IMPACT RESPONSE BASED ON

TIMOSHENKO BEAM THEORY

Eid Khowitar
H. Ronald Riggs
and
Marcelo H. Kobayashi

Research Report UHM/CEE/15-02

June 2015
Copyright © 2015

By

Eid Khowitar

ii
 Acknowledgments
This report is a slightly modified version of the Ph.D. dissertation of Eid Khowitar.

The first author would like to express his deepest gratitude to his advisor Professor H. Ronald
Riggs and Professor Marcelo Kobayashi for their aspiring and genius guidance, invaluably
constructive criticism during this work. This work would not have been accomplished without
their guidance.

The first author would like also to thank the members of his dissertation committee,
Professor Ian N. Robertson, Professor David T. Ma, and Professor M. Asghar Bhatti.

Finally, he would like to thank his family, especially his wife Dr. Asmaa Abu Maziad for her
continuous support.

This material is based upon work supported by the National Science Foundation (via the
Network for Earthquake Engineering Simulations – NEES) under Grant No. 1041666. Any
opinions, findings, and conclusions or recommendations expressed in this material are those of
the authors and do not necessarily reflect the views of the National Science Foundation.

iii
 Abstract
The elastic impact of a translating flexible pole is studied herein. Three scenarios are
considered: 1) transverse impact against a rigid stop, 2) longitudinal impact against a flexible
column and 3) transverse impact against a flexible column. Based on Timoshenko beam theory,
an analytical solution method using mode superposition for the coupled spring-pole or column-
pole system is presented. Any physical set of boundary conditions can be accommodated for the
pole and the column.

For all cases involving axial impacts, the maximum initial impact force is governed by the
local shear deformation in the column and the axial deformation in the pole. However, for
transverse impacts, the maximum initial impact force is governed by the local shear deformation
in the pole and the column. A simple formula for the maximum initial force is derived and shown
to be quite accurate. In no case is the total mass of the pole significant to the initial peak force.
Indeed, based on Euler-Bernoulli beam theory the initial impact force is unbounded as the spring
stiffness increases whereas Timoshenko beam theory has a clear limiting value for the initial
impact force. The impact duration depends on the wave propagation in the pole or the column.

In addition, the energy transfer between kinetic energies and strain energies reveals both the
initial dependence on shear deformation and the transfer of the associated energy to bending
energy. The energy exchange also shows the importance of the inertia of the column in absorbing
a significant part of the initial kinetic energy of the pole. It is shown that the moment of inertia
has a negligible effect on the impact force, which is an interesting conclusion because some
structural finite element codes use a lumped mass matrix that includes translational masses but
not rotational inertias.

For transverse impact, multiple impacts are considered, and the whole collision event is
divided into contact phases and separation phases. It is shown that for all cases the maximum
contact force occurs during later contact phases and its value can reach up to 1.5 times the peak
force in the first contact phase. The impact duration of the first contact phase depends on the
shear wave in the pole or the column according to the mass and wave speed ratios. The total
impulse on the pole ranges between 1.5-1.8 times the initial momentum of the pole, depending

iv
on the stiffness of the column. The energy exchange during the multiple impacts, while it can be
complicated, reveals that for relatively stiff columns the sum of the translational kinetic and
bending strain energies of the pole constitutes approximately 90% of the total energy. In all cases
considered, relatively little net energy has been transmitted to the column at the time of final
separation.

For axial impact, multiple impacts depend on the relative stiffness of the column and the pole,
and also on the inertia of the pole. Hence, the entire collision event for the stiffest column is
characterized by a single impact. However, for the most flexible column all cases involve
multiple impacts. For the case of single impact, most of the kinetic energy of the pole is
transferred into axial strain energy in the pole. However, for the multiple impacts, most of the
kinetic energy of the pole is transferred into bending energy in the column. The maximum
impact force reaches up to 1.9 times the initial peak force and the total impulse reaches up to 1.9
times the initial momentum of the pole. For all cases, the duration of the entire collision event
depend mainly on the wave propagation in the pole.

The impact force and duration depend on the type of impact as well as the end boundary
conditions of the column. For all cases, the axial impact yields larger impact force than that for
the transverse impact according to the stiffness of the column with similar boundary conditions
of the column. The stiffer the column, the larger is the impact force and the smaller the impact
duration. In addition, free end column yields the smallest impact force and duration. However,
pin-end column gives the largest impact duration and fixed-end column gives the largest impact
force.

The dynamic amplification factors for shear force and bending moment depend mainly on the
stiffness of the column and the inertia of the pole. Cases involving stiffer column and larger pole
inertia yield higher dynamic amplification factors. In addition, the dynamic amplification factors
significantly increase for cases involving multiple impacts and always reach their maximum
values at later impacts.

v
 Table of Contents
Acknowledgments………………………………………………………………………..iii

Abstract .............................................................................................................................. iv

Table of Contents ............................................................................................................... vi

List of Tables ...................................................................................................................... x

List of Figures .................................................................................................................... xi

Chapter 1 Introduction ........................................................................................................ 1

1.1 Background ......................................................................................................... 1

1.2 Previous Work .................................................................................................... 2

1.3 Scope of Work .................................................................................................... 5

1.4 Outline................................................................................................................. 5

Chapter 2 Transverse Impact of a Beam on a Flexible Stop............................................... 8

2.1 Introduction ......................................................................................................... 8

2.2 Physical Systems ................................................................................................. 8

2.3 Timoshenko beam model .................................................................................... 9

2.4 Analytical Solution ........................................................................................... 11

2.4.1 Separation of variables and uncoupling equations of motion ................... 12
2.4.2 Solution of the generalized coordinate function ....................................... 13
2.4.3 Solution of the mode shape functions ....................................................... 14
2.4.4 Dimensional boundary conditions ............................................................ 16
2.4.5 Dimensional compatibility conditions at the impact point ....................... 16
2.4.6 Nondimensional boundary conditions ...................................................... 17
2.4.7 Nondimensional compatibility conditions at the impact point ................. 17
2.4.8 Orthogonality condition ............................................................................ 18
2.4.9 Upper bound of the initial impact force .................................................... 20
2.4.10 Impact duration ......................................................................................... 22

vi
 2.4.11 Peak Force using Euler-Bernoulli beam theory ........................................ 22
2.4.12 Mechanical Energy ................................................................................... 22

2.5 Results ............................................................................................................... 25

2.5.1 Effect of spring stiffness ........................................................................... 26
2.5.2 Effect of 𝜏 ................................................................................................. 30
2.5.3 Effect of slenderness ratio 𝜆𝐵 ................................................................... 31
2.5.4 Effect of 𝜂 ................................................................................................. 33
2.5.5 Asymmetric impact of rotating beam ....................................................... 33
2.5.6 Euler vs. Timoshenko ............................................................................... 35
2.5.7 Energy breakdown and beam vibration .................................................... 41

2.6 Summary ........................................................................................................... 45

Chapter 3 Beam Response to Longitudinal Impact by a Pole .......................................... 47

3.1 Introduction ....................................................................................................... 47

3.2 Physical System ................................................................................................ 48

3.3 Equation of Motion ........................................................................................... 49

3.4 Analytical Solution ........................................................................................... 51

3.4.1 Mode Shapes ............................................................................................. 51
3.4.2 Characteristic Equation ............................................................................. 51
3.4.3 Initial Peak Impact Force .......................................................................... 53
3.4.4 Mechanical Energy ................................................................................... 56

