See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/276460912

Dynamic Analysis of Shell Structures with Application to Blast Resistant

Doors

Article in Shock and Vibration · January 2003

DOI: 10.1155/2003/357969

CITATIONS READS

16 46

3 authors, including:

C. G. Koh K.K. Ang

National University of Singapore National University of Singapore
193 PUBLICATIONS 2,553 CITATIONS 88 PUBLICATIONS 1,055 CITATIONS

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Extreme Wave Impact on Floating Structures: A Novel Particle Method with Experimental Validation View project

Engineering Analysis and Application for Mitigating LNG Sloshing on Ship View project

All content following this page was uploaded by K.K. Ang on 16 March 2016.
provided by ScholarBank@NUS
The user has requested enhancement of the downloaded file.
View metadata, citation and similar papers at core.ac.uk brought to you by CORE
Shock and Vibration 10 (2003) 269–279 269
IOS Press

Dynamic analysis of shell structures with

application to blast resistant doors
C.G. Koh∗ , K.K. Ang and P.F. Chan
Department of Civil Engineering, National University of Singapore, 10 Kent Ridge Crescent, Singapore 119260

Received 2 July 2001

Revised 24 September 2002
Accepted 2 May 2003

Abstract. This paper concerns the dynamic analysis of shell structures, with emphasis on application to steel and steel-concrete
composite blast resistant doors. In view of the short duration and impulsive nature of the blast loading, an explicit integration
method is adopted. This approach avoids time-consuming computations of structural stiffness matrix and solving of simultaneous
nonlinear equations. Single-point quadrature shell elements are used, with numerical control to suppress spurious hourglass
modes. Composite shells are handled by an appropriate integration rule across the thickness. Both material and geometric
nonlinearities are accounted for in the formulation. Contact and gap problems are considered using bilinear spring elements
in the finite element analysis. Numerical examples are presented for some benchmark problems and application study to blast
resistant doors. Good correlation is generally obtained between the numerical results based on the software developed and the
results obtained by other means including field blast tests.

Keywords: Dynamics, vibration, shell, blast doors, composite, steel, concrete

1. Introduction fibre-reinforced concrete panels subjected to blast load-

ing. Krauthammer carried out numerical studies on
Blast resistant doors are commonly used in defence the blast response of structural concrete and structural
shelters, ammunition storage dumps and blast-resistant steel connections [7]. The present study focuses on the
structures alike. These doors have to be designed to dynamic analysis of steel blast doors and composite
withstand blast loading and shock waves, so as to pro- blast doors in the form of steel-concrete-steel sandwich
tect occupants and objects inside, and in the case of panels.
ammunition storage, to prevent chain reaction of ex- Geometrically, blast resistant doors may be flat or
plosions from one compartment to another. Inadequate curved, though the doors are usually rectangular in
design of the blast resistant door would undermine the front elevation. The structural resistance of blast doors
operational performance and survivability of the struc- is primarily provided by both membrane and bending
ture and users. On the other hand, an overly conser- actions. Correct modelling of the nonlinear behaviour
vative design would result in incurring unnecessarily of blast resistant doors under severe loading is there-
high costs. fore essential in their safe and cost-effective design.
Recent studies on blast resistant structures mainly Dynamic analysis of shell structures, among all struc-
dealt with reinforced concrete structures, see [1–5] for tural forms, is perhaps the most challenging both math-
example. Lok and Xiao [6] extended the study to steel- ematically and physically. Mathematically, it is rel-
atively difficult to formulate and obtain solutions for
shell structures, particularly when nonlinear dynamic
∗ Corresponding author. Tel.: +65 68742163; Fax: +65 67791635; response is involved. Physically, the response of shell
E-mail: [email protected]. structures involves the interaction between the mem-

