Influence of Strength of Rock Rubble

Cases 12–13 investigate the effect of rock-rubble strength on the penetration. The
numerical results of the penetration depth hpen are shown in Fig. 12.50. It is shown
that hpen decreases greatly and linearly with the increase in the strength of the rock
rubble.
556 12 Concrete Structures Under Projectile Impact

Fig. 12.50 Effect of 3.0

rock-rubble strength on DOP

2.5

hpen (m)
2.0

1.5

1.0
70 80 90 100 110 120 130
Rock-rubble strength (MPa)

12.4.5.5 Influence of Volume Percentage of Rock Rubble

Cases 14–15 investigate the effect of volume percentage of rock rubble on the
penetration. The volume percentage of rock rubble varies from 29.1 to 83%. The
numerical results of the penetration depth hpen are shown in Fig. 12.51. It is shown
that hpen decreases greatly and almost linearly with the increase in Vrk.

12.4.5.6 Influence of Grouted Concrete Strength

Cases 16–18 investigate the effect of grouted concrete strength on the penetration,
and the numerical results are shown in Fig. 12.52. It is indicated that DOP
decreases with the increase in the grouted concrete strength, especially when the
grouted concrete strength is larger than 50 MPa.

Fig. 12.51 Effect of Vrk on 2.3

DOP
2.2
hpen (m)

2.1

2.0

1.9

1.8
0 25 50 75 100
Vrk (%)
12.5 Summary 557

Fig. 12.52 Effect of grouted 2.5

concrete strength on DOP

2.0

hpen (m)
1.5

1.0
-20 0 20 40 60 80 100
Concrete strength (MPa)

12.5 Summary

In this chapter, the impact resistance of four types of concrete structures against
high-speed projectiles is investigated in detail and can be summarized as follows:
(1) In Sect. 12.1, the 7.62-mm API bullet impacting test on bare and rear fabric
(CFRP or UHMWPE)-strengthened UHP-BASFRC panels is conducted. It is
derived that (i) the perforation limit of the bare UHP-BASFRC panel is
experimentally determined as 77.2 mm; (ii) predictions by Chen’s model agree
well with the bullet impact test data of UHP-SFRC panel, and an reduction
coefficient of 0.777 is suggested for UHP-BASFRC targets; and (iii) the above
two kinds of fabrics strengthened at the rear face could reduce the perforation
limit and considerably restrain the crater dimensions as well as catch the ejected
fragments, and the strengthened effect by CFRP is better than that by
UHMWPE.
(2) In Sect. 12.2, twenty-five shots of reduce-scaled projectiles perforation test on
five configurations of monolithic and segmented RC panels with a rear steel
liner is conducted. The impact resistances of layered bare RC and RC/steel
composite targets are analyzed. It is obtained as follows: (i) With the same total
thickness and projectile striking velocity, the stacked segmented targets have
higher impact resistance than the spaced segmented targets; (ii) with the same
total thickness, the impact resistance of the monolithic concrete slab is better
than that of the segmented target; the more the layers of the segmented target,
the more the inferiority of the segmented target has; when the number of the
layers is fixed, the impact resistance of the segmented target is dependent on the
order of the slabs, which is enhanced by putting the thicker RC slab at the rear;
(iii) the relationship of Ir–I0 is almost linear for both bare RC and RC/steel
composite targets, which does not depend on the thickness of the panel; thus,
the ballistic limit can be determined by fitting the test data linearly. Further, an
558 12 Concrete Structures Under Projectile Impact

explicit dimensionless expression for predicting the terminal ballistic parame-

ters of the projectile perforating the RC panel is proposed and well validated;
and (iv) the approach proposed by Grisaro and Dancygier (2014b) for pre-
dicting the ballistic limit of projectile perforating the RC/steel composite target
is evaluated, and the two combined predictions of “NDRC + BRL” and
“NDRC + Rosenberg” coincide better with the test data.
(3) In Sect. 12.3, twenty shots of hard projectile high-speed impact test on the
“SFRHSC panel/steel liner/sandy soil”-layered composite target are carried out.
The impact resistance of the above composite targets as well as the parametric
influences is assessed and discussed. It indicates that within the discussed
parameters’ variation ranges, putting both the sandy soil layer and the steel liner
at the rear of RC targets can enhance the impact resistances of targets, and the
enhanced effect of the sandy soil layer on the impact resistance of the bare RC
target is more influential. Reinforcing steel bars of the concrete panel has no
obvious effect on the impact resistances of the composite target. Finally, based
on the perforation model of RC/steel targets, a practical approach for predicting
the ballistic limits of RC/steel/sandy soil composite targets is proposed by
considering the enhancing effect of the sandy soil layer on the impact
resistance.
(4) In Sect. 12.4, a 3D finite element analysis approach is presented to study the
projectile penetration into rock-rubble overlays with the consideration of the
randomness of rock rubble in shape and spatial distribution. The generation
algorithm of 3D rock rubble with random size and shape is firstly proposed.
The taking and dropping and compacting algorithms for the random distribu-
tion of all particles of rock rubble in an overlay are then developed. Thirdly, the
mapping algorithm is presented to generate a finite element grid, which can be
easily implemented into the finite element software, such as LS-DYNA, to
perform the numerical simulation of projectile penetration. The numerical
results are compared with the related test data, and a good agreement is
obtained. Finally, the above validated analysis approach is employed to analyze
the Small Diameter Bomb (SDB) penetration into rock-rubble overlays,
focusing on the effects of impact position at the target and oblique angles of the
projectile, grouted concrete strength as well as size, strength, and volume
percentage of rock rubble on the penetration depth. It is indicated that the size,
strength, and volume percentage of rock rubble as well as the grouted concrete
strength are critical to the projectile penetration.
References

ABAQUS: Analysis user’s manual (2004) Version 6.5

Abbas H, Paul DK, Codbole PN, Nayak GC (1996) Aircraft crash upon outer containment of
nuclear power plant. Nucl Eng Des 160:13–50
Abdel-Kader M, Fouda A (2014) Effects of reinforcement on the response of concrete panels to
impact of hard projectile. Int J Impact Eng 63:1–17
Abrams DA (1917) Effect of rate of application of load on the compressive strength of concrete.
ASTM J 17(2):364–367
ACE (1946) Fundamentals of protective structures. Report AT120 AT1207821, Army Corps of
Engineers, Office of the Chief of Engineers
Adeli H, Amin AM (1985) Local effects of impactors on concrete structures. Nucl Eng Des
88:301–317
Adley MD, Moxley RE (1996) PENCURV/ABAQUS: a simply coupled penetration trajectory/
structural dynamics model for deformable projectiles impacting complex curvilinear targets.
Technical report SL-96-6. U.S. Army Engineer Waterways Experiment Station, Vicksburg,
MS39180
Alco Products Inc (1955) Hazards summery report for the Army Packaged Power Reactor,
APAE-2
Alekseevskii VP (1966) Penetration of a rod into a target at high velocity. Combust Explos Shock
2:63–66
Alia A, Souli M (2006) High explosive simulation using multi-material formulations. Appl Therm
Eng 26(10):1032–1042
Almansa EM, Cánovas MF (1999) Behaviour of normal and steel fiber-reinforced concrete under
impact of small projectiles. Cem Concr Res 29:1807–1814
Almusallam TH, Siddiqui NA, Iqbal RA, Abbas H (2013) Response of hybrid-fiber reinforced
concrete slabs to hard projectile impact. Int J Impact Eng 58:17–30
Almusallam T, Al-Salloum Y, Alsayed S, Iqbal R, Abbas H (2015) Effect of CFRP strengthening
on the response of RC slabs to hard projectile impact. Nucl Eng Des 286:211–226
Altair HyperWorks (2010) HyperMesh, Version 11.0
Amde AM, Mirmiran A, Walter TA (1997) Local damage assessment of turbine missile impact on
composite and multiple barriers. Nucl Eng Des 178(1/2):145–156
Amirikian A (1950) Design of protective structures. Report NT-3726, Bureau of Yards and Docks,
Department of the Navy
Anderson CE Jr, Bodner SR (1988) Ballistic impact: the status of analytical and numerical
modeling. Int J Impact Eng 7(1):9–35
Anderson CE Jr, Walker JD, Hauver GE (1992a) Target resistance for long-rod penetration into
semi-infinite targets. Nuclear Eng Des 138(1):93–104
Anderson CE Jr, Morris BL, Littlefield DL (1992b) A penetration mechanics database. SwRI
report 3593/001, AD-A246351, Southwest Research Institute, San Antonio, TX

© Science Press, Beijing and Springer Nature Singapore Pte Ltd. 2017 559
Q. Fang and H. Wu, Concrete Structures Under Projectile Impact,
DOI 10.1007/978-981-10-3620-0
560 References

Ansari F, Li QB (1998) High-strength concrete subjected to triaxial compression. ACI Mater J 95