3.5 Results ............................................................................................................... 58

3.5.1 Parameter Range ....................................................................................... 58
3.5.2 Axial Impact of a Pole against a Column ................................................. 59
3.5.3 Comparison of Timoshenko versus Euler Models .................................... 65
3.5.4 Effect of Rotary Inertia ............................................................................. 68
3.5.5 Energy Breakdown.................................................................................... 69

3.6 Summary ........................................................................................................... 73

Chapter 4 Transverse Impact of a Horizontal Beam on a Vertical Column .................... 75

vii
 4.1 Introduction ....................................................................................................... 75

4.2 Physical System ................................................................................................ 76

4.3 Equations of Motion ......................................................................................... 78

4.4 Analytical Solution ........................................................................................... 79

4.4.1 Separation of Variables ............................................................................. 79
4.4.2 Initial Peak Impact Force .......................................................................... 82
4.4.3 Mechanical Energy ................................................................................... 83

4.5 Solution Methodology ...................................................................................... 85

4.6 Numerical Results ............................................................................................. 85

4.6.1 Physical Dimensions ................................................................................. 85
4.6.2 Force-Time History................................................................................... 87
4.6.3 Energy-Time History ................................................................................ 89
4.6.4 Impulse-Time History ............................................................................... 92
4.6.5 Energy Decomposition by Modes ............................................................. 94
4.6.6 3D Energy Density:................................................................................... 96

4.7 Computational Aspects ..................................................................................... 99

4.8 Summary ......................................................................................................... 101

Chapter 5 Additional Scenarios ...................................................................................... 103

5.1 Introduction ..................................................................................................... 103

5.2 Axial Impact.................................................................................................... 104

5.2.1 Wood pole hitting concrete column ........................................................ 104
5.2.2 Wood pole hitting steel column .............................................................. 107
5.2.3 Wood pole hitting wood column............................................................. 110

5.3 Transverse Impact ........................................................................................... 114

5.4 Maximum Shear and Bending Envelopes ....................................................... 117

5.4.1 Envelope versus time .............................................................................. 118
5.4.2 Envelope versus space ............................................................................ 123

viii
 5.5 Summary ......................................................................................................... 128

Chapter 6 Conclusion ...................................................................................................... 131

6.1 Response of a Beam Hitting Transversely a Stop ........................................... 131

6.2 Beam Response to Longitudinal Impact by a Pole ......................................... 132

6.3 Transverse Impact of a Horizontal Beam on a Vertical Column .................... 133

6.4 Additional Scenarios ....................................................................................... 134

ix
 List of Tables
Table 1: Dimensional Properties ....................................................................................... 60

Table 2: Nondimensional Parameters ............................................................................... 60

Table 3: Length Ratios ...................................................................................................... 60

Table 4: Dimensional Properties ....................................................................................... 86

Table 5: Nondimensional Parameters ............................................................................... 87

Table 6: Maximum dynamic amplification factors for shear force for axial impact ...... 127

Table 7: Maximum dynamic amplification factors for bending moment for axial impact
................................................................................................................................. 128

Table 8: Maximum dynamic amplification factors for shear force for transverse impact
................................................................................................................................. 128

Table 9: Maximum dynamic amplification factors for bending moment for transverse impact
................................................................................................................................. 128

x
 List of Figures
Figure 1: Schematic of beam and stop spring ..................................................................... 9

Figure 2. Impact zone at time t * ....................................................................................... 21

Figure 3: Impact force for approximately "rigid" impact with 𝜏 = 10, 𝜂2 = 1, 𝜆𝑏 = 120
................................................................................................................................... 28

Figure 4: Impact force for a range of 𝑘, with 𝜂2 = 1, 𝜆𝐵 = 120 and 𝜏 = 10 ................. 29

Figure 5: Impact force for a range of 𝑘, with 𝜂2 = 1, 𝜆𝐵 = 120 and 𝜏 = 20/9 ............. 29

Figure 6: Impact force for a range of 𝜏, with 𝑘 = 1200, 𝜂2 = 1, 𝜆𝐵 = 120 ................... 31

Figure 7: Impact force for a range of 𝜆𝐵, with 𝑘 = 1200, 𝜂2 = 1, 𝜏 = 10 ..................... 32

Figure 8: Impact force for a range of 𝜆𝐵, with 𝑘 = 10, 𝜂2 = 1, 𝜏 = 10 .......................... 32

Figure 9: Impact force for 𝐿𝑟 = 120, 𝜏 = 10 with soft and stiff springs and 𝜂2 = 1 and 𝜂2 =
0.01 ........................................................................................................................... 33

Figure 10: Impact force for impact at quarter span, 𝜆𝐵 = 120, 𝜏 = 10, 𝜂2 = 1 𝑎𝑛𝑑 𝑘 = 1200
(All plots coincide) ................................................................................................... 34

Figure 11: Branches of dimensionless group speed 𝑐𝑔 = 𝑐𝑔 ∗/𝑐𝑠 as a function of the

dimensionless wave number 𝜉 = 𝜉 ∗ 𝑟 for 𝜏 = 10 ................................................. 36

Figure 12: Comparison of impact forces between Timoshenko and Euler-Bernoulli beam theories
for different 𝑘, with 𝜆𝐵 = 120, 𝜏 = 10, 𝜂2 = 1 ....................................................... 41

Figure 13: Time evolution of potential and kinetic energy components (log scale) for 𝜆𝐵 =
120, 𝜏 = 10, 𝜂2 = 1 and 𝑘 = 1200 ......................................................................... 43

Figure 14: Space-time plot of potential energy density for 𝜆𝐵 = 120, 𝜏 = 10, 𝜂2 = 1 and 𝑘 =
1200 .......................................................................................................................... 44

xi
Figure 15: Space-time plot of kinetic energy density for B  120 ,   10 ,   1 , k  1200
2

................................................................................................................................... 45

Figure 16: Schematic of beam/column and pole............................................................... 49

Figure 17: Impact zone at time t ....................................................................................... 54

Figure 18: Impact force for a wood pole hitting fixed-fixed concrete columns ............... 62

Figure 19: Impact force for a wood pole hitting fixed-fixed steel columns ..................... 63

Figure 20: Impact force for a wood pole hitting fixed-fixed wood columns .................... 64

Figure 21: Impact force for a wood pole hitting steel columns (left axis) and the end rotation of
the pinned column (right axis) for the case of C2P2 ................................................ 65

Figure 22: Impact force for a wood pole hitting fixed-end concrete (C), steel (S), and wood (W)
columns for case C1P2 based on Timoshenko (T) and Euler (E) beam theories ..... 67

Figure 23: Impact force for a wood pole hitting a fixed-end wood column for case C1P2 for
different ratios of the contact stiffness to the axial pole stiffness, for Timoshenko (T) and
Euler (E) theories ...................................................................................................... 67

Figure 24: Effect of rotary inertia on impact force for L/r=120 and   20 / 9 with   1 and
2

 2  0.01................................................................................................................... 69

Figure 25: Shear and bending strain energy in the wood column for case C1P2 ............. 71

Figure 26: Component energies and total energy for the case C1P2 with a wood column72

Figure 27: Time variation of energy densities for case C1P2 with a wood column ......... 73

Figure 28: Schematic of pole and column ........................................................................ 78

Figure 29: Impact force time histories .............................................................................. 89

Figure 30: Component energies and total energy ............................................................. 92

xii
Figure 31: Impulse time histories a wood pole hitting columns of different lengths and materials
................................................................................................................................... 94