ISSN 1070-9622/03/$8.00  2003 – IOS Press. All rights reserved

270 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors

brane mode and bending mode which, in general, are locities are transformed to the co-rotational coordinate
of different orders of magnitude in stiffness effects. system. The velocity strains for membrane stretch-
Analytical approaches do not apply to nonlinear ing, bending and shear are computed. This involves
problems in general, and numerical approaches have to straightforward strain-displacement matrices evaluated
be used. With rapid advances in computer hardware, at the origin of the co-rotational coordinate system. As
the finite element method has proven to be a versa- the element is under-integrated (in plane), a stabiliza-
tile and powerful tool for nonlinear dynamic analysis tion approach called perturbation hourglass control as
of shell structures. Many shell theories and solution proposed in [12] is adopted to suppress the spurious
methods have been developed in the last few decades. singular modes.
The finite element formulation of shell elements can Since blast resistant doors may be composite in ma-
largely be categorised into two approaches: (a) direct terial, it is necessary to split the element in several
approach based on a selected shell theory and (b) ap- plies that are not necessarily made of the same ma-
proach based on degenerated continuum element first terial. Different plies may have different thicknesses.
proposed by Ahmad et al. [8]. A good literature review In this regard, the approach for the explicit dynamic
of shell elements may be found in [9]. analysis of laminated composite shells [11] is appro-
As this study involves short-duration loading, ex- priate. It is assumed herein that the bonding between
plicit dynamic analysis, which avoids the time consum- plies remains intact with no delamination even though
ing simultaneous equation solver, is preferred. For ex- deformation may be large. The focus of the analysis
plicit analysis, 4-node shell element derived by the de- is the global behaviour (e.g. deflection) of the struc-
generated continuum approach is an attractive choice. ture, rather than the local behaviour (e.g. strain vari-
In contrast, higher-order elements are sensitive to mesh ation through the thickness). Therefore, the assump-
distortion and less stable in an explicit dynamic anal- tion that the plane section across the thickness remains
ysis [10]. Reduced integration with single quadra- plane is made. To account for possible large thickness
ture point is preferred for reason of computational ef- strain, the thickness is updated using log-strain rather
ficiency, but numerical means have to be introduced to than engineering strain. Geometric nonlinear effects
suppress hourglass modes. due to large deformations are included by means of the
updated Lagrangian formulation.
In blast resistant door designs, steel and concrete ma-
In terms of dynamic analysis, there are two possi-
terials are commonly used. A combination of steel and
ble approaches: the implicit and explicit integration
concrete would take advantage of high strength of steel
methods. In the implicit methods, element stiffness
and added inertia of concrete in-fill. The sandwich-like
matrices have to be computed, assembled and stored in
structure would require composite elements in the anal-
the global stiffness matrix. The equations of motion
ysis. In this regard, the explicit dynamic approach de-
are converted into a set of algebraic equations by some
veloped by Koh et al. [11] for 4-node composite shells
numerical integration schemes such as the Newmark-
is adopted herein for the dynamic analysis of blast re-
Wilson family [13] and then solved by a simultaneous
sistant doors. The constitutive models adopted for steel
equation solver. Since the system is nonlinear, an iter-
and concrete are, respectively, Von-Mises yield crite-
ative scheme such as the Newton-Raphson method is
rion with hardening and a scalar-damage model (es- required in the equation solving until certain conver-
tablished in the framework of continuum mechanics). gence criterion is met. As the system is nonlinear and
Bilinear spring elements are used to handle contact and coupled, the equation solving process normally forms
gap between the door and door frame. Nonlinear effects the bulk of the computational effort. Thus, though the
due to geometry updates are accounted for by means of implicit methods with appropriate parameters can be
the updated Lagrangian formulation. unconditionally stable, the cost per time step is often
high.
Explicit methods, on the other hand, do not require
2. Composite shell elements and dynamic analysis the formation of stiffness matrices. The equations of
motion in the global co-ordinate system may be written
In this study, a bilinear four-node quadrilateral shell as
element with single quadrature point [12] is adopted
in the finite element modelling of blast resistant doors. M Ü + C U̇ (t) + P (U , t) = F (t) (1)
In essence, a co-rotational coordinate system is estab- where U , P , F , M and C are the displacement vec-
lished for each element. The nodal coordinates and ve- tor, internal force and external force vectors, mass and
 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors 271