(6):747–755
ANSYS/Autodyn-2D and 3D, Version 6.1 (2007) User documentation, ANSYS Inc, Canonsburg,
PA
Aoude H, Dagenais FP, Burrell RP, Saatcioglu M (2015) Behavior of ultra-high performance fiber
reinforced concrete columns under blast loading. Int J Impact Eng 80:185–202
Arros J, Doumbalski N (2007) Analysis of aircraft impact to concrete structures. Nucl Eng Des
237(12):1241–1249
Attard MM, Setunge S (1996) Stress-strain relationship of confined and unconfined concrete. ACI
Mater J 93(5):432–441
Attaway SW, Brown KN, Mello FJ, Heinstein MW, Swegle JW, Ratner JA, Zadoks RI (1998)
PRONTO3D Users instructions: a transient dynamic code for nonlinear structural analysis.
SAND98-1361, Sandia National Laboratories, Albuquerque, NM
Austim CF, Halsey CC, Clodt RL (1982) Protection systems development. Technical report,
Tyndall Air Force Base
AUTODYN Theory Manual (Revision 4.3) (2003) Century Dynamics, Inc
Ayisit O (2008) The influence of asymmetries in shaped charge performance. Int J Impact Eng 35
(12):1399–1404
Bahar LY, Rice JS (1978) Simplified derivation of the reaction-time history in aircraft impact on a
nuclear power plant. Nucl Eng Des 49:263–268
Bangash MYH (1989) Concrete and concrete structures: numerical modelling and application.
Elsevier Applied Science, London
Bangash MYH (1993) Impact and explosion: structural analysis and design. CRC Press, Boca
Raton
Barnett SJ, Lataste JF, Parry T, Millard SG, Soutsos MN (2010) Assessment of fibre orientation in
ultra high performance fibre reinforced concrete and its effect on flexural strength. Mater Struct
43:1009–1023
Barr P (1990) Guidelines for the design and assessment of concrete structures subjected to impact.
Report, UK Atomic Energy Authority, Safety and Reliability Directorate, HMSO, London
Bazant ZP (1992) Fracture mechanics of concrete structures. In: Breckenridge CO (ed)
State-of-the-art and Proceedings of 1st international conference (FraMCos 1). Elsevier
Applied Science, Amsterdam
Ben-Dor G, Dubinsky A, Elperin T (2006) Applied high-speed plate penetration dynamics.
Springer, Berlin
Ben-Dor G, Dubinsky A, Elperin T (2009) Ballistic properties of multilayered concrete shields.
Nucl Eng Des 239(10):1789–1794
Ben-Dor G, Dubinsky A, Elperin T (2010) Estimation of perforation thickness for concrete shield
against high-speed impact. Nucl Eng Des 240(5):1022–1027
Ben-Dor G, Dubinsky A, Elperin T (2013) High-speed penetration dynamics. Engineering models
and methods. Published by World Scientific Publishing Co. Pte. Ltd., Singapore
Beppu M, Miwa K, Itoh M, Katayama M, Ohno T (2008) Damage evaluation of concrete plates by
high-velocity impact. Int J Impact Eng 35(12):1419–1426
Bernard RS, Creighton DC (1978) Non-normal impact and penetration: analysis for hard targets
and small angles of attack. Technical report SL-78-14, U.S. army engineer waterways
experimental station soils and pavements laboratory, USA
Bernard RS, Creighton DC (1979) Projectile penetration in soil and rock: analysis for non-normal
impact. Technical report SL-79-15, army engineer waterways experiment station structures
laboratory, USA, pp 13–26
Berriaud C, Sokolovsky A, Gueraud R, Dulac J, Labrot R (1978) Local behaviour of reinforced
concrete walls under missile impact. Nucl Eng Des 45:457–469
Beth RA (1941) Penetration of projectiles in concrete. PPAB interim report no. 3
Béton CEI (1993) CEB-FIP model code 1990: design code. Thomas Telford, UK
References 561

Bischoff PH, Perry SH (1991) Compressive behavior of concrete at high strain rates. Maters Struct
24(6):425–450
Bishop RF, Hill R, Mott NF (1945) The theory of indentation and hardness tests. Proc Phys Soc 57
(Part 3):147–159
Bludau C, Keuser M, Kustermann A (2006) Perforation resistance of high-strength concrete
panels. ACI Struct J 103(2):188–195
BNFL (2003) Reinforced concrete slab local damage assessment. R3 impact assessment
procedure, Appendix H, vol 3. Magnox Electric plc & Nuclear Electric Limited
Booker PM, Cargile JD, Kistler BL, Saponara VL (2009) Investigation on the response of
segmented concrete targets to projectile impact. Int J Impact Eng 36(7):926–939
BPC (1974) Design of structures for missile impact. Topical report BC-TOP-9-A, Bechtel Power
Corporation, USA
Brandt AM (2008) Fibre reinforced cement-based (FRC) composites after over 40 years of
development in building and civil engineering. Compos Struct 86:3–9
Byers RK, Yarrington P, Chabai AJ (1978) Dynamic penetration of soil media by slender
projectiles. Int J Eng Sci 16(11):835–844
Candappa DC, Sanjayan JG, Setunge S (2001) Complete triaxial stress-strain curves of
high-strength concrete. J Mater Civil Eng 13(13):209–215
Cargile JD, Giltrud ME, Luk VK (1993) Perforation of thin unreinforced concrete slabs. Sandia
National Laboratory, SAND-93-9150C
Cauberg N, Piérard J, Remy O (2008) Ultra high performance concrete: mix design and practical
applications. Tailor Made Concr Struct New Solutions our Soc: 1085–1087
CEB-FIB Model Code 1990 (1993) Design code. Thomas Telford, Lausanne, Switzerland
Chang WS (1981) Impact of solid missiles on concrete barriers. J Struct Div ASCE 107(ST2):
257–271
Chang JZ, Lu YG, Sun CJ (2009) Experimental study for pre-hole efficiency on masonry wall
target by shaped charge. Ordnance Mat Sci Eng 32(6):39–41 (in Chinese)
Chanteret PY, Reck B (2013) Shaped charge jet penetration into masonry materials. Proceedings
of 27th International on Symposium Ballistics, Freiburg, Germany, pp 1516–1527
Chelapati CV, Kennedy RP, Wall IB (1972) Probabilistic assessment of hazard for nuclear
structures. Nucl Eng Des 19:333–364
Chen DC (1982) Plasticity in reinforced concrete. McGraw-Hill Book Company, New York, USA
Chen EP (1989) Penetration into dry porous rock: a numerical study on sliding friction simulation.
Theor Appl Fract Mech 11(2):135–141
Chen XW (2005) Mechanics of structural design of EPW (I): The penetration/perforation theory
and the analysis on the cartridge of projectile. Chin J Explosion Shock Waves 25(6):499–505
(in Chinese)
Chen XW, Li QM (2002) Deep penetration of a non-deformable projectile with different
geometrical characteristics. Int J Impact Eng 27(6):619–637
Chen XW, Li QM (2003) Perforation of a thick plate by rigid projectiles. Int J Impact Eng 28
(7):743–759
Chen XW, Li QM (2004a) Transition from nondeformable projectile penetration to semi
hydrodynamic penetration. J Engng Mech 130(1):123–127
Chen XW, Fan SC, Li QM (2004b) Oblique and normal perforation of concrete targets by a rigid
projectile. Int J Impact Eng 30(6):617–637
Chen G, Chen ZF, Tao JL, Niu W, Zhang QP, Huang XC (2005) Investigation and validation on
plastic constitutive parameters of 45 steel. Chin J Explosion Shock Waves 25(5):451–456 (in
Chinese)
Chen XW, Zhang FJ, Yang SQ, Xie RZ (2006) Mechanics of structural design of EPW (III):
investigations on the reduced-scale tests. Chin J Explosion Shock Waves 26(2):105–114 (in
Chinese)
562 References

Chen JL, Li XD, Liu KX (2008a) Experimental research on parameters of constitutive model for a
cement mortar. Acta Scientiarum Naturalium Universitatis Pekinensis 44(5):689–694 (in
Chinese)
Chen XW, Li XL, Huang FL, Wu HJ, Chen YZ (2008b) Normal perforation of reinforced concrete
target by rigid projectile. Int J Impact Eng 35(10):1119–1129
Chen XW, He LL, Yang SQ (2010) Modeling on mass abrasion of kinetic energy penetrator. Eur J
Mech A Solids 29(1):7–17
Chern JC, Yang HJ, Chen HW (1992) Behavior of steel fiber reinforced concrete in multiaxial
loading. ACI Mater J 89(1):32–40
Corbett GG, Reid SR, Johnson W (1996) Impact loading of plates and shells by free-flying
projectiles: a review. Int J Impact Eng 18(2):141–230
Dancygier AN (1997) Effect of reinforcement ratio on the resistance of reinforced concrete to hard
projectile impact. Nucl Eng Des 172(1–2):233–245
Dancygier AN (1998) Rear face damage of normal and high-strength concrete elements caused by
hard projectile impact. ACI Struct J 5(3):291–304
Dancygier AN, Yankelevsky DZ (1996) High strength concrete response to hard projectile impact.
Int J Impact Eng 18(6):583–599
Dancygier AN, Yankelevsky DZ, Haegermann C (2007) Response of high performance concrete
plates to impact of non-deforming projectiles. Int J Impact Eng 34(11):1768–1779
Dancygier AN, Katz A, Benamou D, Yankelevsky DZ (2014) Resistance of double-layer
reinforced HPC barriers to projectile impact. Int J Impact Eng 67:39–51
Davis RN, Neely AM, Jones SE (2004) Mass loss and blunting during high-speed penetration.
J Mech Eng Sci Proc Inst Mech Eng 218:1053–1062
Degen PP (1980) Perforation of reinforced concrete slabs by rigid missiles. J Struct Div ASCE,
106(7)
DOE (2006) Accident analysis for aircraft crash into hazardous facilities. U.S. Washington D.C.
Department of Energy Standard DOE-STD-3014-2006
Dritter K, Gruner P (1976) Calculation of the total force acting upon a rigid wall by projectiles.
Nucl Eng Des 37:231–244
Duvaut G, Lions IJ (1972) Les inequations en mechanique et en physique. Dunod, Paris, France
Ellis BD, DiPaolo BP, McDowell DL, Zhou M (2014) Experimental investigation and multi scale
modeling of ultra-high performance concrete panels subject to blast loading. Int J Impact Eng
69:95–103
Elshenawy T, Li QM (2013) Influences of target strength and confinement on the penetration depth
of an oil well perforator. Int J Impact Eng 54:130–137
Erengil ME, Cargile DJ (2002) Advanced projectile concept for high speed penetration of concrete
targets. Proceedings of 20th international symposium on ballistics, Orlando
Espinosa HD, Zavattieri PD, Emore GL (1998) Adaptive FEM computation of geometric and
material nonlinearities with application to brittle failure. Mech Mater 29(3):275–305
Etse G, Willam K (1994) Fracture energy formulation for inelastic behavior of plain concrete.
J Eng Mech 120(9):1983–2011
Fan SC, Zhou XQ (2000) Constitutive model for reinforced concrete against penetration. Final
report, PTRC-CSE/DSO/2000.01, Protective Technology Research Centre, School of Civil and
Structural Engineering. Nanyang Technological University, Singapore
Fan JB, Zu J, Xu P, Wang Y (2012) Impact deceleration signal processing for concrete target
penetration. J Detection Control 34(4):1–5 (in Chinese)
Fanella DA, Naaman AE (1985) Stress-strain properties of fiber reinforced mortar in compression.
ACI J 82(4):475–483
Fang Q, Zhang JH (2012) Three-dimensional numerical modelling of concrete-like materials
subjected to dynamic loadings. In: Advances in protective structures, pp 33–64, CRC Press,
USA
Fang Q, Zhang JH (2013) Three-dimensional modelling of steel fiber reinforced concrete material
under intense dynamic loading. Constr Build Mater 44:1318–1323
References 563