Figure 32: Modal contribution of energy of the pole for short concrete column (left) and long
wood column (right) ................................................................................................. 96

Figure 33: Time variation of energy densities of the pole for short concrete column (left) and
long wood column (right) ......................................................................................... 99

Figure 34: Force time history for 3.6 m concrete column obtained from analytical and FEA
solution.................................................................................................................... 101

Figure 35: Component energies and total energy for a wood pole hitting axially a concrete
column..................................................................................................................... 105

Figure 36: Impulse for a wood pole hitting axially a concrete column .......................... 106

Figure 37: Force-time history for case C2P2 of a wood pole hitting axially a steel column107

Figure 38: Energy components and total energy for a wood pole hitting axially a steel column
................................................................................................................................. 109

Figure 39: Impulse for a wood pole hitting axially a steel column ................................ 110

Figure 40: Impact force for a wood pole hitting axially a wood column ....................... 111

Figure 41: Energy components and total energy for a wood pole hitting axially a wood column
................................................................................................................................. 113

Figure 42: Impulse for a wood pole hitting axially a wood column ............................... 114

Figure 43: Impact force for 9 m wood pole hitting transversely a 6 m wood column.... 115

Figure 44: Energy components and total energy for a 9 m wood pole hitting transversely a 6 m
wood column ........................................................................................................... 116

Figure 45: Impulse for a 9 m wood pole hitting transversely a 6 m wood column ........ 117

xiii
Figure 46: Maximum shear force (left) and bending moment (right) amplification factors for
axial impact ............................................................................................................. 119

Figure 47: Maximum shear force (left) and bending moment (right) amplification factors for
transverse impact .................................................................................................... 122

Figure 48: Maximum shear force (left) and bending moment (right) amplification factors for
axial impact ............................................................................................................. 125

Figure 49: Maximum amplification factors for transverse impact ................................. 127

xiv
 Chapter 1
Introduction

1.1 Background
A significant threat to structures in the tsunami inundation zone is impact from debris driven
by the tsunami flow (NRC 2004). A proper characterization of these forces is especially
important to life-safety related to vertical tsunami evacuation shelters (FEMA 2012). Debris
driven by tsunami waves can cause catastrophic damages to coastal buildings. Debris is
composed of different materials, such as building fragments, boats, vehicles, wood poles, docks,
and shipping containers. In March 2011, huge tsunami waves hit the coast of Japan resulting in
approximately 16,000 deaths, 6,000 injured, and damage to the buildings. 5 million tons of
debris swept in the Pacific Ocean, 70 % of them sank near the coast of Japan. The remaining
30% of the debris moved in the ocean and some of them appeared in the coasts of the United
States and Canada in late 2011 (Alaska 2013; Toro 2012).

On September 29, 2009, an earthquake of magnitude 8.0 hit the islands of Samoa.
Consequently, destructive tsunami waves were generated and washed the islands, causing
damages to the buildings and fatalities. It has been reported that almost 137 people died and 310
were injured. In addition, some of the buildings suffered damages due to hydrodynamic and
debris impact loads (Robertson et al. 2010).

On December 26, 2004 a huge earthquake hit Sumatra and was followed by devastating
tsunami waves in the Indian Ocean. Final estimation showed that 226,226 people died, including
49,648 missing people, and nearly two million people were forced to evacuate. One of the main
causes of damage to buildings was debris impact, especially exposed structures (Rossetto et al.
2006).

Hurricane Katrina is considered one of the deadliest storms that hit the United States.
Approximately 1,836 people died and millions became homeless (Zimmermann 2012) . In
August 2005, the storm reached its peak (storm category 5) before it hit the Gulf Coast of
Louisiana, Mississippi, and Alabama. It lost some of its strength at the time of landfall in

1
Louisiana/ Mississippi border; however, the storm surge did not abate rapidly. Many structures
experienced significant damages due to floating or mobile debris, such as shipping containers,
boats, and barges (Robertson et al. 2007).

From the examples mentioned above, it is clear that debris impact can cause profound
damage to coastal structures and loss of lives. The situation would be even more dramatic if
dangerous chemical or radioactive substances were to spread in the surroundings. Unfortunately,
many coastal areas are not designed to resist the damages caused by tsunamis. Relatively little
research has been devoted to tsunami-driven debris, although recent tsunamis have illustrated the
potential for structural damage from such debris. Low velocity impact of high mass, water-
driven debris on civil-type structures has received attention primarily related to flood-borne
woody debris (Haehnel and Daly 2002; Matsutomi 2009), barge collision on bridge piers
(Consolazio et al. 2006; Consolazio and Cowan 2005; Consolazio et al. 2009) and navigation
locks (Arroyo-Caraballo and Ebeling 2006).

The objective of this work is to improve our understanding of, and predictive capabilities for,
tsunami-driven debris impact forces on structures. The focus is on low velocity and high mass
woody debris, which is one of the more prevalent substantial debris in tsunami flow for industrial
and densely populated coastal areas. Impact forces specified by current codes and standards are
based on rigid body dynamics, while the work herein includes the flexibility of the debris as well
as the structural member (column) for more accurate prediction of impact forces.

1.2 Previous Work

Many studies on the impact response of beams have been carried out. The studies are often
motivated by a number of important scenarios associated with mechanical equipment, such as
piping systems, heat exchangers and valves, and fall into three primary categories: a (typically
rigid) mass impacting a beam, a beam hitting a rigid or flexible stop, and two beams hitting each
other. An early study was Timoschenko (1914), who used an Euler beam, a Hertzian contact
model, and mode superposition to obtain a solution for a beam hit by a mass. Boley and Chao
(1955) used the Laplace transform to study the behavior of a semi-infinite beam under transverse
impact. The impact was modeled as a “sudden load” and numerical integration was used to
compute the definite integrals in the analytical solution. These authors briefly compared

2
numerical results for Timoshenko and Euler-Bernoulli beam theories under these conditions, and
obtained poor correlation for the initial force propagation. Goyder and Teh (1989) used a single-
degree-of-freedom spring model to evaluate repeated impacts of a tube in a loose support. In an
attempt to reduce the computational complexity of impact simulation, Evans et al. (1991)
developed an efficient numerical integration scheme for a mass hitting a round Kirchhoff plate,
based on a Hertzian contact model and mode superposition.

Xing et al. (2002) and Vinayaravi et al. (2013) studied impact of a rigid mass on a beam.
Vinayaravi et al. (2013) investigated the damping as a result of repeated impact of a mass on a
cantilever beam at the tip, including repeated impacts. The physical system was modeled as a
two degree-of-freedom spring-mass-damper system. They observed that the ‘damping’ depends
on the number of effective impacts (impact with higher relative velocity between the beam and
the mass) and not on the total number of impacts. Xing et al. (2002) used Timoshenko beam
theory and modal superposition to study the impact of a rigid mass hitting a beam at midspan.
They used conservation of linear momentum to estimate the initial impact force and observed
that the initial value of the impact force depends on the shear wave propagation in the beam and
not on the flexural wave propagation. However, when the flexural wave returns to the impact
point, the impact force starts to have small vibrations, and when the shear wave returns to the
impact point, separation occurs between the mass and the beam; hence, there was only one
impact. Ervin and Wickert (2007) used modal superposition to model multiple impacts of a mass
on a fixed Euler beam hitting a compliant stop.