damping matrices, respectively. Note that no stiffness σ

matrix is required. With the assumption of lumped
mass matrix and mass-proportional damping matrix, To account for gap
the equations of motion are uncoupled. Expressing the
acceleration and velocity in terms of displacement, the
displacement vector can be solved for the current step
without a simultaneous equation solver. The drawback ε
is that explicit methods are conditionally stable; thus
a small time step (smaller than a critical value) has
to be used. This, however, is not a hindrance in this
study because the time step used for impulsive (blast)
loading is normally small in order to accurately capture Fig. 1. Stress-strain relationship of contact/gap element.
the loading function. In such applications, the over-
all computational cost for explicit methods is generally As the explicit scheme is conditionally stable, the time
less than implicit methods. step has to be less than a critical value to ensure numer-
In the present study, a simple yet effective explicit ical stability. The critical time step is computed based
method known as the central difference method is cho- on a safe estimate as follows [14]
sen [13]. Let subscripts n and n − 1 denote the current
∆tcritical = (8)
and previous time steps, respectively. The velocity for
 0.5
the i-th degree of freedom is replaced by the following ρ(1 − v 2 )/E
central difference approximation: Ls
2 + 0.83(1 + 1.5L2s /h2 )(1 − v)
Ui,n+1 − Ui,n where ρ is the density, E Young’s modulus, v Possion’s
U̇i,n+0.5 = (2)
∆tn ratio, h element thickness, and L s the shortest distance
where ∆tn denotes time step used at the n-th step between adjacent nodes.
(current step) and subscript n + 0.5 denotes the specific The stresses are computed at all the integration points
time midway between the n-th step and (n + 1)-th step. through the thickness. In general, composite layer
Similarly, thicknesses are arbitrary and need not necessarily be
Ui,n − Ui,n−1 symmetrical with respect to the mid-surface of the ele-
U̇i,n−0.5 = (3) ment. Each layer can be further divided into sub-layers
∆tn−1
(of the same material), depending on the relative thick-
Note that the formulation allows variable time steps. ness and desired numerical accuracy. In this instance,
For certain situations where long-term response is the commonly used rule of Gauss quadrature is not
needed, the computational time would be reduced sub- convenient for integration through the thickness. Fur-
stantially with the application of a variable time step thermore, this rule is not necessarily the most accurate
scheme. The velocity and acceleration at the current strategy for integration of nonlinear stress through the
step are given by: thickness [15] and it is thus justifiable to use a simpler
1 integration rule. In the present study, the mid-point
U̇i,n = (U̇i,n+0.5 + U̇i,n−0.5 ) (4)
2 rule is used, i.e. integration points are the midpoints of
sub-layers and quadrature weights are proportional to
1
Üi,n = (U̇i,n+0.5 − U̇i,n−0.5 ) (5) the sub-layer thicknesses, though not equal in general.
∆tn For ease of reference, this is termed as the composite
where integration rule. This is different from the centroidal
1 method as presented in [15], which divides the element
∆tn = (∆tn−1 + ∆tn ) (6)
2 thickness into sub-layers of equal thickness.
Making use of the above equations, it can be readily
shown that Eq. (1) yields

 3. Material models
2Mi − Ci ∆tn
U̇i,n+0.5 = U̇i,n−0.5
2Mi + Ci ∆tn In the design of blast resistant doors, steel and con-

 (7)
2∆tn crete materials are commonly used for reason of cost-
+ (Fi − Pi )
2Mi + Ci ∆tn effectiveness though more advanced materials are avail-
272 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors

Dynamic Finite element Material

analysis modelling models

Steel:
Von-Mises
4-node shell model
Explicit element with
approach single quadrature &
hourglass control Concrete:
Scalar
damage
model
Updated Composite
Lagrangian layers
Contact/gap
element

Software BASS

Fig. 2. Main features of software development.