Fang Q, Kong XZ, Hong J, Wu H (2014a) Prediction of projectile penetration and perforation by
finite cavity expansion method with the free-surface effect. Acta Mech Solida Sin 27(6):
597–611
Fang Q, Kong XZ, Wu H, Gong ZM (2014b) Determination of Holmquist-Johnson-Cook
consitiutive model parameters of rock. Eng Mech 31(3):197–204 (in Chinese)
Fang Q, Hong J, Zhang JH, Chen L, Ruan Z (2014c) Issues of SHPB test on concrete-like material.
Eng Mech 31(5):1–14 (in Chinese)
Farnam Y, Mohammadi S, Shekarchi M (2010a) Experimental and numerical investigations of low
velocity impact behavior of high-performance fiber-reinforced cement based composite. Int J
Impact Eng 37(2):220–229
Farnam Y, Moosavi M, Shekarchi M, Babanajad SK, Bagherzadeh A (2010b) Behaviour of slurry
infiltrated fibre concrete (SIFCON) under triaxial compression. Cem Concr Res 40:1571–1581
Feng J, Li WB, Wang XM, Song ML, Ren HQ, Li WB (2015) Dynamic spherical cavity
expansion analysis of rate-dependent concrete material with scale effect. Int J Impact Eng
84:24–37
Field JE, Walley SM, Proud WG, Goldrein HT, Siviour CR (2004) Review of experimental
techniques for high rate deformation and shock studies. Int J Impact Eng 30(7):725–775
Forrestal MJ (1986) Penetration into dry porous rock. Int J Solids Struct 22:1485–1500
Forrestal MJ, Hanchak SJ (2002) Penetration limit velocity for ogive-nose projectiles and
limestone targets. Trans ASME J Appl Mech 69(6):853–854
Forrestal MJ, Longcope DB (1990) Target strength of ceramic materials for high-velocity
penetration. J Appl Phys 67(8):3669–3672
Forrestal MJ, Luk VK (1988) Dynamic spherical cavity-expansion in a compressible elastic-plastic
solid. Trans ASME J Appl Mech 55:275–279
Forrestal MJ, Luk VK (1992) Penetration into soil targets. Int J Impact Eng 12(3):427–444
Forrestal MJ, Luk VK (1992) Penetration of 7075-T651 aluminum targets with ogival-nose rods.
Int J Solids Struct 29:1729–1736
Forrestal MJ, Piekutowski AJ (2000) Penetration experiments with 6061-T6511 aluminium targets
and spherical-nose steel projectiles at striking velocities between 0.5 and 3.0 km/s. Int J Impact
Eng 24:57–67
Forrestal MJ, Tzou DY (1997) A spherical cavity-expansion penetration model for concrete
targets. Int J Solids Struct 34:4127–4146
Forrestal MJ, Warren TL (2008) Penetration equations for ogive-nose rods into aluminum targets.
Int J Impact Eng 35(8):727–730
Forrestal MJ, Longcope DB, Norwood FR (1981) A model to estimate forces on conical
penetrators into dry porous rock. J Appl Mech 48:25–29
Forrestal MJ, Okajima K, Luk VK (1988) Penetration of 6061-T651 aluminum targets with rigid
long rods. ASME J Appl Mech 55:755–760
Forrestal MJ, Brar NS, Luk VK (1991) Penetration of strain-hardening targets with rigid
spherical-nose rods. Trans ASME J Appl Mech 58(1):7–10
Forrestal MJ, Cargile JD, Tzou RDY (1993) Penetration of concrete targets. Sandia National
Laboratory, SAND-92-2513C
Forrestal MJ, Altman BS, Cargile JD, Hanchak SJ (1994) An empirical equation for penetration
depth of ogive-nose projectiles into concrete targets. Int J Impact Eng 15:395–405
Forrestal MJ, Tzou DY, Askari E, Longscope BD (1995) Penetration into ductile metal targets
with rigid spherical-nose rods. Int J Impact Eng 16(5):699–710
Forrestal MJ, Frew DJ, Hanchak SJ, Brar NS (1996) Penetration of grout and concrete targets with
ogive-nose steel projectiles. Int J Impact Eng 18(5):465–476
Forrestal MJ, Frew DJ, Hickerson JP, Rohwer TA (2003) Penetration of concrete targets with
deceleration-time measurements. Int J Impact Eng 28(5):479–497
Forrestal MJ, Togami TC, Baker WE, Frew DJ (2003) Performance evaluation of accelerometers
used for penetration experiments. Exp Mech 43(1):90–96
564 References

Forrestal MJ, Børvik T, Warren TL (2010) Perforation of 7075-T651 aluminum armor plates with
7.62 mm APM2 bullets. Exp Mech 50(8):1245–1251
Foster JT, Frew DJ, Forrestal MJ, Nishida EE, Chen W (2012) Shock testing accelerometers with a
Hopkinson pressure bar. Int J Impact Eng 46:56–61
Frano RL, Forasassi G (2011) Preliminary evaluation of aircraft impact on a near term nuclear
power plant. Nucl Eng Des 241(12):5245–5250
Frew DJ, Hanchak SJ, Green ML, Forrestal MJ (1998) Penetration of concrete targets with
ogive-nose steel rods. Int J Impact Eng 21(6):489–497
Frew DJ, Forrestal MJ, Hanchak SJ (2000) Penetration experiments with limestone targets and
ogive-nose steel projectiles. J Appl Mech 6:841–845
Frew DJ, Forrestal MJ, Cargile JD (2006) The effect of concrete target diameter on projectile
deceleration and penetration depth. Int J Impact Eng 32(10):1584–1594
Fu JP, Chen ZG, Hou XC, Li SQ, Li SC, Wang JW (2013) Simulation and experimental
investigation of jetting penetrator charge at large stand-off distance. Defence Technology 9
(2):91–97
Fullard K, Baum MR, Barr P (1991) The assessment of impact on nuclear power plant structures in
the United Kingdom. Nucl Eng Des 130:113–120
Gao SQ, Jin L, Liu HP (2004) Dynamic response of a projectile perforating multi-plate concrete
targets. Int J Solids Struct 41(18–19):4927–4938
Gebara JM, Pan AD, Anderson JB (1993) Analysis of rock-rubble overlay protection of structures
by the finite block method. In: The sixth international symposium interaction of nonnuclear
munitions with structure, Panama City Beach, Florida, pp 33–43
Gebbeken N, Ruppert M (2000) A new material model for concrete in high-dynamic hydrocode
simulations. Arch Appl Mech 70(7):463–478
Gebbeken N, Greulich S, Pietzsch A (2006) Hugoniot properties for concrete determined by
full-scale detonation experiments and flyer-plate-impact tests. Int J Impact Eng 32:2017–2031
Gelman MD, Richard BN, Ito YM (1987) Impact of AP projectile into array of large caliber
boulders. Technical report no. SL-87-30, U.S. Army Engineer Waterways Experiment Station,
Vicksburg, Mississippi
Girgin ZC, Anoglu N, Anoglu E (2007) Evaluation of strength criteria for very-high-strength
concretes under triaxial compression. ACI Struct J 104(3):277–283
Gold VM, Vradis GC, Pearson JC (1996) Penetration of concrete by eroding projectiles:
experiments and analysis. J Eng Mech ASCE 122(2):145–152
Goldsmith W (1999) Non-ideal impact projectile on targets. Int J Impact Eng 22:95–395
Gomeza JT, Shukla A (2001) Multiple impact penetration of semi-infinite concrete. Int J Impact
Eng 25(10):976–979
Goodier JN (1965) On the mechanics of indentation, cratering in solid targets of strain-hardening
metal by impact of hard and soft spheres. AIAA proceedings of the seventh symposium on
hypervelocity impact, vol III, pp 215–259
Grady DE (1996a) Dynamic decompression properties of concrete from Hugoniot states 3 to 25
GPa. Sandia National Laboratories, USA
Grady DE (1996b) Shock equation of state properties of concrete. International conference on
structures under shock and impact, Udine, Italy
Gran JK, Frew DJ (1997) In-target stress measurements from penetration experiments into
concrete by ogive-nose steel projectiles. Int J Impact Eng 19(8):715–726
Graybeal BA (2007) Material property characterization of ultra-high performance concrete. VA,
USA: FHWA-HRT-06-103, U.S. Department of Transportation, pp 1–186
Greenberger M (1961) An a priori determination of serial correlation in random numbers. Math
Comput 15:383–389
Grisaro H, Dancygier AN (2014a) A modified energy method to assess the residual velocity of
non-deforming projectiles that perforate concrete barriers. Int J Prot Struct 5(3):307–321
Grisaro H, Dancygier AN (2014b) Assessment of the perforation limit of a composite RC barrier
with a rear steel liner to impact of a non-deforming projectile. Int J Impact Eng 64:122–136
References 565