Yin et al. (2007) and Wang and Kim (1996) studied the transient behavior of a cantilever
beam hitting a rod. Both papers used Euler-Bernoulli beam theory and modal superposition to
solve the beam equation of motion and the one-dimensional wave equation for the axial response
of the rod. In Yin et al. (2007), the beam tip was subjected to a periodic force. The repeated
‘impact’ is divided into three phases: pre-impact, impact, and separation. The results showed that
the impacts induce high frequency response. Wang and Kim (1996) observed that representing
the valve stop as a spring yields good results for a short stop; however, for a relatively long stop
the results deviate greatly from the exact solution in which the inertia of the rod is considered.
Wagg and Bishop (2002) considered a cantilever Euler beam hitting a rigid stop. They assumed a

3
coefficient of restitution and focused on pre- and post-impact, rather than on the response during
impact. Again, normal modes were used.

Packzkowski (2012) studied the axial impact of a projectile hitting a rigid wall. The author
was able to predict the impact force and duration using one-dimensional wave equation to model
the projectile. The results showed that the impact force does not depend on the total mass of the
projectile and that the impact duration is the time taken by the axial wave to propagate in the
projectile and come back to the impact point. Khowitar et al. (2014) studied longitudinal (axial)
impact of a wood pole against a column. The one dimensional axial wave equation of motion
was used for the pole and Timoshenko beam theory was used for the column. A stiff spring was
placed between the pole and the column to model a finite contact stiffness. The modal
superposition method was used to solve for the analytical response. A closed form equation for
the instantaneous peak impact force was also derived. It was shown that the instantaneous impact
force is dominated by the shear wave in the column, and hence, Euler-Bernoulli beam theory
yields inaccurate results except for quite low values of contact stiffness. A study on the energy
transfer during the impact was also conducted.

Dorogy and Rittel (2008) conducted experiments and a three-dimensional finite element
analysis of a free-free beam subjected to transverse impact by a Hopkinson bar at four locations
along the beam. They considered elastic-plastic behavior of the beam, with bilinear isotropic
hardening as well as geometrical nonlinearity, in the numerical analysis. They observed that
most of the plastic deformation occurs very soon after impact. Symmetrical impact has the
maximum impact force and dissipated energy due to plastic deformation. Approximately 76% of
the initial energy is converted into plastic deformation and approximately 21% is transferred into
kinetic energy of the beam. Langley (2012) developed a statistical solution for randomly
impacting objects assuming that the force time history is approximated by a quarter sine curve.
He was able to predict the impact force and duration of two identical Euler beams hitting each
other with a non-linear spring at the impact point. Ervin (2009) studied multiple impacts of two
beams subject to base excitation at the supports. Euler-Bernoulli beam theory and modal
superposition were used to obtain the response. A spring was placed between the beams at the
impact point to account for the contact stiffness. She studied the effect of different parameters on
the multiple impact response spectra. Christoforou and Yigit (1998) found the main parameters

4
on which the impact force depends for two special impact cases. The first case was half space
impact, in which the target mass is very large compared to the mass of the impactor. In this case
the flexural wave is not reflected back to the impact point from the boundary of the target and the
deformation is localized around the impact point. Therefore, they used an infinite beam and
Timoshenko beam theory to find the peak force. They noticed that the main parameter affecting
the impact force is the normalized impact velocity. In the second case, the mass of the impactor
was very large compared to the target. In this case the problem was quasi-static, in which a load
is applied at the impact point. The mass of the target is neglected and the problem can be seen as
a single degree of freedom system with contact stiffness and static beam stiffness. They found
that the only parameter affecting the impact force in this case is the ratio of the beam static
stiffness to the contact stiffness.

1.3 Scope of Work

The objective of this work is to improve our understanding of, and predictive capabilities for,
tsunami-driven debris impact forces on structures. A low velocity, high mass, in-air debris
impact theory based on flexible body dynamics is developed.

Significant amount of work has been carried out on the beam impact. However, most of this
work is based on the Euler-Bernoulli theory to obtain the response, which is proved later to be
inadequate and yields inaccurate results for the axial and the transverse impact. It will be shown
in this study that the impact force depends mainly on the shear deformation of the beam for
transverse impact, and hence Timoshenko beam theory is used to account for the shear
deformation and the rotary inertia. On the other hand, the multiple impacts for beam-beam
transverse impact have not received much attention, but will be discussed in detail herein.

1.4 Outline
Three different scenarios are considered in the following chapters: 1) a pole hitting
transversely a flexible stop, 2) a pole hitting longitudinally a column, and 3) a pole hitting
transversely a column. Each of the chapters are written to stand essentially by themselves.
Therefore, they are a complete statement of the problem, and therefore there is some overlap
between the chapters.

5
 In Chapter 2, the transverse impact of a pole hitting a massless spring is discussed. In this
case, the pole can have arbitrary initial translational and rotational velocities, and both
Timoshenko and Euler-Bernoulli beam theories are considered. A detailed parameter study is
carried out to investigate the effect of the nondimensional parameters on the impact force time
history.

In Chapter 3, axial impact of a pole hitting a flexible column is investigated. The column can
have any arbitrary end boundary conditions. A simple design formula for the peak value of the
impact force and the impact duration is derived. Moreover, the energy exchange is discussed as it
gives insight on the behavior of the impact force.

In Chapter 4, transverse impact of a pole hitting a column is investigated. The behavior of the
impact force and relevant phenomena are studied for multiple impacts. Simple formulas for the
peak impact force and duration are derived. To understand better the behavior of the impact force
during the multiple impacts, the energy exchange among various components as well as the force
impulse are investigated.

In Chapter 5, additional scenarios for axial and transverse impacts that supplement the results
in the previous two chapters are considered. The axial impact simulations presented in Chapter 3
only considered the first contact phase. In this chapter, the same scenarios are simulated but
through the entire impact event, capturing the multiple impacts, if any. The maximum impact
force, which normally does not occur during the first contact phases, is sought. Moreover, the
main differences between cases that exhibit single and multiple impacts will be clarified from the
energy exchange. For transverse impact, free-end and pin-end column boundary conditions are
considered to supplement the fixed-end boundary conditions in Chapter 4. The free-end
condition, while not particularly relevant for a pole hitting a column, has some interest in the
transverse impact of two poles, for example.

In an attempt to assess the values of the initial peak force obtained herein for design purposes,
it is useful to give an insight on whether applying this force as a static load can bound the
dynamic response. The values of the maximum shear force and bending moment are obtained by
applying a static load of a value equal to the initial peak force at mid-span of a fixed-end column.

6
These values are then compared to the values of the shear force and the bending moments from
the dynamic analyses.

Finally, in Chapter 6, conclusions of the main concepts and major findings are presented.

The results of this study will contribute to improving community resilience to tsunamis. The
results have the potential to impact significantly the specification of design forces for debris in
codes and standards. Debris impact is an especially important design consideration for tsunami
shelters, fuel and chemical storage tanks, and port and industrial facilities, all of which may
unavoidably be located in tsunami inundation zones. In addition, the results will be applicable to
hurricane-driven, water-borne debris, and to some extent to barge and ship collisions of bridge
piers, docks, and navigation locks. The models are a significant improvement over that which is
currently used for design forces.