able. The respective constitutive models are Von-Mises f (σ) = σx2 + σy2 − σx σy + 3σxy
2
(10)
model and a scalar damage model. The hybrid use of
For stress update, the key issue is often the integra-
steel and concrete would take advantage of the high
tion of the flow rule in each time step. A commonly
strength of steel and large inertia of concrete in-fill to
adopted approach is the backward Euler method which
enhance the performance of blast resistant doors. The is stable and usually efficient [16]. This involves itera-
following two types of blast resistant doors are consid- tive solutions to achieve convergence by means of, say,
ered in this study. the Newton-Raphson scheme which is adopted in this
(A) Steel-air-steel (SAS) door – The door comprises study. Non-iterative schemes are also available if the
two “skin” plates, stiffened by side plates along advantage of high concurrency for the explicit dynamic
the four edges and some stiffener beams (e.g. analysis is to be exploited towards vectorization and
I or C-channel cross section) at intermediate parallel computing [11].
locations. All the plates and beams are made of
steel. 3.2. Scalar damage model for concrete
(B) Steel-concrete-steel (SCS) door – The door is
similar to the SAS door, except that the void is The concrete model adopted here is a scalar dam-
filled with concrete. age model developed in the framework of the contin-
uum damage mechanics [17]. It is relatively simple
in formulation and easy to implement in the finite ele-
3.1. Elastoplastic model with hardening for steel ment analysis. More importantly, the strain rate effect
on concrete behaviour is accounted for. The compu-
Steel usually exhibits a work-hardening behaviour tational model is partially verified by an experimental
when stressed into the elastoplastic zone, and the investigation program involving impact tests of con-
isotropic Von Mises material model is appropriate. The crete specimens. This model is appropriate for steel-
yield function is generally in the form of f y (σ, κ, . . .), concrete composite doors where concrete in-fill serves
where κ is the hardening parameter which depends on mainly to increase the inertia (rather than strength) and
the accumulated plastic strain in some manner. The severe concrete damage is not expected.
Von Mises criterion for the assumed plane stress state In general, the mechanical properties of concrete are
can be written as enhanced with increasing strain rate. The domain of
F (σ, κ) = f (σ) − σY2 (κ) (9) stress and strain are divided into two parts, one for the
low (quasi-static) to intermediate rate (ε̇
30s −1 , and
where σY is the yield stress in pure shear, and the yield the other, from intermediate to high rates. The CEB-
function (f ) is given by 1988 recommendations [18] are adopted with regards to
 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors 273

start

input data

assemble mass matrix

compute mass-proportional damping matrix

compute external loads

material
compute internal forces
models

compute hourglass forces

t = t + ∆t

update effective nodal forces

compute hourglass forces

compute vel/disp/acc.

output results

save data for future analysis

end

Fig. 3. Computational procedure of dynamic analysis.

the rate-dependent concrete material properties under this concrete damage model, concrete is assumed to
compressive and tensile actions. remain isotropic up to failure. The positive strains de-
The rate dependent damage model is formulated us- termine the growth of damage that is mainly associated
ing the concept of internal variable to represent the ex- with opening of micro-cracks. The damage consists of
tent of concrete damage. The stress-strain relationship both tension and compression components, defined as
of the concrete model is given as follows
D = αt Dt + αc Dc (12)
σij = (1 − D)Eijkl εkl (11)
where subscripts t and c denote tension and compres-
where σij and εij are the components of the stress sion, respectively, and α t and αc are non-dimensional
and strain tensors, respectively, E ijkl denote the initial functions of the principal strains. More details on the
moduli, and D is a scalar damage parameter that ranges damage model adopted can be found in [17] which
from 0 (virgin material) to 1 (asymptotic failure). In covers experimental verification of the model.
274 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors

P Symmetry

2θ
Symmetry

(a) Geometry of the spherical cap (b) Finite element mesh

Fig. 4. Geometry of the spherical shell and its finite element mesh.

0 0.2 0.4 0.6 0.8 1

0.00

-0.02
Displacement ( in )

-0.04

-0.06

-0.08

-0.10 Liu (1985)

present solution
-0.12
Time ( ms )

Fig. 5. Vertical displacement at centre of spherical cap – linearly elastic material.