Grove B, Heiland J, Walton I (2008) Geologic materials’ response to shaped charge penetration.
Int J Impact Eng 35(12):1563–1566
Gwaltney RC (1968) Missile generation and protection in light water-cooled reactor power plants,
ORNL NSIC-22. Oak Ridge National Laboratory, Oak Ridge
Habel K, Gauvreau P (2008) Response of ultra-high performance fiber reinforced concrete
(UHPFRC) to impact and static loading. Cem Concr Compos 30(10):938–946
Haldar A, Hamieh H (1984) Local effect of solid missiles on concrete structures. J Struct
Div ASCE 110(5):948–960
Hallquist JO (2007) LS-DYNA keyword user’s manual. Livermore Software Technology
Corporation, Version, p 971
Hanchak SJ, Forrestal MJ, Young ER, Ehrgott JQ (1992) Perforation of concrete slab with 48MPa
(7ksi) and 140MPa (20ksi) unconfined compressive strength. Int J Impact Eng 12(1):1–7
Hao H, Hao Y, Li ZX (2012) Numerical quantification of factors influencing high-speed impact
tests of concrete material. IAPS Special Publication, pp. 97–130
Hartmann T, Pietzsch A, Gebbeken N (2010) A hydrocode material model for concrete. Int J
Protect Struct 1(4):443–468
Hashimoto J, Takiguchi K, Nishimura K, Matsuzawa K, Tsutsui M, Ohashi Y, Kojima I, Torita H
(2005) Experimental study on behavior of RC panels covered with steel plates subjected to
missile impact. In: Proceedings of 18th international conference on international association for
structural mechanics in reactor technology (SMIRT), Beijing
He LL, Chen XW (2011) Analyses of the penetration process considering mass loss. Eur J Mech A
Solids 30(2):145–157
He X, Xu XY, Sun GJ, Shen J, Yang JC, Jin DL (2010a) Experimental investigation on projectiles
high-velocity penetration into concrete targets. Chin J Explosion Shock Waves 30(1):1–6 (in
Chinese)
He LL, Chen XW, He X (2010b) Parametric study on mass loss of penetrators. Acta Mech Sin 26
(4):585–597
He T, Wen HM, Guo XJ (2011) A spherical cavity expansion model for penetration of
ogival-nosed projectiles into concrete targets with shear-dilatancy. Acta Mech Sin 27:
1001–1012
Herrmann W (1969) Constitutive equation for the dynamic compaction of ductile porous materials.
J Appl Phys 40:2490–9
Heuze FE (1990) An over review of projectile penetration into geological materials with emphasis
on rocks. Int J Rock Mech Min Sci Geomech Abstr 27(1):1–14
Hill R (1948) A theory of earth movement near a deep underground explosion. Memo no. 21–48,
Armament Research Establishment, Front Halstead, Kent, UK
Holmquist TJ, Johnson GR, Cook WH (1993) A computational constitutive model for concrete
subjected to large strains, high strain rates and high pressures. In: Proceedings of 14th
international symposium on Ballistics, Quebec, Canada, pp 591–600
Hopkins HG (1960) Dynamic expansion of spherical cavities in metals. In: Sneddon IN, Hill R
(eds) Progress in soli mechanics, vol 1. North-Holland Publishing Co., Amsterdam, New York
(Chapter 3)
Hornyik K (1977) Analytic modeling of the impact of soft missiles on protective walls, 4th
SMiRT, USA, Paper no. J7/3
Hou ZG (2006) Research on concrete strength under triaxial stresses. Dissertation, Hebei
University of Technology, Hebei (in Chinese)
Hu CM, He HL, Hu SS (2003) A study on dynamic mechanical behaviors of 45 steel. Explo Shock
Waves 23(2):188–192 (in Chinese)
Huang FL, Wu HJ, Jin QK, Zhang QM (2005) A numerical simulation on the perforation of
reinforced concrete targets. Int J Impact Eng 32(1):173–187
Huang FL, Zhang LL, Duan ZP (2008) Experimental study about shaped charge with large cone
angle penetration to concrete target. Chin J Explosion Shock Waves 28(1):17–22 (in Chinese)
566 References

Huang JR, Liu RC, He X, Sun GJ, Xu P (2009) A new data processing technique for measured
penetration overloads. Explos Shock Waves 29(5):555–560 (in Chinese)
Huerta M, Vigil MG (2006) Design, analyses, and field test of a 0.7 m conical shaped charge. Int J
Impact Eng 32(8):1201–1213
Hughes G (1984) Hard missile impact on reinforced concrete. Nucl Eng Des 77:23–35
IAEA (2003) Safety standard series: external events excluding earthquakes in the design of nuclear
power plants. Safety guide no. NS-G-1.5
Imran I, Pantazopoulou SJ (1996) Experimental study of plain concrete under triaxial stress. ACI
Mater J 93(6):589–601
Iqbal MA, Rai S, Sadique MR, Bhargava P (2012) Numerical simulation of aircraft crash on
nuclear containment structure. Nucl Eng Des 243:321–335
Iravani S (1996) Mechanical properties of high-performance concrete. ACI Mater J 93(5):416–426
Islam MJ, Swaddiwudhipong S, Liu ZS (2012) Penetration of concrete targets using a modified
Holmquist-Johnson-Cook material model. Int J Comp Meth 9(4):1–19
Itoh M, Katayama M, Rainsberger R (2005) Computer simulation of an F-4 Phantom crashing into
a reinforced concrete wall. WIT transactions on modelling and simulation, pp 207–217
Jan AT, Henrik S (2004) Penetration into concrete by truncated projectiles. Int J Impact Eng 30
(4):447–464
Jankov ZD, Shanahan JA, White MP (1976) Missile tests of quarter-scale reinforced concrete
barriers. In: Proceedings of a symposium on tornadoes, assessment of knowledge and
implications for man. Texas Technical University, Lubbock, TX
Jena PK, Ramanjeneyulu Siva KK, Balakrishna B (2009) Ballistic studies on layered structures.
Mater Des 30:1922–1929
Jiang H, Chorzepa MG (2014) Aircraft impact analysis of nuclear safety-related concrete
structures: a review. Eng Fail Anal 46:118–133
Johnson GR, Cook WH (1983) A constitutive model and data for metals subjected to large strains,
high strain rates and high temperatures. In: Proceedings of the 7th international symposium on
Ballistics. The Hague, Netherlands, pp 541–547
Johnson GR, Stryk RA (2003) Conversion of 3D distorted elements into meshless particles during
dynamic deformation. Int J Impact Eng 28(9):947–966
Johnson GR, Beissel SR, Holmquist TJ, Frew DJ (1998) Computed radial stresses in a concrete
target penetrated by a steel projectile. In: Jones N, Talaslidis DG, Brebbia CA, Manolis GD
(eds) Structures under shock and impact V. Computational Mechanics Publications,
Southampton, UK and Boston, USA, pp 793–806
Johnson GR, Stryk RA, Beissel SR, Holmquist TJ (2002) An algorithm to automatically convert
distorted finite elements into meshless particles during dynamic deformation. Int J Impact Eng
27(10):997–1013
Jones N (2011) Structural impact. Second edition. Cambridge University Press, Cambridge
Jones SE, William KR (2000) On the optimal nose geometry for a rigid penetrator, including the
effects of pressure-dependent friction. Int J Impact Eng 24(4):403–415
Jones SE, William KR (2008) Hypervelocity impact penetration mechanics. Int J Impact Eng 35
(12):1654–1660
Jones SE, Foster JC, Toness OA, DeAngelis RJ, Rule WK (2002) An estimate for mass loss from
high velocity steel penetrators. In: Moody FJ (ed) Proceedings of the ASME PVP-435
conference on thermal-hydraulic problems, sloshing phenomena, and extreme loads on
structures. ASME, New York, pp 227–237
Kar AK (1978) Local effects of tornado generated missiles. ASCE J Struct Div 104(ST5):809–816
Kar AK (1979) Impactive effects of Tornado missiles and aircraft. J Struct Div 105(ST1):2243–
2260
Karen L, Scrivener R, James K (2008) Innovation in use and research on cementitious material.
Cem Concr Res 38(2):128–136
Kennedy RP (1966) Effects of an aircraft crash into a concrete reactor containment building.
Holmes & Narver Inc, Anaheim, CA
References 567

Kennedy RP (1976) A review of procedures for the analysis and design of concrete structures to
resist missile impact effects. Nucl Eng Des 37(2):183–203
Kojima I (1991) An experimental study on local behaviour of reinforced concrete slabs to missile
impact. Nucl Eng Des 130:121–132
Kong XZ, Fang Q, Wu H (2014) Terminal ballistics study of deformable projectile penetrating
brittle material targets for free-surface and crack region effects. Acta Armamentarii 35(6):814–
821 (in Chinese)
Kong XZ, Fang Q, Wu H, Peng Y (2016a) Numerical predictions of cratering and scabbing in
concrete slabs subjected to projectile impact using a modified version of HJC material model.
Int J Impact Eng 95:61–71
Kong XZ, Wu H, Fang Q, Ren GM (2016b) Analyses of rigid projectile penetration into UHPCC
target based on an improved dynamic cavity expansion model. Constr Build Mater 126:759–
767
Kong XZ, Wu H, Fang Q, Peng Y (2017) Rigid and eroding projectile penetration into concrete
targets based on an extended dynamic cavity expansion model. Int J Impact Eng 100:13–22
Kupfer H, Hilsdorf H, Rush H (1969) Behaviour of concrete under biaxial stresses. ACI J 23:656–
665
Lan S, Guo Z (1997) Experimental investigation of multiaxial compressive strength of concrete
under different stress paths. ACI Mater J 94(5):427–434
Lan B, Wen HM (2010) Alekseevskii-Tate revisited: an extension to the modified hydrodynamic
theory of long-rod penetration. Sci China Ser E 53(5):1364–1373
Langberg H, Markeset G (1999) High performance concrete-penetration resistance and material
development. In: Proceedings of the 9th international symposium on interaction of the effects
of munitions with structures, Berlin-Strausberg, pp 933–941
Langheim H, Pahl H, Schmolinske E, Stilp AJ (1993) Subscale penetration tests with bombs and
advanced penetration against hardened structures. The sixth international symposium on
interaction of nonnuclear munitions with structures, Panama City Beach, Florida, pp 12–17
Lawver D, Tennant D, Mould J, Levine H (2002) Impact of aircraft engines into reinforced
concrete walls. ASME 2002 pressure vessels and piping conference, pp 209–218
Lee HK, Kim SE (2013) Erosion sensitivity assessment of SC walls under aircraft impact. In:
Proceedings of 22nd conference on structural mechanics in reactor technology, San Francisco
Lee K, Han SE, Hong JW (2013) Analysis of impact of large commercial aircraft on a prestressed
containment building. Nucl Eng Des 265:431–449
Lee K, Jung JW, Hong JW (2014) Advanced aircraft analysis of an F-4 Phantom on a reinforced
concrete building. Nucl Eng Des 273:505–528
Leppänen J (2006) Concrete subjected to projectile and fragment impacts: Modelling of crack
softening and strain rate dependency in tension. Int J Impact Eng 32(11):1828–1841
Li Q, Ansari F (1999) Mechanics of damage and constitutive relationships for high-strength
concrete in triaxial compression. J Eng Mech 125(1):1–10
Li Q, Ansari F (2000) High-strength concrete in triaxial compression by different size of
specimens. ACI Mater J 97(6):684–689
Li QM, Chen XW (2003) Dimensionless formulae for penetration depth of concrete targets
impacted by rigid projectiles. Int J Impact Eng 28(1):93–116
Li QM, Flores-Johnson EA (2011) Hard projectile penetration and trajectory stability. Int J Impact
Eng 38(10):815–823
Li J, Hao H (2014) Numerical study of concrete spall damage to blast loads. Int J Impact Eng
68:41–55
Li QM, Meng H (2003) About the dynamic strength enhancement of concrete-like materials in a
split Hopkinson pressure bar test. Int J Solids Struct 40(2):343–360
Li QM, Tong DJ (2003) Perforation thickness and ballistic limit of concrete target subjected to
rigid impact. ASCE J Eng Mech 129(9):1083–1091
Li QM, Weng HJ, Chen XW (2004) A modified model for the penetration into moderately thick
plates by a rigid, sharp-nosed projectile. Int J Impact Eng 30(2):193–204
568 References