7
 Chapter 2
Transverse Impact of a Beam on a Flexible Stop

2.1 Introduction
The investigation herein is motivated by the threat of debris strikes on buildings, such as by a
floating log propelled by a tsunami or hurricane storm surge and striking a column. As such, the
focus is on the impact force during single impact. A previous study (Kobayashi et al. 2012;
Paczkowski et al. 2012) considered axial impact, and this study focuses on transverse impact. In
this chapter, an initial model is investigated that involves a free-free beam with initial
translational and rotational velocities striking a flexible spring anywhere within the span. The
response of the beam is investigated to guide the models adopted in subsequent chapters for
more accurate models of the pole-column impact. The discussion is restricted to linear impact
response. An approach similar to Xing et al. (2002) is followed to develop analytical solutions
for both Timoshenko and Euler beam theories, but the approach adopted herein is more general
and covers a broader range of problems. The range of applicability of the two beam theories is
investigated. Of primary interest is the high resolution capture of the impact force time history.

This chapter is organized as follows. First, the governing equations are defined and
nondimensionalized. Then the analytical solutions, which are based on superposition of the
normal modes of vibration, are developed. A parameter study is conducted to explore the effect
of several key nondimensional parameters on the impact behavior, and to evaluate the range of
validity of the Euler beam model for this application. The final section contains conclusions from
the present study.

2.2 Physical Systems

A schematic of the system is shown in Figure 1. The figure indicates a pole with an initial
translational and angular velocity striking a spring with a stiffness k s* . The pole is assumed to be
elastic and homogenous, with constant cross-sectional properties.

8
 𝑘𝑠∗

𝑥∗
𝐿∗1 𝜔0∗ 𝐿∗2
𝐿∗
𝑣0∗
Figure 1: Schematic of beam and stop spring

2.3 Timoshenko beam model

Under the assumptions above, the equations of motion for a Timoshenko beam model can be
written as

 2 w*   2 w*   * * *
A   GA  2    ks w   0 (2.1)
t * x* 
2
 x*

 2  w*   2
Im   GA  *     EI *2  0 (2.2)
t *  x  x
2

where w ( x , t ) is the transverse displacement of the pole,  ( x , t ) is the rotation of the cross
* * * * *

section of the pole,  is the mass density, A is the cross-sectional area, E is the Young's elastic

modulus, G is the shear modulus, I is the cross-sectional area moment of inertia, I m   2  I is

the sectional rotary inertia of the beam,  is the shear coefficient, k s* is the spring stiffness,  is
*

the Kronecker delta, and L*  L*1  L*2 is the length of the pole. The impact point can be at any

arbitrary point along the pole ( x  0 ).

*

The boundary conditions at the beam ends are null shear force and null moment. The shear
force at the free ends is specified by

 * 
 GA  w*    0
 x  x*  L*1
(2.3)
 * 
 GA  w*    0
 x  x*  L*2

and the bending moment at the free ends is specified by

9
   
EI  *  0
 x  x*  L*1
(2.4)
  
EI  *  0
 x  x*  L*2

At impact, the beam's center of mass is assumed to move at a speed of v 0* , while the beam

rotates at a rate of 0* .

To reduce the number of parameters while providing broader scope for each solution, these
model equations are cast in nondimensional form using the same nondimensionalization as used
by Xing et al. (2002):

x  x* / r
w  w* / r
L  L* / r  B
(2.5)
L1,2  L*1,2 / r
t  t *c0 / r
k s  k s* /  EA / r 

where r  I / A is the radius of gyration of the beam section and c 0  E /  is the speed of
sound in the beam.

The nondimensionalization in Eqs. (2.5), based on r and r / c 0 , is convenient to develop the

solution. However, the length scale L* and time scale L* / c s , where c s  G /  is the shear
speed in the beam, are more meaningful physically. Therefore, we will use the more physically
relevant parameters to discuss the results.

The space derivatives of the kinematic variables are

10
 w* wr x wr 1 w
  
x* x x* x r x
 2 w*  w  2 w x*  2 w 1
  
x* x* x x 2 x x 2 r
2

(2.6)
  x  1
 
x* x x* x r
 2   1  2 1 x*  2 1
  
x* x* x r x 2 r x x 2 r 2
2

The time derivatives of the kinematic variables are

w* wr t wr c0 w

   c0
t * t t * t r t
 2 w*  w  2 w t *  2 w c02
 c  c0  2
t * t t 2 t t r
0
t *
2

(2.7)
  t  c0
 
t * t t * t r
 2   c0  2 c0 t *  2 c02
  
t * t * t r t 2 r t t 2 r 2
2

These parameters can be used to express the equations of motion in nondimensional form.
Substituting the non-dimensional parameters in Eq. (2.1) results in

 2 w 1   2 w  
  0 (2.8)
t 2   x 2 x 

where   E /  G   c02 / cs2 . Substituting the non-dimensional parameters in Eq. (2.2) results in

 2 1  w  
2
2       0 (2.9)
t 2   x  x
2

2.4 Analytical Solution

An analytical solution to the Timoshenko model above is obtained using the method of
modal superposition.

11
 2.4.1 Separation of variables and uncoupling equations of motion
Because of the discontinuity in the shear force at the impact point, the displacement and
rotation fields are divided into two parts:

 
 w ,  L1  x  0
w 
 w , 0  x  L2

 ,  L1  x  0
 

  
  , 0  x  L2


Each of the displacements and the rotations are functions of x and t . These displacements can be
expressed as the product of two functions; one is dependent on x and the other is dependent on
t:

w( x, t )   ( x)q(t )
(2.10)
 ( x, t )   ( x)q(t )

where  (x ) and  (x ) are the mode shape functions and q (t ) is the generalized coordinate
function.

Substitution of Eqs. (2.10) into Eq. (2.8) yields

1
 q   q  q   0

 q  q     
q      
   2
q 

where  is the non-dimensional natural frequency of the beam and the dimensional natural
frequency *    c 0 / r . Thus,

q  2q  0 (2.11)

      2  0 (2.12)

Substitution of Eqs. (2.10) into Eq. (2.9) results in

12
 1  1
        2 2    0 (2.13)
  

Taking the first derivative of Eq. (2.12) yields

      2   0 (2.14)

Rearranging terms in Eq. (2.13) and taking the second derivative with respect to x results in

       2 2  1
(2.15)
    (4)   2 2  1 

Substitution of Eqs. (2.15) into Eq. (2.14) yields the differential equation for the rotational mode
shape function

 (4)   2 2   2     4 2   2   0 (2.16)

Following the same procedure, the differential equation for the translational mode shape
function is

 (4)   2 2   2     4 2   2   0 (2.17)

2.4.2 Solution of the generalized coordinate function

t
Let q  e . Substitution into Eq. (2.11) results in

V2 +w2 = 0

The roots of the above equation are 1,2  i  , and thus the solution becomes

q  c1eit  c2eit

Using the Euler formulas, the solution can be written as

q  c1  cos  t  i sin  t   c2  cos  t  i sin  t 

  c1  c2  cos  t   ic1  ic2  sin  t

Thus,

13
 q  C1 cos  t  C2 sin  t (2.18)

2.4.3 Solution of the mode shape functions

x
Let   e . Substitution into Eq. (2.17) results in

 4   2 2   2   2   4 2   2  0

The roots of this equation are

1,2  i   2  2     4   2  2    / 2
2

 i 

3,4     2  2     4   2  2    / 2
2

 