3.3. Contact/Gap modelling 4. Numerical study and discussion

There are two types of blast resistant doors in prac-

tice, namely the air-tight door and blast-tight door. In Instead of using a general purpose software, a spe-
the former case, rubber gaskets are used between the cific PC-based software is developed as a research tool
door and door frames for airtightness, giving rise to a to implement the explicit dynamic analysis of blast re-
structural gap between the door and the door frame. sistant doors with the main features as illustrated in
In the latter case, the door is in contact with the door Fig. 2. In particular, the composite shell element as
frame (no gap) when closed. In either case, the con- mentioned earlier is incorporated. The flowchart of the
tact and gap behaviour is modelled by using nonlin-
software called BASS (Blast Analysis of Shell Struc-
ear spring elements in the finite element analysis. The
stress-strain relationship of a typical contact/gap spring tures) is illustrated in Fig. 3. The software has the op-
is illustrated in Fig. 1. The tension part represents the tion of using Gauss integration rule for non-composite
situation when the door and the frame are not in con- shells or the composite integration rule otherwise. The
tact, causing no stress in the element. When the door following numerical examples are presented to verify
comes into contact with the door frame, the element the formulation and implementation (more numerical
becomes very stiff under compressive action. examples can be found in [19]).
 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors 275

0 0.2 0.4 0.6 0.8 1

0.04

0.00
Displacement (in)

-0.04

-0.08

-0.12
Liu (1985)
present solution
-0.16
Time (ms)

Fig. 6. Vertical displacement at centre of spherical cap – Von Mises material.

(a) (b)
Fig. 7. Single-panel blast door: (a) schematic diagram; (b) finite element mesh.

Example 1: Cantilever beam The present solution agrees well with the published re-
Consider a cantilever beam with rectangular cross sults [12] as presented in Table 1. It is noted that the
section and parameters as follows [12]: length = present solution is closer to the beam element solution
0.254 m, width = thickness = 2.54 cm, Young’s modu- as compared to other solutions as summarised in [12].
lus = 82.8 MPa, density = 12.87 kg/m 3 and Possion’s
ratio = 0.2. The beam is subjected to a suddenly ap- Example 2: Spherical cap
plied pressure loading is analysed using the software Clamped all around the edge, a spherical cap is sub-
BASS. The applied pressure loading is 69 Pa (0.01 jected to a suddenly applied pressure of 4.14 MPa (600
psi), and the material is assumed to be linearly elastic. psi). The geometric and material properties are as fol-
276 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors

Table 1
Dynamic response of cantilever beam
Element type Number of elements Max. deflection (mm) Fundamental period (ms)
Euler beam element 5 0.646 5.812
Triangular plate elements 20 0.612 5.662
Quadrilateral plate elements 5 0.623 5.68
Present solution (with software BASS) 5 0.65 5.835

Table 2
Comparison of maximum displacements
Point Maximum displacement Maximum Displacement
on Outer Skin Plate (mm) on Inner Skin Plate (mm)
BASS LUSAS BASS LUSAS
A 12.9 8.2 12.8 8
B 23.6 24.8 22.4 17.6
C 12.1 6.9 12.1 6.1
D 22.5 24.6 6.4 4
E 1.1 1.1 1 1

lows [20]: internal radius = 0.5657 m, thickness = 1000

1.041 cm, semi-angle (θ) = 26.27 ◦, Young’s modulus
800
Pressure (kPa)