Li QM, Reid SR, Wen HM, Telford AR (2005) Local impact effects of hard missiles on concrete
targets. Int J Impact Eng 32(1):224–284
Li QM, Reid SR, Ahmad-Zaidi AM (2006) Critical impact energies for scabbing and perforation
of concrete target. Nucl Eng Des 236(11):1140–1148
Li QM, Ye ZQ, Ma GW, Reid SR (2007) Influence of overall structural response on perforation of
concrete targets. Int J Impact Eng 34:926–941
Li QM, Lu YB, Meng H (2009) Further investigation on the dynamic compressive strength
enhancement of concrete-like materials based on split Hopkinson pressure bar tests. Part II:
numerical simulations. Int J Impact Eng 36(12):1335–1345
Li WB, Wang XM, Li WB (2010) The effect of annular multi-point initiation on the formation and
penetration of an explosively formed penetrator. Int J Impact Eng 37(4):414–424
Li JZ, Lv ZJ, Zhang HS, Huang FL (2013) Perforation experiments of concrete target with residual
velocity measurements. Int J Impact Eng 57:1–6
Liang B, Chen XW, Ji YQ, Huang HJ, Gao HY, Li XL, Huang HY (2008) Experimental study on
deep penetration of reduced-scale advanced earth penetrating weapon. Chin J Explosion and
Shock Waves 28(1):1–9 (in Chinese)
Linderman RB, Fakhari M, Rotz JV (1973) Design of structures for missile impact. BC-TOP-9,
Rev. 1, Bechtel Power Corporation, San Francisco
Liu WK, Jun S, Zhang YF (1995) Reproducing kernel particle methods. Int J Numer Meth Fl 20
(8–9):1081–1106
Liu RC, Wu B, Zhang XZ, Shen GS, Niu XL (2002) Tests on resisting projectiles penetration of
high strength volume steel fiber concrete. Chin J Explosion and Shock Waves 22(4):368–372
(in Chinese)
Liu Y, Ma A, Huang F (2009) Numerical simulations of oblique-angle penetration by deformable
projectiles into concrete targets. Int J Impact Eng 36(3):438–446
Longcope DB, Tabbara MR, Jung J (1999) Modeling of oblique penetration into geologic targets
using cavity expansion penetrator loading with target free-surface effects. SAND99-1104,
Sandia National Laboratories, Albuquerque, NM
Lu X, Hsu CT (2006) Behavior of high strength concrete with and without steel fiber
reinforcement in triaxial compression. Cem Concr Res 36(9):1679–1685
Lu YB, Li QM (2011a) A correction methodology to determine the strain-rate effect on the
compressive strength of brittle materials based on SHPB testing. Int J Prot Struct 2(1):127–138
Lu YB, Li QM (2011b) About the dynamic uniaxial tensile strength of concrete-like materials.
Int J Impact Eng 38:171–80
Luk VK, Forrestal MJ (1987) Penetration into semi-infinite reinforced-concrete targets with
spherical and ogival nose projectiles. Int J Impact Eng 6:291–301
Luk VK, Forrestal MJ, Amos DE (1991) Dynamic spherical cavity expansion of strain-hardening
materials. Trans ASME J Appl Mech 58(1):1–6
Lundgren RG (1994) High velocity penetrator. SAND94-2724C, Conf-9411142-1
Luo X, Sun W, Chan SYN (2000) Characteristics of high-performance steel fiber-reinforced
concrete subject to high velocity impact. Cem Concr Res 30(6):907–914
Maalej M, Quek ST, Zhang J (2005) Behavior of hybrid-fiber engineered cementitious composites
subjected to dynamic tensile loading and projectile impact. J Mater Civil Eng 17(2):143–152
Máca P, Sovják R, Vavřinik T (2013) Experimental investigation of mechanical properties of
UHPFRC. Procedia Eng 65:14–19
Máca P, Sovják R, Konvalinka P (2014) Mix design of UHPSFRC and its response to projectile
impact. Int J Impact Eng 63:158–163
Macek RW, Duffey TA (2000) Finite cavity expansion method for near-surface effects and
layering during earth penetration. Int J Impact Eng 24(3):239–258
Magallanes JM, Wu YC, Malvar LJ, Crawford JE (2010) Recent improvements to release III of the
K&C concrete model. In: 11th international LS-DYNA users conference, pp 6–8
References 569

Mahmud GH, Yang ZJ, Hassan AMT (2013) Experimental and numerical studies of size effects of
ultra high performance steel fibre reinforced concrete (UHPFRC) beams. Constr Build Mater
48:1027–1034
Malvar LJ, Crawford JE (1998) Dynamic increase factors for concrete. In: Proceedings of the 28th
DDESB seminar, Orlando, FL. ANSI Std, pp 1–17
Malvar LJ, Ross CA (1998) Review of strain rate effects for concrete in tension. ACI Mater J 95
(6):735–739
Malvar LJ, Simons D (1996) Concrete material modeling in explicit computations. In:
Proceedings, workshop on recent advances in computational structural dynamics and high
performance computing, USAE waterways experiment station, Vicksburg, MS, pp 165–194
Malvar LJ, Crawford JE, Wesevich JW, Simons D (1997) A plasticity concrete material model for
DYNA3D. Int J Impact Eng 19(9):847–873
Malvar LJ, Crawford JE, Morrill KB (2000) K&C concrete material model Release III-automated
generation of material model input. Karagozian and case structural engineers, Technical report
TR-99-24.3
Mao L, Barnett S, Begg D, Schleyer G, Wight G (2014) Numerical simulation of ultra high
performance fibre reinforced concrete panel subjected to blast loading. Int J Impact Eng 64:
91–100
Markovich N, Kochavi E, Ben-Dor G (2011) An improved calibration of the concrete damage
model. Finite Elem Anal Des 47(11):1280–1290
Marsh SP (1980) LASL shock Hugoniot data. University of California Press, California
Meyer CS (2011) Development of geomaterial parameters for numerical simulations using the
Holmquist-Johnson-Cook constitutive model for concrete. ARL-TR-5556, U.S. Army
Research Laboratory, USA
Mu ZC, Zhang W (2011) An investigation on mass loss of ogival projectiles penetrating concrete
targets. Int J Impact Eng 38(6):770–778
Murphy MJ (1983) Shaped charge penetration in concrete: a unified approach. No. UCRL-53393.
Lawrence Livermore National Lab, CA, USA
Murphy MJ, Baum DW, Clark DB, Mcguire EM, Simonson SC (2000) Numerical simulation of
damage and fracture in concrete from shaped charge jets. Lawrence Livermore National
Laboratory (LLNL), Livermore, CA, UCRL-JC-140416
Murphy MJ, Randers-Pehrson G, Kuklo RM, Rambur TA, Switzer LL, Summes MA (2003)
Experiments and simulations of penetration into granite by an aluminum shaped charge.
J Phys IV 110(9):603–608
Mutalib AA (2011) Damage assessment and prediction of FRP strengthened RC structures
subjected to blast and impact loads. University of West Australia. Doctoral thesis
Naaman AE, Homrich JR (1989) Tensile stress-strain properties of SIFCON. ACI Mater J 86
(3):244–251
Naaman AE, Reinhardt HW (1996) Characterization of high performance fiber reinforced cement
composites-HPFRCC. High performance fiber reinforced cement composites, vol 2. Ann
Arbor, pp 1–24
Nataraja MC, Dhang N, Gupta AP (1999) Stress-strain curves for steel-fiber reinforced concrete
under compression. Cem Concr Compos 21:383–390
NDRC (1946) Effects of impact and explosion. Summary technical report of division 2, vol 1,
National Defence Research Committee, Washington, DC
NEI (2011) Methodology for performing aircraft impact assessments for new plant designs.
Washington, DC
Nia AA, Zolfaghari M, Khodarahmi H, Nili M, Ghorbankhani AH (2014) High velocity
penetration of concrete targets with eroding long-rod projectiles: an experiment and analysis.
J Protect Struct 5(1):47–63
NRC (2009) Consideration of aircraft impacts for new nuclear power reactors. RIN 3150-AI19,
Washington, DC
570 References