The mode shape function then becomes

  x   a1ei x  a2ei x  A3e x  A4e x

Using the Euler formulas in the mode shapes function gives

  x   a1  cos  x  i sin  x   a2  cos  x  i sin  x   A3e x  A4e  x

  a1  a2  cos  x   a1i  a2i  sin  x  A3e x  A4e  x

Thus, the mode shape function for the displacement is

  x   A1 cos  x  A2 sin  x  A3e x  A4e x (2.19)

Using the same procedure, the mode shape function for the rotation becomes

  x   B1 cos  x  B2 sin  x  B3e x  B4e x (2.20)

A relationship can be found between B i and A i by substituting the mode shape functions in
Eq. (2.12):

14
   A cos  x   A sin  x   A e   A e  
2
1
2
2
2
3
x 2
4
 x

  B sin  x   B cos  x   B e   B e  
1 2 3
x
4
 x

   A cos  x  A sin  x  A e  A e   0
2
1 2 3
x
4
 x

Rearranging terms

  A   B    A  cos  x    A   B    A  sin  x 
2
1 2
2
1
2
2 1
2
2
(2.21)
 A   B    A  e   A   B    A  e  0
2
3 3
2
3
x 2
4 4
2
4
 x

Eq. (2.21) is satisfied only if each coefficient of cos  x , sin  x , e  x , and e  x is null. Hence,

  2 A1   B2   2 A1  0
  2 A2   B1   2 A2  0
 2 A3   B3   2 A3  0
 2 A4   B4   2 A4  0

This leads to

 2   2
B2  A1    A1

  2   2
B1  A2    A2

 2   2
B3  A3   A3

 2   2
B4  A4   A4


where

 2   2
 

 2   2
 


Hence, the mode shapes on each side of the discontinuity at x  0 are

 
 
  A1 cos  x  A2 sin  x  A3 e x  A4 e x 


    x   x 
(2.22)
  ( A1 sin  x  A2 cos  x)   ( A3 e  A4 e ) 
 
 
   

15
 Note that  is always real, but  can be real or imaginary. A special mode shape must be

considered when   1    ,   0 , in which the mode shapes can be obtained by taking

the limit of Eq. (2.22). We note that these eigenfunctions are consistent with Xing et al. (2002),
who considered only the case  = 1.

2.4.4 Dimensional boundary conditions

The dimensional boundary conditions for the shear force at the ends of the beam are

  * 


 GA  *    0
 x 
  x*  L1*
  * 


 GA  *    0
 x 
  x*  L*2

The dimensional boundary conditions for the bending moment at the ends of the beam are

 
EI * 0
x x*  L1*

 
EI 0
x* x*  L*1

2.4.5 Dimensional compatibility conditions at the impact point

The dimensional compatibility condition for the displacement at the impact point is

 *  0   *  0
 

The dimensional compatibility condition for the rotation at the impact point is

   0     0

The dimensional compatibility condition for the bending moment at the impact point is

   
EI  EI
x * x * 0
x * x * 0

The difference in the shear forces at the impact point equals the force in the spring, thus the

16
dimensional compatibility condition for the shear force at the impact point is

  *    * 
 

GA  *     GA  *   
 x 
 k s*av* (0)
 x  x * 0   x * 0

 
in which av* (0)   *  0    *  0  / 2 .
 

2.4.6 Nondimensional boundary conditions

The nondimensional boundary conditions for the shear force at the ends of the beam are

   
    0
 x  x  L1
   
    0
 x  x L2

The nondimensional boundary conditions for the bending moments at the ends of the beam are

 
0
x x  L1

 
0
x x  L2

2.4.7 Nondimensional compatibility conditions at the impact point

The nondimensional compatibility condition for the displacement at the impact point is

   0     0

The nondimensional compatibility condition for the rotation at the impact point is

   0     0

The nondimensional compatibility condition for the bending moment at the impact point is

   

x x 0
x x 0

The nondimensional compatibility conditions for the shear force at the impact point is

17
       ks* *
    av (0)
 x x  x 0  GA
ks  EA / r 
 av (0) r
 GA
  ksav (0)

2.4.8 Orthogonality condition

The dimensional orthogonality condition can be written as

0 L*2

 (m   I m  j )dx   (mi*  *j  I m i j )dx*  0

* *   *  

i j i
 L*1 0

where m   A is the mass per unit length of the beam. The nondimensional form of the
orthogonality condition would be

0 L2


 L1
(i j   2 i j )dx   (i j   2 i j )dx  0
0

The coefficients of the mode shape functions A i and A i , i  1, 2,3, 4 are 8 coefficients,
which is the same as the number of boundary and compatibility conditions. Hence, the
coefficients can be determined exactly by enforcing satisfaction of the boundary and
compatibility conditions.

The four boundary conditions and four compatibility equations result in a system of
equations in the eight coefficients plus the natural frequencies  . For there to be a non-null
solution, the characteristic equation must be zero:

18
  a11 a12 a13 a14 0 0 0 0 
 
 a21 a22 a23 a24 0 0 0 0 
 0 0 0 0 a35 a36 a37 a38 
 
 0 0 0 0 a45 a46 a47 a48 
 0    0     
 
 1 0 1 1 1 0 1 1 
  0      0     
 

 1 k k 1 k k 
  (k )       (k )     
 2 2 2 2 2 2 

L
with a11  (    )sin(L1 ) ; a12  (    ) cos(L1 ) ; a13  e L1 (   ) ; a14  e 1 (   ) ;

a21    cos(L1 ) ; a22    sin(L1 ) ; a23  e L1 ; a24  eL  ; a35  (    )sin(L2  ) ;
1

a36  (    )cos(L 2  ) ; a37  eL2 (   ) ; a38  e L2 (   ) ; a45    cos(L2  ) ;

a46    sin(L2  ) ; a47  eL2 ; and a48  e L2

The values of  that satisfy this equation are the natural frequencies of the system.

General formulas of the generalized function coefficients C1 and C2 can be determined using
the initial conditions and the property of orthogonality of the mode shapes. These formulas are

Iv
C1  I w cos nt0  sin nt0
n
Iv
C2  I w sin nt0  cos nt0
n

with

L2

 w
 L1
0 n   2 0 n dx
Iw  L2

   2 n2 dx
2
n
 L1

19
 L2

 v
 L1
0 n   20 n dx
Iv  L2

   2 n2 dx
2
n
 L1

The parameters t0 , w0 , q 0 , v0 , w 0 are the dimensionless initial time, transverse displacement,

cross-section rotation, transverse velocity, and cross-section rotational velocity respectively, and
f n is the nth dimensionless translational mode shape for the pole, and y n is the nth rotational mode
shape for the pole

For the case of rigid body motion (   0 ), the time variation of the generalized coordinate is

modified, as are the coefficients C1 and C2 . In this case the mode shapes  ,  are  x,1 for

rotational motion, and the generalized time function will be

qn  C1t  C2

with

C1  I v
C2  I w  I vt0

2.4.9 Upper bound of the initial impact force

Figure 2 shows a schematic of the deflected shape of the system for a very short time after
first impact. The time is so short that the effective pole length, 2cs t * , is so small that the

*
deformation is dominated by shear. The shear strain in the pole is   , in which  is the
*
*
cs t
displacement of the pole at the impact point relative to the as-yet unaffected centerline.
Accordingly, the force required to cause the shear deformation is

*
F *  2 GA  2 GA (2.23)
cs t *

20
 𝛥∗

𝑣0∗

𝑐𝑠 𝑡 ∗
𝑣0∗ 𝑡 ∗ 𝐹𝑠∗

Figure 2. Impact zone at time t *

The force in the contact spring is

Fs*  k *  *  v0t * 

which can be rearranged to

Fs*
*  *
 v0t * (2.24)
k

Substituting Eq. (2.24) into Eq. (2.23) gives

F *cs t * Fs*
 *  v0t * (2.25)
2 GA k

From equilibrium of forces at the impact point,

F *   Fs* (2.26)

Substituting Eq. (2.26) into Eq. (2.25) gives

 c t* 1
F *  s  *   v0t *
 2 GA k 

21
and re-arranging terms results in

v0t *
F 
*
(2.27)
 cs t * 1
  *
 2 GA k 

By taking the limit of Eq. (2.27) as k   , an upper bound of the initial impact force is
*

2 GAv0
F*   2  G  Av0  2 ksh mv0 (2.28)
cs

in which ksh   GA L and m   AL* . Note that Eq. (2.28) is exact for rigid stops.
*

2.4.10 Impact duration

The impact duration for the first impact event is td  L / cs (impact at the center), which is
*

the time the shear wave in the pole takes to return to the impact point from the free end.