= 72.4 GPa, density = 2778 kg/m 3 , Poisson’s ratio =

0.3, yield stress = 169 MPa, and hardening parameter
600
= 0.
By virtue of symmetry, only a quarter of the sphere is 400
considered and is discretised into12 elements (Fig. 4).
Both linear and nonlinear analyses are carried out. The 200
problem was analysed by Liu [20] using 8-node de-
generated thick-shell elements and an implicit dynamic 0
analysis approach. The same material model is adopted 0 2 4 6 8 10
by Liu [20] and in the present study, and no rate effect Time (ms)
is considered in this example. Figure 5 shows the ver-
tical displacement at the centre of the sphere for lin- Fig. 8. Idealized blast pressure for Example 3.
early elastic material, in comparison with Liu’s result.
For the nonlinear case, the Von Mises yield criterion as Length L = 2.2 m
described earlier is considered but without hardening. Width b = 1.3 m
The corresponding dynamic response is presented in Young’s modulus E = 200 GPa
Density ρ = 7800 kg/m3
Fig. 6 and, due to softening effect (yielding), is gener- Poisson’s ratio v = 0.2
ally larger than the response for linearly elastic mate- Yielding stress σY = 374#MPa
rial as shown in Fig. 5. Note that in [20], the results Ultimate stress σult = 440 MPa
were computed using 6 layers (accurate up to 5th order Hardening parameter κ = 1.65 × 10−4
polynomial in the thickness direction), whereas in the
present study, 5 Gauss points (accurate up to 9th order with the blast pressure) and inner skin plate are mod-
polynomial) are used. Therefore, the present solution elled by the shell elements. Stiffeners are modelled as
should be more accurate than the results given in [20]. beam elements in this study as they deform predomi-
nantly in flexural bending, though shell elements may
Example 3: Numerical study of blast resistant door also be used. The door is surrounded by a rigid door
In this example, a single-panel rectangular SAS door frame around all four edges, with two hinges on the left
subjected to blast loading is analyzed. The door consid- edge. There is an initial gap of 1 mm between door and
ered is made of two 9-mm mild steel skin plates, stiff- door frame. There is no initial gap between latch and
ened by four side plates along the edges and three C- door frame. Figure 7(b) shows the finite element mesh
channels at intermediate locations, as shown schemati- used in the numerical study. The parameters used in
cally in Fig. 7(a). The outer skin plate (in direct contact this example are listed below:
 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors 277

0.030

0.025
Displacement (m) B
0.020 D
0.015
A
0.010
C
0.005

0.000
E
-0.005
0 2 4 6 8
Time (ms)

Fig. 9. Displacements at selected points on the outer skin plate.

0.025
B
0.020
Displacement (m)

0.015 A
0.010 C
0.005 D
0.000 E
-0.005

-0.010
0 2 4 6 8
Time (ms)

Fig. 10. Displacements at selected points on the inner skin plate.

An idealized pressure loading as shown in Fig. 8 is 1000

Pressure (kPa)

used. The displacements at critical points on the outer 800

and inner skin plates are presented in Figs 9 and 10, 600
respectively. The displacements at points A, C and E 400
at the stiffeners are smaller than the other two points 200
(B and D) away from the stiffeners, as expected. It is
0
also observed that the outer skin plate undergoes larger
0 2 4 6
deformation than the inner skin plate.
Numerical results for up to 8 ms are obtained by Time (ms)
the software BASS. The maximum displacements of Fig. 11. Idealized blast pressure used for Example 4.
the skin plates are presented in Table 2. These results
are found to be generally in good agreement with the
results obtained by an implicit finite element analysis
software called LUSAS [21], particularly for the two the explicit approach is found to be about twenty times
most critical points B and D where the deformations are more efficient than the implicit approach to achieve
large. In terms of computational time, BASS based on roughly the same accuracy.
278 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors

Table 3
Details of blast resistant doors used in field tests
Door specimens Plate thickness Clear distance between outer Stiffeners
(mm) and inner skin plates (mm) (C-channels)
SAS1 and SCS1 6 50 50 × 25 × 3
SAS2 and SCS2 6 100 100 × 50 × 5
SAS3 and SCS3 9 100 100 × 50 × 5
SAS4 and SCS4 9 150 150 × 50 × 5

Table 4
Maximum and permanent displacements at centre of blast door
Door specimen Maximum displacement (mm) Permanent displacement (mm)
Field test BASS Field test BASS
SAS1 >85 180 N.A. 150
SCS1 42 36 N.A. 18
SAS2 >85 71 N.A. 51
SCS2 35 11 26 2
SAS3 N.A. 45 33 28
SCS3 10 6 1 1
SAS4 38 24 N.A. 14
SCS4 15 4 0 0
Note: N.A. = Not available due to instrumentation problem.