O’Neil EF, Neeley BD, Cargile JD (1999) Tensile properties of very-high-strength concrete for
penetration-resistant structures. Shock Vib 6(5–6):237–245
Ong CW, Boey CW, Hixson RS, Sinibaldi JO (2003) Advanced layered personnel armor. Int J
Impact Eng 38:369–383
Ortiz M (1996) Computational micromechanics. Comput Mech 18(5):321–338
Osipenko KY (2009) Penetration of a body of revolution into an elasto-plastic medium. Mech
Solids 44(2):311–321
Ottosen NS (1977) A failure criterion for concrete. ASCE J Eng Mech Div 103(4):527–535
Peng Y, Wu H, Fang Q, Gong ZM, Kong XZ (2015a) A note on the deep penetration and
perforation of hard projectiles into thick targets. Int J Impact Eng 85:37–44
Peng Y, Fang Q, Wu H, Gong ZM, Kong XZ (2015b) Discussion on the resistance forcing
function of projectiles penetrating into concrete targets. Eng Mech 32(4):112–119 (in Chinese)
Peng Y, Wu H, Fang Q, Liu JZ, Gong ZM (2016a) Impact resistance of basalt aggregated
UHP-SFRC/fabric composite panel against small caliber arm. Int J Impact Eng 88:201–213
Peng Y, Wu H, Fang Q, Liu JZ, Gong ZM (2016b) Flat nosed projectile penetrating into
UHP-SFRC target: experiment and analysis. Int J Impact Eng 93:88–98
Peng Y, Wu H, Fang Q, Liu JZ, Gong ZM (2016c) Residual velocities of projectiles after normally
perforating the thin ultra-high performance steel fiber reinforced concrete slabs. Int J Impact
Eng 97:1–9
Perzyna P (1966) Fundamental problems in viscoplasticity. Adv Appl Mech, vol 9. Academic
Press, New York, pp 243–377
Petr M, Radoslav S, Tomáš V (2013) Experimental investigation of mechanical properties of
UHPFRC. Procedia Eng 65:14–19
Petrangeli G (2010) Large airplane crash on a nuclear plant: design study against excessive
shaking of components. Nucl Eng Des 240:4037–4042
Petry L (1910) Monographies de Systemes d’Artillerie. Brussels
Pi AG, Huang FL (2007) Dynamic behavior of a slender projectile on oblique penetrating into
concrete target. Chin J Explosion Shock Waves 27(4):331–337 (in Chinese)
Piekutowski AJ, Forrestal MJ, Poormon KL, Warren TL (1999) Penetration of 6061-T6511
aluminium targets by ogive-nose steel projectiles with striking velocities between 0.5 and 3.0
km/s. Int J Impact Eng 23:723–734
Piotrowska E, Malecot Y, Ke Y (2014) Experimental investigation of the effect of coarse aggregate
shape and composition on concrete triaxial behavior. Mech Mater 79:45–57
Polanco-Loria M, Hopperstad OS, Børvik T, Berstad T (2008) Numerical predictions of ballistic
limits for concrete slabs using a modified version of the HJC concrete model. Int J Impact Eng
35(5):290–303
Poon CS, Shui ZH, Lam L (2004) Compressive behavior of fiber reinforced high-performance
concrete subjected to elevated temperatures. Cem Concr Res 34(12):2215–2222
Qian LX, Yang YB, Liu T (2000) A semi-analytical model for truncated-ogive-nose projectiles
penetration into semi-infinite concrete targets. Int J Impact Eng 24(9):947–955
Ranjan R, Banerjee S, Singh RK, Banerji P (2014) Local impact effects on concrete target due to
missile: an empirical and numerical approach. Ann Nucl Energy 68:262–275
Rebentrost M, Wight G (2009) Investigation of UHPFRC slabs under blast loads. In:
Toutlemonde F, Resplendino J (eds) Proceedings of designing and building with UHPFRC:
state-of-the-art, designing and building with UHPFRC. Marseille, Wiley, France, pp 363–376
Reid SR, Wen HM (2001) Predicting penetration, cone cracking, scabbing and perforation of
reinforced concrete targets struck by flat-faced projectiles. UMIST report
ME/AM/02.01/TE/G/018507/Z
Ren GM, Wu H, Fang Q, Liu JZ, Gong ZM (2016) Triaxial compressive behavior of UHPCC and
applications in the projectile impact analyses. Constr Build Mater 113:1–14
Resnyansky AD, Weckert S (2009) Response of an ultrahigh performance concrete to shaped
charge jet. In: Proceedings of 8th international conference on Shock and Impact Loads on
Structures. Adelaide, Australia, pp 28–36
References 571

Resnyansky AD, Katselis G, Wildegger-Gaissmaier AE (2004) Experimental and numerical study

of the shaped charge jet perforation against concrete target. In: Proceedings of 21st
international symposium on Ballistics, CD-ROM proceedings, additional entries, paper1
Richard P, Cheyrezy M (1995) Composition of reactive powder concrete. Cem Concr Res 25
(7):1501–1511
Riedel W, Thoma K, Hiermaier S (1999) Penetration of reinforced concrete by BETAB-500
numerical analysis using a new macroscopic concrete model for hydrocodes. In: Proceedings of
9th international symposium on interaction of the effect of munitions with structures.
Strausberg, Berlin, pp 315–322
Riedel W, Nöldgen M, Straßburger E, Thoma K, Fehling E (2010) Local damage to ultra high
performance concrete structures caused by an impact of aircraft engine missile. Nucl Eng Des
240(10):2633–2642
Riera JD (1968) On the stress analysis of structures subjected to aircraft impact forces. Nucl Eng
Des 8:415–426
Riera JD (1980) A critical reappraisal of nuclear power plant safety against accidental aircraft
impact. Nucl Eng Des 57:193–206
Riera JD, Zorn NF, Schuëller GI (1982) An approach to evaluate the design load time history for
normal engine impact taking into account the crash-velocity distribution. Nucl Eng Des 71
(3):311–316
Rodriguez-Nikl T, Lee C, Hegemier G, Seible F (2012) Experimental performance of concrete
columns with composite jackets under blast loading. Struct Eng 138:81–89
Rohani B (1987) Penetration of kinetic energy projectiles into rock-rubble/boulder overlays.
Technical report, U.S. Army Engineer Waterways Experiment Station, Vicksburg, Mississippi
Rong ZD, Sun W (2012) Experimental and numerical investigation on the dynamic tensile
behavior of ultra-high performance cement based composites. Constr Build Mater 31:168–173
Rong ZD, Sun W, Zhang YS (2010) Dynamic compression behavior of ultra-high performance
cement based composites. Int J Impact Eng 37(5):515–520
Rosenberg Z, Dekel E (2009) Analytical solution of the spherical cavity expansion process. Int J
Impact Eng 36(2):193–198
Rosenberg Z, Dekel E (2010) Revisiting the perforation of ductile plates by sharp-nosed rigid
projectiles. Int J Solids Struct 47(22–23):3022–3033
Rosenberg Z, Dekel E (2012) Terminal ballistics. Springer, Berlin
Rosenberg Z, Dekel E (2016) Terminal ballistics, 2nd edn. Springer, Berlin
Rosenberg Z, Kositski R (2016) Modeling the penetration and perforation of concrete targets by
rigid projectiles. Int J Prot Struct 7(1):1–22
Rosenberg Z, Marmor E, Mayseless M (1990) On the hydrodynamic theory of long-rod
penetration. Int J Impact Eng 10(1):483–486
Rossi P, Arca A, Parant E (2005) Fakhri P. Bending and compressive behaviors of a new cement
composite. Cem Concr Res 35(1):27–33
Rotz JV (1976) Evaluation of tornado missile impact effects on structures. In: Proceedings of a
symposium on tornadoes, assessment of knowledge and implications for man. Texas Technical
University, Lubbock, TX
Sadique MR, Iqbal MA, Bhargava P (2013) Nuclear containment structure subjected to
commercial and fighter aircraft crash. Nucl Eng Des 260:30–46
Sagals G, Orbovic N, Blahoianu A (2011) Sensitivity studies of reinforced concrete slabs under
impact loading. Presented at the transactions of the 21st SMiRT, New Delhi, India
Sagomonyan AY (1974) Penetration: penetration of solids into compressible continuous Media.
Moscow State University Press, Moscow (in Russian)
Samudre SS, Nair UR, Gore GM (2009) Studies on an improved plastic bonded explosive
(PBX) for shaped charges. Propell Explos Pyrot 34(2):145–150
Samuely FJ, Hamann CW (1939) Civil protection. The Architectural Press, USA
Saran S, Ayısıt O, Yavuz MS (2013) Experimental investigations on aluminum shaped charge
liners. Procedia Eng 58:479–486
572 References

Satapathy SS (1997) Application of cavity expansion analysis to penetration problems. Ph.D.