2.4.11 Peak Force using Euler-Bernoulli beam theory

The (time) Laplace transform provides a convenient approach to obtain the limiting impact
force for an Euler-Bernoulli beam. Indeed, using the initial value theorem, a simple calculation
shows that

d 3W
F  lim s 3  lim
 
2 2 sv0 ei 2l2 s sin 2l2 s  sinh   2l2 s   
 s (2.29)
e2 1 l2 s  e2 1 l2 s  4ei 2l2 s  e2i
1/4 3/4
s  s 
dx 2l2 s
1

where s is the independent variable in the transformed space. Thus, as t  0 , i.e. s   , the
instantaneous impact force is unbounded according to Euler-Bernoulli beam theory. This result is
consistent with Eq. (2.28) by letting the shear stiffness become large, given that Euler theory
corresponds to infinite shear stiffness.

2.4.12 Mechanical Energy

The energy exchange during impact provides insight on the mechanics of impact force
generation and development. The nondimensional mechanical energy of the Timoshenko beam
can be written as

22
 H  Kw  K  V  V  Vk (2.30)

where

1  w  1  w 
0 2 2 L 2

Kw     dx     dx (2.31)
 L1
2  t  0
2  t 

 2     2   
0 2 L2 2

K  
 L1
  dx   
2  t  0
 dx
2  t 
(2.32)

0 2 L
1 1
V    2 dx    2 dx (2.33)
 L1
2 0
2

0 2 L
1 2 1
V    B dx    B2 dx (2.34)
 L1
2 0
2

1 2
Vk  kw  0, t  (2.35)
2

w 
   is the section shear strain and  B  is the beam curvature. This mechanical
x x
energy is conserved during the beam motion and the analysis of the exchange of the energy
among its components determines the impact-force history.

The exchange of energy is locally dependent on the evolution of the energy density for each
kinetic and potential energy component. The governing equations for each energy density can be
obtained directly from the model equations and kinetic considerations. Accordingly, let us
consider the following form of the Timoshenko model:

 1 
  kw (2.36)
t 2 x

  B 
2   (2.37)
t x 

w 
where   and   denote the transverse velocity and the section angular velocity,
t t

23
respectively. Also, taking time derivatives of the kinematic relations defining the shear strain and
the curvature, and assuming sufficient regularity of the solution to allow for the interchanging of
time and space derivatives, we obtain

 
  (2.38)
t x

 B 
 (2.39)
t x

The governing equation for the energy density for each component can be derived by
multiplying each of Eqs. (2.36) - (2.39) by its evolving variable. For instance, to obtain the
1
governing equation for  w   2 , we multiply Eq. (2.36) by  to obtain
2

 w   Vk
   (2.40)
t  x t

Similarly, we have

  
2  B  (2.41)
t x 

   
  (2.42)
t  x 

 
 B (2.43)
t x

1 1 1
in which   2 ,    2 and    B2 . It follows that at the impact point, the spring
2 2 2
energy changes at a rate proportional to the jump in the shear power, i.e.,

dVk  (0, t )
  (0, t )   (0, t ) (2.44)
dt 

The energy exchange between the other components is more complex. Away from the origin,
 w exchanges energy with  through the local shear power term

24
 1     
    (2.45)
 x  x  x

1 
This term,   , represents a local source of power from the shear strain which, when
 x
integrated, accounts for the power being generated at the beam ends. Besides this exchange with
 w , the shear energy density  also directly trades local specific energy with the rotational

kinetic energy  through .


Finally, in addition to the latter exchange,  trades energy with the potential energy density

 due to bending through the rotational local source

1   
  B    B   B (2.46)
 x x x

An analogous manipulation of the Euler-Bernoulli model equations yields the following

equations for the kinetic and potential energy density components:

 wE  2 E dVk
    (2.47)
t 2 x 2 dt

E  2
 E 2 (2.48)
t x

1 2 E 1 2 2w
in which    ,    E , and  E  2 . Moreover, we have
E

x
w
2 2

dVk    
  (0, t )  E |(0 ,t )  E |(0 ,t )  (2.49)
dt  x x 

2.5 Results
As stated previously, the primary motivation for this work is to determine the force from
debris impact. It is interesting to investigate how the impact force varies with practical values of
the dimensionless parameters. Regardless of where along the beam impact occurs the impact

25
force time history is the same until the waves have time to return from the end of the shorter
segment. In addition, the force time history is independent of the initial angular velocity because
the impact duration is very small and the beam rotation is negligible within the impact duration.
As a result, we focus on impact at midspan, which would seem to lead to longer contact
durations, and on zero initial angular velocity.

The pole has a weight of 342 kg (754 lb), length of 9.14 m (30 ft), and a diameter of 0.305 m
(1 ft). This is somewhat less than the prototypical 1,000 lb wood pole suggested by a common
U.S. design guideline (ASCE 2010). Therefore, for the parameter study the beam has a
slenderness ratio B = 120,  = 1, and a shear coefficient  = 0.9, which is commonly used for a
solid circular section (although more accurate estimates are available (Cowper 1966)). For an
isotropic material,   2 1    /  , in which  is the Poisson ratio. However, wood is not an

isotropic material, and wood properties vary widely. We adopt here a value of  = 10 for our
prototypical beam, which is at the low end of values suggested in Yoshihara et al. (1998).

Note that the dimensional force at time zero for rigid impact is F*  2  G Av0* . Based on G
= 0.89 GPa (129 ksi) given in Yoshihara et al. (1998) for spruce, the estimated rigid impact force
for our prototypical beam impacting at a physically reasonable 10 km/h (6.2 mph) is F*  300 kN
(67,000 lb) with an approximate duration of 8 ms (milliseconnds).

Unless otherwise stated below, Timoshenko beam theory is used. Results for Euler-Bernoulli
beam theory will be compared subsequently. Also, in the following, for convenience, the impact
force is non-dimensionalized by the theoretical impact force based on rigid impact given by Eq.
(2.28), F * / F* , time is nondimensionalized as t  t * / td*  t *cs / L* and nondimensional space is

defined as x  x / L . For impact at midspan of the beam, td*  L* / cs , which is the time it takes
* *

a shear wave to propagate to the free end and back to the point of impact. It is a measure of the
approximate duration of impact.