Example 4: Comparison with field test results for C-channel stiffeners Side plates
blast resistant doors
Field tests of blast resistant doors of both SAS and
SCS types were carried out, each under 100 kg bare
charge of TNT with a stand-off distance of 5 m. The 1200 mm
peak pressure loading is computed by a computer pro-
gram called CONWEP [22] and its shape is idealized
as linear, as shown in Fig. 11. The dimensions of all
test specimens are 2.2 m × 1.2 m, as shown in Fig. 12.
Each door specimen comprises two skin plates, three 4 x 550 mm Skin plates
C-channel stiffeners, and side plates all around the four (top & bottom)
edges. A total of eight specimens were blast tested
Fig. 12. Schematic diagram of blast door specimens for field tests
in the field: four SAS doors and four SCS doors. In (plan view).
the case of SCS doors, plain concrete of grade 30 (de-
signed to have compressive strength of 30 MPa at 28
days conforming to British Standard BS 5328 [23]) was age of instruments during the blast test. Furthermore,
used as in-fill. Table 3 shows the geometrical dimen- it should be noted that the specified blast time history
sions of the specimens. The specimens were vertically (peak value generated by CONWEP and idealized lin-
held by clamping their two shorter edges on a specially early) used in the numerical simulation studies differs
constructed concrete supporting frame. Displacement from the real blast in the field tests. In this light, the
gauges were installed behind the door specimen to mea- dynamic analysis by means of software BASS gives
sure the maximum and permanent deflection of the
reasonably good correlation with the blast test results,
specimen. The field test results are presented in Table 4
as shown in Table 4. The other possible source of
and compared with the numerical results obtained by
BASS. discrepancy is the modeling of actual boundary con-
Response measurement in the field is generally a dif- ditions, which play an important role in the displace-
ficult task and particularly so for blast tests where the ment response of blast resistant doors. Both the nu-
peak response occurs in a very short time. Some field merical and field test results confirm the effectiveness
test results may be inaccurate due to measurement er- of the concrete in-fill in that the displacement response
rors or even unavailable. In fact, some results for per- of SCS blast doors is considerably smaller than that of
manent displacements were not available due to dam- SAS blast doors.
 C.G. Koh et al. / Dynamic analysis of shell structures with application to blast resistant doors 279