thesis, University of Texas at, Austin
Schuler H, Mayrhofer C, Thoma K (2006) Spall experiments for the measurement of the tensile
strength and fracture energy of concrete at high strain rates. Int J Impact Eng 32:1635–1650
Schwalbe LA, Wingate CA, Stofleth JH, Greene RW (1996) Experiment and computation studies
of rod-deployment mechanisms. In: Proceedings of 16th international symposium on Ballistics,
San Francisco, pp 1–9
Schwer L (2010) An introduction to the Winfrith Concrete Model. Schwer Engineering &
Consulting Services
Sfer D, Carol I, Gettu R, Etse G (2002) Study of the behavior of concrete under triaxial
compression. J Eng Mech 128(2):156–163
Shi ZY, Tang WH, Zhao GM, Zhang RQ (2005) Similarity study of the penetration depth for the
concrete targets. J Ballistics 17(1):62–66 (in Chinese)
Shi Y, Li Z, Hao H (2010) A new method for progressive collapse analysis of RC frames under
blast loading. Eng Struct 32(6):1691–1703
Shi CC, Wang MY, Li J, Li MS (2014) A model of depth calculation for projectile penetration into
dry sand and comparison with experiments. Int J Impact Eng 73:112–122
Siefert A, Henkel FO (2014) Nonlinear analysis of commercial aircraft impact on a reactor
building-comparison between integral and decoupled crash simulation. Nucl Eng Des
269:130–135
Silling SA, Forrestal MJ (2007) Mass loss from abrasion on ogive-nose steel projectiles that
penetrate concrete targets. Int J Impact Eng 34(6):1814–1820
Sirijaroonchai K, EI-Tawil S, Parra-Montesinos (2010) Behavior of high performance fiber
reinforced cement composites under multi-axial compressive loading. Cem Concr Compos
32:62–72
Sjøl H, Teland JA, Kaldheim Ø (1998) Penetration into concrete with 12 mm projectiles.
FFI/NOTAT-98/04392 (in Norwegian)
Sjøl H, Teland JA, Kaldheim Ø (2000) Penetration into concrete-analysis of experiments with 12
mm projectiles. FFI/RAPPORT-2000/04414
Sliter GE (1980) Assessment of empirical concrete impact formulas. J Struct Div ASCE 106
(ST5):1023–1045
Sovják R, Vavřiník T, Máca P, Zatloukal J, Konvalinka P, Song Y (2013a) Experimental
investigation of ultra-high performance fiber reinforced concrete slabs subjected to deformable
projectile impact. Procedia Eng 65:120–125
Sovják R, Vogel F, Beckmann B (2013b) Triaxial compressive strength of ultra high performance
concrete. Acta Polytecknica 53(6):901–905
Sovják R, Vavřiník T, Zatloukal J, Máca P, Mičunek T, Frydrýn M (2015) Resistance of slim
UHPSFRC targets to projectile impact using in-service bullets. Int J Impact Eng 76:166–177
Steinberg DJ, Cochran SG, Guinan MW (1980) A constitutive model for metals applicable at
high-strain rate. J Appl Phys 51(3):1498–1504
Stronge WJ (1994) Planar impact of rough compliant bodies. Int J Impact Eng 15(4):435–450
Stronge WJ (2000) Impact mechanics. Cambridge University Press, Cambridge, pp 35–38
Stronge WJ (2013) Smooth dynamics of oblique impact with friction. Int J Impact Eng 51:36–49
Sugano T, Tsubota H, Kasai Y, Koshika N, Orui S, Von Riesemann WA, Bickel DC, Parks MB
(1993) Fullscale aircraft impact test for evaluation of impact force. Nucl Eng Des 140:373–385
Sugano T, Tsubota H, Kasai Y, Koshika N, Ohnuma H, Von Riesemann WA, Bickel DC,
Parks MB (1993a) Local damage to reinforced concrete structures caused by impact of aircraft
engine missiles Part 1. Test program, method and results. Nucl Eng Des. 140:387–405
Sugano T, Tsubota H, Kasai Y, Koshika N, Ohnuma H, Von Riesemann WA, Bickel DC,
Parks MB (1993b) Local damage to reinforced concrete structures caused by impact of aircraft
engine missiles Part 2. Evaluation of test results. Nucl Eng Des 140:407–423
Taerwe LR (1992) Influence of steel fibers on strain-softening of high-strength concrete. ACI
Mater J 88(6):54–60
References 573

Tai YS (2009) Flat ended projectile penetrating ultra-high strength concrete plate target. Theor
Appl Fract Mec 51(2):117–128
Tate A (1967) A theory for the deceleration of long rods after impact. J Mech Phys Solids 15
(6):387–399
Tate A (1969) Further results in the theory of long rod penetration. J Mech Phys Solids 17(3):141–
150
Tate A (1977) A possible explanation for the hydrodynamic transition in high speed impact. Int J
Mech Sci 19(2):121–123
Tate A (1979) A simple estimate of the minimum target obliquity required for the ricochet of a
high speed long rod projectile. J Phys D: Appl Phys 12:1825–1829
Tate A (1986) Long rod penetration models-Part II. Extensions to the hydrodynamic theory of
penetration. Int J Mech Sci 28(9):599–612
Taylor LM, Flanagan DP (1989) PRONTO3D a three-dimensional transient solid dynamics
program. SAND87-1912. Sandia National Laboratories, Albuquerque, NM
Taylor LM, Chen EP, Kuszmaul JS (1986) Microcrack-induced damage accumulation in brittle
rock under dynamic loading. J Comput Methods Appl Mech Eng 55:301–320
Teland JA, Sjøl H (2004) Penetration into concrete by truncated projectiles. Int J Impact Eng 30
(4):447–464
Teng TL, Chu YA, Chang FA, Shen BC, Cheng DS (2008) Development and validation of
numerical model of steel fiber reinforced concrete for high-velocity impact. Comp Mater Sci 42
(1):90–99
Tetsuo S, Atsushi K, Tomonori O, Hajime T, Masatoshi U, Nobutaka I (1997) Experimental and
numerical simulation of double-layered RC plates under impact loadings. Nucl Eng Des 176
(3):195–205
Thai DK, Kim SE (2014) Failure analysis of reinforced concrete walls under impact loading using
the finite element approach. Eng Fail Anal 45:252–277
Thai DK, Kim SE (2015) Failure analysis of UHPFRC panels subjected to aircraft engine model
impact. Eng Fail Anal 57:88–104
Thai DK, Kim SE (2016) Prediction of UHPFRC panels thickness subjected to aircraft engine
impact. Case Stud Struct Eng 5:38–53
TM 5-855-1 (1986) Fundamentals of protective design for conventional weapons. Technical
manual. US Department of Army, Washington, DC
Tsubota H, Kasai Y, Koshika N, Morikawa H, Uchida T, Ohno T, Kogure K (1993) Quantitative
studies in impact resistance of reinforced concrete panels with steel liners under impact
loading, Part 1: Scaled model impact tests. In: Proceedings of 12th international conference on
international association for structural mechanics in reactor technology (SMIRT), Stuttgart
Tu Z, Lu Y (2009) Evaluation of typical concrete material models used in hydrocodes for high
dynamic response simulations. Int J Impact Eng 36(1):132–146
Tu Z, Lu Y (2010) Modifications of RHT material model for improved numerical simulation of
dynamic response of concrete. Int J Impact Eng 37(10):1072–1082
Underwood JM (1995) Effectiveness of yaw-inducing deflection grids for defeating advanced
penetrating weapons. Technical report no. ESL-TR-92-61, Air Force Civil Engineering
Support Agency
Unosson M (2000) Numerical simulations of penetration and perforation of high performance
concrete with 75 mm steel projectile. Technical report no. FOA-R-00-01634-311-SE, Defence
Research Establishment Weapons and Protection Division, Tumba, Sweden
Unosson M, Nilsson L (2006) Projectile penetration and perforation of high performance concrete:
experimental results and macroscopic modeling. Int J Impact Eng 32(7):1068–1085
Vossoughi F, Ostertag CP, Monteiro PJM, Johoson GC (2007) Resistance of concrete protected by
fabric to projectile impact. Cem Concr Res 37(1):96–106
Vu XH, Daudeville L, Malecot Y (2011) Effect of coarse aggregate size and cement paste volume
on concrete behavior under high triaxial compression loading. Constr Build Mater 25:3941–
3949
574 References

Walker JD, Anderson CE Jr (1995) A time-dependent model for long-rod penetration. Int J Impact
Eng 16(1):19–48
Walter TA, Wolde-Tinsae AM (1984) Turbine missile perforation of reinforced concrete. J Struct
Eng ASCE 110(10):2439–2455
Wang W (1997) Stationary and propagative instabilities in metals-a computational point of view.
PhD, TU Delft, The Netherlands
Wang CZ, Guo ZH, Zhang XQ (1987) Experimental investigation of biaxial and triaxial
compressive concrete strength. ACI Mater J 84(2):92–100
Wang DR, Ge T, Zhou ZP, Wang MY (2006) Investigation of calculation method for
anti-penetration of reactive power steel fiber concrete (RPC). Explos Shock Waves 26
(004):367–372 (in Chinese)
Wang CH, Chen PY, Xu XC (2007a) Filtering of penetration deceleration data and determining of
penetration deceleration on the rigid-body. Explos Shock Waves 27(5):416–419 (in Chinese)
Wang ZL, Li YC, Shen RF, Wang JG (2007b) Numerical study on craters and penetration of
concrete slab by ogive-nose steel projectile. Comput Geotech 34(1):1–9
Wang C, Ma TB, Ning JG (2008) Experimental investigation of penetration performance of
shaped charge into concrete targets. Acta Mech Sin 24(3):345–349
Wang YN, Huang FL, Duan ZP (2010) Bending of projectile with small angle of attack during
high-speed penetration of concrete targets. Chin J Explosion and Shock Waves 30(6):598–606
(in Chinese)
Wang ZL, Wu J, Wang JG (2010) Experimental and numerical analysis on effect of fibre aspect
ratio on mechanical properties of SRFC. Constr Build Mater 24(4):559–565
Wang KH, Ning JG, Li ZK, Geng BG, Zhou G (2013) Ballistic trajectory of high-velocity
projectile obliquely penetrating concrete target. Chin J High Press Phys 27(4):561–566 (in
Chinese)
Wang KH, Geng BG, Chu Z, Zhou G, Li M, Gu RH, Ning JG (2014) Experimental studies on
structural response and mass loss of high-velocity projectiles penetrating into reinforced
concrete targets. Chin J High Pressure Phys 28(1):61–68 (in Chinese)
Wang C, Wang WJ, Ning JG (2015) Investigation on shaped charge penetrating into concrete
targets. Chin J Theor Appl Mech 47(4):672–686 (in Chinese)
Warren TL, Forrestal MJ (1998) Effects of strain hardening and strain-rate sensitivity on the
penetration of aluminum targets with spherical-nosed rods. Int J Solids Struct 35:3737–3753
Warren TL, Poormon KL (2001) Penetration of 6061-T6511 aluminum targets by ogive-nosed
VAR 4340 steel projectiles at oblique angles: experiments ans simulation. Int J Impact Eng. 25
(10):993–1022
Warren TL, Tabbara MR (1997) Spherical cavity-expansion forcing function in PRONTO 3D for
application to penetration problems. SAND97-1174, Sandia National Laboratories,
Albuquerque, NM
Warren TL, Fossum AF, Frew DJ (2004a) Penetration into low strength (23 MPa) concrete: target
characterization and simulations. Int J Impact Eng 30:477–503
Warren TL, Hanchak SJ, Poormon KL (2004b) Penetration of limestone targets by ogive-nosed
VAR 4340 steel projectiles at oblique angles: experiments and simulations. Int J Impact Eng 30
(10):1307–1331
Warren TL, Forrestal MJ, Randles PW (2014) Evaluation of large amplitude deceleration data
from projectile penetration into concrete targets. Exp Mech 54(2):241–253
Weerheijm J, Van Doormaal JCAM (2007) Tensile failure of concrete at high loading rates: new
test data on strength and fracture energy from instrumented spalling tests. Int J Impact Eng 34
(3):609–626
Wen HM, Xian YX (2015) A unified approach for concrete impact. Int J Impact Eng 77:84–96
Wen HM, Yang Y (2014) A note on the deep penetration of projectiles into concrete. Int J Impact
Eng 66:1–4
References 575