5. Conclusions [4] R.K. Otani and T. Krauthammer, Assessment of reinforcement

details for blast containment structures, ACI Structural Journal
94 (1997), 124–132.
This paper illustrates the use of explicit dynamic [5] T. Krauthammer and R.K. Otani, Mesh, gravity and load ef-
analysis for the analysis and design of actual blast resis- fects on finite element simulations of blast loaded reinforced
tant doors. The main characteristic of the quadrilateral concrete structures, Computers and Structures 63 (1997),
shell element used is that only single quadrature point 1113–1120.
[6] T.S. Lok and J.R. Xiao, Steel-fibre-reinforced concrete panels
is needed in the plane of each element. This results in a exposed to air blast loading, Proceedings of the Institution of
simplified formulation and low computational cost per Civil Engineers – Structures and Buildings 134 (1999), 319–
element-time step. The shell element allows the exten- 331.
sion to treat composite materials using the composite [7] T. Krauthammer, Blast-resistant structural concrete and steel
connections, International Journal of Impact Engineering 22
integration rule. (1999), 887–910.
In terms of nonlinearity, three aspects have been in- [8] S. Ahmad, B.M. Irons and O.C. Zienkiewicz, Analysis of thick
cluded. (A) For material nonlinearity, Von Mises yield and thin shell structures by curved finite elements, Int. J. Num.
Meth. Engrg. 2 (1970), 419–451.
criterion with isotropic hardening is adopted for steel,
[9] H. Stolarski, T. Belytschko and S.-H. Lee, A review of shell
whereas a scalar damage model developed for concrete finite and corotational theories, Computational Mechanics Ad-
is used. (B) Geometric nonlinearity due to possible vances 2 (1995), 125–212.
large displacement motion is accounted for by means of [10] Y.G. Zhu and T. Zacharia, A new one-point quadrature, quadri-
lateral shell element with drilling degrees of freedom, Com-
Lagrangian formulation. (C) The contact/gap behavior
puter Methods in Applied Mechanics and Engineering 136
is modeled using bilinear spring elements. (1996), 165–203.
The computational procedure has been coded lead- [11] C.G. Koh, D.R.J. Owen and D. Peric, Explicit dynamic analy-
ing to the software called BASS. The numerical study sis of elasto-plastic laminated composite shells: implementa-
tion of non-iterative stress update schemes for the HOFFMAN
presented includes comparison with published solu- yield criterion, Comp. Mech. 16 (1995), 307–314.
tions for some benchmark problems of beams and [12] T. Belytschko, J.L. Lin and C.S. Tsay, Explicit algorithms for
shells, with another finite element (implicit) software the nonlinear dynamics of shells, Comp. Meth. Appl. Mech.
and with field measurements for actual blast resistant Engrg. 42 (1984), 225–251.
[13] K.-J. Bathe, Finite Element Procedures, Prentice-Hall, New
doors. In general, the reasonably good correlation vali- Jersey, USA, 1996.
dates the numerical procedure and software implemen- [14] T.Y. Tsui and P. Tong, Stability of transient solution of mod-
tation. The software developed is useful in the para- erately thick plate by finite difference method, AIAA Journal
metric studies to optimize the design of blast resistant 9 (1971), 2062–2063.
[15] C.J. Burgoyne and M.A. Crisfield, Numerical integration strat-
doors. egy for plates and shells, Int. J. Num. Meth. Engrg. 29 (1990),
105–121.
[16] J.C. Simo and R.L. Taylor, A return mapping algorithm for
Acknowledgment plane stress elastoplasticity, Int. J. Num. Meth. Engng. 31
(1986), 1069–1086.
[17] C.G. Koh, Z.J. Liu and S.T. Quek, Numerical and experi-
The research grant and technical assistance (particu- mental studies of concrete damage under impact, Magazine of
larly on design data and field tests) provided by the Min- Concrete Research 53 (2001), 417–427.
[18] Comité Euro-International du Béton (CEB), Concrete struc-
istry of Defence, Singapore, are gratefully acknowl- tures under impact and impulsive loading, Synthesis Report.
edged. Bulletin d’Information No. 187, Lausanne, 1998.
[19] P.F. Chan, Dynamic Analysis of Shell Structures with Appli-
cation to Blast Doors. M.Eng. Thesis, National University of
Singapore, Singapore, 1999.
References [20] G.Q. Liu, Nonlinear and Transient Finite Element Analysis of
General Reinforced Concrete Plates and Shells, Ph.D. Thesis,
[1] S.C. Woodson and S.A. Kiger, Stirrup requirements for blast- University College of Swansea, UK, 1985.
resistant slabs, Journal of Structural Engineering, ASCE 114 [21] LUSAS Finite Element Software, Version 12, FEA Ltd. Sur-
(1988), 2057–2069. rey, UK, 1998.
[2] F.B.A. Beshara and K.S. Virdi, Time integration procedures for [22] Fundamentals of Protective Design for Conventional
finite element analysis of blast-resistant reinforced concrete Weapons, (Software CONWEP), US Dept of the Army, Tech-
structures, Computers and Structures 40 (1991), 1105–1123. nical Manual, TM5-855-1, 1987.
[3] P.T. Nash, C.V.G. Vallabhan and T.C. Knight, Spall damage to [23] BS 5328: 1999 Concrete. Part 1: 1997 Guide to specifying
concrete walls from close-in cased and uncased explosions in concrete (amended May 1999), British Standards Institution,
air, ACI Structural Journal 92 (1995), 680–688. London, 1999.
 International Journal of

Rotating
Machinery

The Scientific
Engineering
Journal of Advances in
Journal of Mechanical
World Journal
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
Sensors
Hindawi Publishing Corporation
Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

International Journal of
Distributed Advances in
Sensor Networks
Hindawi Publishing Corporation
Civil Engineering
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

Advances in Submit your manuscripts at

OptoElectronics http://www.hindawi.com

Journal of
Robotics
Hindawi Publishing Corporation
Hindawi Publishing Corporation Volume 2014 http://www.hindawi.com Volume 2014
http://www.hindawi.com

VLSI Design

Modelling &
International Journal of Simulation
International Journal of Navigation and in Engineering Advances in
Chemical Engineering Observation Acoustics and Vibration
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014
Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014

Journal of

Control Science
and Engineering

International Journal of Journal of

Active and Passive Antennas and Electrical and Computer
Electronic Components Propagation Shock and Vibration Engineering
Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation Hindawi Publishing Corporation
http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014 http://www.hindawi.com Volume 2014

View publication stats