Wen HM, Yang Y, He T (2010) Effects of abrasion on the penetration of ogive-nosed projectiles
into concrete targets. Lat Am J Solids Stru 7:413–422
Werner S, Thienel K-C, Kustermann A (2013) Study of fractured surfaces of concrete caused by
projectile impact. Int J Impact Eng 52:23–27
Whiffen P (1943) UK road research laboratory note no. MOS/311
Willam KJ, Warnke EP (1975) Constitutive model for the triaxial behavior of concrete. In:
International association of bridges and structural engineers, seminar on concrete structures
subjected to triaxial stresses, Paper III-1, IABSE Proceedings 19, Bergamo, Italy
Wille K, Naaman AE (2012) Pullout behavior of high-strength steel fibers embedded in
ultra-high-performance concrete. ACI Mater J 109(4):479–488
Wille K, El-Tawil S, Naaman AE (2014) Properties of strain hardening ultra high performance
fiber reinforced concrete (UHP-FRC) under direct tensile loading. Cem Concr Comp 48:53–66
Winnicki A, Pearce CJ, Bićanić N (2001) Viscoplastic Hoffman consistency model for concrete.
Comput Struct 79(1):7–19
Wolf JP, Bucher KM, Skrikerud PE (1978) Response of equipment to aircraft impact. Nucl Eng
Des 47:169–193
Wu JH, Yu HY, Li Q, Jiang YD (2007) Experimental study on axial property for concrete with
constant surrounding pressure ratio. J Exp Mech 22(2):142–148 (in Chinese)
Wu C, Oehlers DJ, Rebentrost M, Leach J, Whittaker AS (2009) Blast testing of ultra-high
performance fibre and FRP-retrofitted concrete slabs. Eng Struct 31:2060–2069
Wu H, Fang Q, Zhang YD, Gong ZM (2012a) Semi-theoretical analyses of the concrete plate
perforated by a rigid projectile. Acta Mech Sin 28(6):1630–1643
Wu HJ, Qian F, Huang FL, Wang YN (2012b) Projectile nose mass abrasion of high-speed
penetration into concrete. Adv Mech Eng Article ID 296503:1–9
Wu HJ, Huang FL, Wang YN, Duan ZP, Pi AG (2012c) Experimental investigation on projectile
nose eroding effect of high-velocity penetration into concrete. Acta Armamentarii 33(1):48–55
(in Chinese)
Wu Y, Magallanes JM, Choi HJ, Crawford JE (2012d) Evolutionarily coupled finite-element
mesh-free formulation for modeling concrete behaviors under blast and impact loadings. J Eng
Mech ASCE 139(4):525–536
Wu H, Chen XW, He LL, Fang Q (2014a) Stability analyses of the mass abrasive projectile
high-speed penetrating into concrete target part I: engineering model for the mass loss and
nose-blunting of the ogive-nosed projectile. Acta Mech Sin 30(6):933–942
Wu H, Chen XW, Fang Q, He LL (2014b) Stability analyses of the mass abrasive projectile
high-speed penetrating into concrete target part II: structural stability analyses. Acta Mech Sin
30(6):943–955
Wu Y, Wang D, Wu CT (2014c) Three dimensional fragmentation simulation of concrete
structures with a nodally regularized meshfree method. Theor Appl Fract Mec 72:89–99
Wu H, Fang Q, Peng Y, Gong ZM, Kong XZ (2015a) Hard projectile perforation on the
monolithic and segmented RC panels with a rear steel liner. Int J Impact Eng 76:232–250
Wu H, Chen XW, Fang Q, Kong XZ, He LL (2015b) Stability analyses of the mass abrasive
projectile high-speed penetrating into concrete target Part III: Terminal ballistic trajectory
analyses. Acta Mech Sin 31(4):558–569
Wu H, Fang Q, Chen XW, Gong ZM, Liu JZ (2015c) Projectile penetration of ultra-high
performance cement based composites at 510m/s to 1320m/s. Constr Build Mater 74(15):188–
200
Wu H, Fang Q, Gong J, Liu JZ, Zhang JH, Gong ZM (2015d) Projectile impact resistance of
corundum aggregated UHP-SFRC. Int J Impact Eng 84:38–53
Wu H, Fang Q, Gong ZM, Peng Y (2015e) Hard projectile impact on layered SFRHSC composite
target. Int J Impact Eng 84:88–95
Xie J, Elwi AE, MacGregor JG (1995) Mechanical properties of three high-strength concretes
containing silica fume. ACI Mater J 92:135–143
576 References

Xie HP, Dong YL, Li SP (1996) Study of a constitutive model of elasto-plastic damage of concrete
in axial compression test under different pressures. J China Coal Soc 21(3):265–270 (in
Chinese)
Xiong YB (2009b) Research on constitutive parameters of concrete based on the
Johnson-Holmquist concrete model. Dissertation, Northwest Institute Nuclear of
Technology, Xi’an (in Chinese)
Xiong YB, Chen JJ, Hu YL (2009a) Preliminary identification of sensitive parameters in
Johnson-Holmquist concrete constitutive model. J China Ordnance 30(2 suppl):145–148 (in
Chinese)
Xiong YB, Hu YL, Xu J, Chen JJ (2010) Determining failure surface parameters of the
Johnson-Holmquist concrete constitutive model. J China Ordnance 31(6):746–751 (in Chinese)
Xu H (2013) A new computational constitutive model for concrete subjected to dynamic loadings.
Dissertation, University of Science and Technology of China, Hefei (in Chinese)
Xu JB, Lin JD (2002) Penetration of steel bar projectiles into concrete targets. Explos Shock
Waves 22(2):174–178 (in Chinese)
Xu K, Lu Y (2006) Numerical simulation study of spallation in reinforced concrete plates
subjected to blast loading. Comput Struct 84(5):431–438
Xu H, Wen HM (2013) Semi-empirical equations for the dynamic strength enhancement of
concrete-like materials. Int J Impact Eng 60:76–81
Xu Z, Hao H, Li HN (2012) Mesoscale modelling of dynamic tensile behaviour of fibre reinforced
concrete with spiral fibres. Cement Concrete Res 42(11):1475–1493
Yan SH (2001) Study of penetration theory and experiments on high-strength fiber reinforced
concrete. Dissertation, PLA University of Science and Technology, Nanjing (in Chinese)
Yan SH, Qian QH, Zhou ZS, Lu YS, Yin FL (2000) Experimental study of equation of state for
high-strength concrete and high-strength fiber concrete. J PLA Univ Sci Technol 1(6):49–53
(in Chinese)
Yan DM, Lin G, Xu P (2007) Dynamic strength and deformation of concrete in triaxial stress
states. Eng Mech 24(3):58–64 (in Chinese)
Yankelevsky DZ (1997) Local response of concrete slabs to low velocity missile impact. Int J
Impact Eng 19(4):331–343
Youch DF (2006) Efficient calculation of earth penetrating projectile trajectories. Naval
Postgraduate School. Monterey, California. Master thesis
Zhang WH (2010) Investigation of microstructure formation mechanism and dynamic mechanical
behavior of UHPCC. Doctoral thesis, Southeast University, Nanjing, China (in Chinese)
Zhang MH, Shim VPW, Lu G, Chew CW (2005) Resistance of high-strength concrete to projectile
impact. Int J Impact Eng 31(7):825–841
Zhang MH, Sharif MSH, Lu G (2007a) Impact resistance of high-strength fiber-reinforced
concrete. Mag Concrete Res 59(3):199–210
Zhang WD, Chen LJ, Xiong JJ, Ma YC (2007b) Ultra-high g deceleration-time measurement for
the penetration into steel target. Int J Impact Eng 34(3):436–447
Zhang M, Wu HJ, Li QM, Huang FL (2009) Further investigation on the dynamic compressive
strength enhancement of concrete-like materials based on split Hopkinson pressure bar tests
part I: experiments. Int J Impact Eng 36(12):1327–1334
Zhao J, Chen XW, Jin FN, Xu Y (2010a) Depth of penetration of high-speed penetrator with
including the effect of mass abrasion. Int J Impact Eng 37(9):971–979
Zhao J, Chen XW, Jin FN, Xu Y (2010b) Study on the penetration depth of penetrator with
including the effect of mass loss. Chin J Theor Appl Mech 42(2):212–218 (in Chinese)
Zhao J, Chen XW, Jin FN, Xu Y (2011) Cartridge design of a high-speed projectile considering
mass abrasion. Chin J Explosion Shock Waves 31(5):481–489 (in Chinese)
Zhao J, Chen XW, Jin FN, Xu Y (2012) Analysis on the bending of a projectile induced by
asymmetrical mass abrasion. Int J Impact Eng 39(1):16–27
References 577

Zhou N, Ren HQ, Shen ZW (2006a) Experimental on the projectile penetration concrete targets
and reinforced concrete targets. Chin J Uni Sci-Tech 36(10):1021–1027 (in Chinese)
Zhou X, Hao H, Kuznetsov VA, Waschl J (2006b) Numerical calculation of concrete slab response
to blast loading. Transac Tianjin Univ 12(Suppl.):94–99
Zukas JA (1990) Survey of computer codes for impact simulations. In: Zukas JA (ed) High
velocity impact dynamics. Wiley, New York, pp 693–714