2.5.1 Effect of spring stiffness

First, we consider a range of 𝑘̅ = 𝑘 ∗ /(𝜅𝐺𝐴/𝐿∗ ), which is the ratio of the spring stiffness 𝑘 ∗
to a measure of the shear stiffness of the beam. A convergence study was carried out to

26
determine what value of 𝑘̅ is sufficiently close to rigid impact. As shown in Figure 3, there is
only a modest difference between 𝑘̅ = 1200 and 1800, and 𝑘̅ = 1200 is chosen to represent
rigid impact for the current study. However, even 𝑘̅ = 600 is quite stiff, and the primary
difference is a slightly lower initial impact force and a slightly longer rise time. We note that
contact durations are all less than 1 for this range of 𝑘̅.

1.2

1
Nondimensional Force

0.8
k = 1800
0.6 k = 1200
k = 600

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
Nondimensional time

(a) Impact force over entire duration

27
 1.2

Nondimensional Force
0.8

0.6

0.4
k = 1800
k = 1200
0.2 k = 600

0
0 0.02 0.04 0.06 0.08 0.1
Nondimensional time

(b) Close up of initial peak impact force

Figure 3: Impact force for approximately "rigid" impact with 𝜏 = 10, 𝜂2 = 1, 𝜆𝑏 = 120

The value of k  1 is taken as the smallest stiffness considered. Figure 4 and Figure 5 show

the impact force for different values of k and two values of  . (As explained subsequently,  =
20/9 represents a reasonable lower bound.) For very soft springs, the response resembles the

response of a single-degree-of-freedom oscillator. However, this quickly changes as k increases.

For all cases with k  10 , the response is qualitatively similar, with the initial peak impact force

increasing toward the limit of 1 as k increases. The second peak in the force also increases, and
can exceed one for large spring stiffnesses and   10 . There are also high frequency oscillations
that start near the end of contact.

28
 1.2

k=1
1
k=5
Nondimensional force

k = 10
0.8
k = 20

0.6 k = 100
k = 600

0.4 k = 1200

0.2

0
0 0.5 1 1.5 2 2.5
Nondimensional time

Figure 4: Impact force for a range of 𝑘̅, with 𝜂2 = 1, 𝜆𝐵 = 120 and 𝜏 = 10

0.8
k=1
Nondimensional force

k=5

0.6 k = 10
k = 20
k = 100
0.4
k = 600
k = 1200
0.2

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Nondimensional time

Figure 5: Impact force for a range of 𝑘̅, with 𝜂2 = 1, 𝜆𝐵 = 120 and 𝜏 = 20/9

For stiff springs it can be seen that the force drops off quickly from the initial peak. The
initial response is dominated by localized shear behavior around the point of impact, and the

29
force decreases as flexure begins to occur and strain energy is converted from shear strain energy
to bending strain energy.

2.5.2 Effect of 𝜏
As mentioned previously, for an isotropic beam 𝜏 = 2(1 + 𝜈)/𝜅. It is interesting to see how
the impact force varies based on different Poisson’s ratio and κ. For 0 ≤ 𝜈 ≤ 0.5 and κ = 0.5
(circular or square tube), 5/6 (rectangular cross section) and 0.9 (solid circular cross section),
values 𝜏 = 20/9, 3, 4, 5, and 6 are considered. The value 20/9 corresponds to ν = 0 and 𝜅 = 0.9,
and therefore represents a reasonable lower bound. The value 6 corresponds to 𝜈 = 0.5 and
𝜅 = 0.5, and therefore represents a reasonable upper bound for an isotropic beam of nearly
incompressible material. The impact force time histories are shown in Figure 6. The dependence
of the impact force on τ does not appear to be as strong as the dependence on spring stiffness. A
consistent trend is that the initial peak (nondimensional) impact force increases with increasing τ.
However, even for the range considered, the minimum peak is still greater than 0.88. At the end
of contact, the variation is larger. In general, the second peak increases with increasing τ,
although not consistently. For example, the second peak for 𝜏 = 5 is about 0.9, while the second
peak for 𝜏 = 6 is only about 0.4.

30
 1.2


1 


Nondimensional Force

0.8 


0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
Nondimensional time

Figure 6: Impact force for a range of 𝜏, with 𝑘̅ = 1200, 𝜂2 = 1, 𝜆𝐵 = 120

2.5.3 Effect of slenderness ratio 𝜆𝐵

The effect of slenderness ratio is shown in Figure 7 for nearly rigid impact where 𝑘̅ = 1200
and Figure 8 for a soft spring where 𝑘̅ = 10. One interpretation for these plots is to assume that
L* remains constant, so that the horizontal axis represents the same point in time for all cases,
and let r vary. It can be seen then that as the radius of gyration increases (bending stiffness
increases relative to shear stiffness), the importance of the shear deformation increases and the
drop in impact force is slower.

31
 1.2

1   
  
Nondimensional force

0.8
 
 
0.6
 

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
Nondimensional time

Figure 7: Impact force for a range of 𝜆𝐵 , with 𝑘̅ = 1200, 𝜂2 = 1, 𝜏 = 10

0.6

0.5 

Nondimensional force

0.4 

0.3 

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1
Nondimensional time

Figure 8: Impact force for a range of 𝜆𝐵 , with 𝑘̅ = 10, 𝜂2 = 1, 𝜏 = 10

32
 2.5.4 Effect of 𝜂
The effect of mass moment of inertia is shown in Figure 9. As can be seen, the moment of
inertia has a negligible effect on the impact force for the soft spring case. In the case of nearly
rigid impact, its effect on the whole trend of the impact force-time curve is negligible except for
some oscillations near the end of the impact. This conclusion is interesting, because some
structural finite element codes use a lumped mass matrix that includes translational masses but
not rotational inertias.

1.2


k = 1200,  = 1
1

k = 1200,  = 0.01

Nondimensional Force

k = 10,  = 1
0.8

k = 10,  = 0.01

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
Nondimensional time

Figure 9: Impact force for 𝐿⁄𝑟 = 120, 𝜏 = 10 with soft and stiff springs and 𝜂2 = 1 and 𝜂2 =
0.01

2.5.5 Asymmetric impact of rotating beam

The focus thus far has been on the impact at midspan, which would seem to lead to longer
contact durations, and on zero initial angular velocity. Figure 10 shows the impact force history
for an asymmetric impact at quarter span for different initial angular velocities. As before, the
impact force in this plot is nondimensionalized by the force for the impact on a rigid structure
(Eq. (2.28)), but here using the velocity at the impact point when the beam is rotating. The plot

33
shows that the initial impact dynamics is dominated by the vibrational motion in the immediate
neighborhood of the impact. So regardless of where along the beam impact occurs, the impact
force-time history is the same until the waves have time to return from the end of the shorter
segment. In addition, the time history of the impact force is independent of the initial angular
velocity because the impact duration is very small and the variation in the beam rotation is
negligible within the impact duration. The main effect of the asymmetric impact is in lowering
the second peak of the impact force because of the asynchrony of returning waves. Although not
shown, it should be remarked that the vibrational motion of the beam depends on the position of
the impact as well as on the initial rotation. For instance, the dependence of vibration history on
the impact location is evident in the dispersed waves: the second force peak is broader and is
preceded by ripples of lower wave number for asymmetric impact when compared with the
symmetric impact.

Figure 10: Impact force for impact at quarter span, 𝜆𝐵 = 120, 𝜏 = 10, 𝜂2 = 1 𝑎𝑛𝑑 𝑘̅ = 1200
(All plots coincide)

34
 2.5.6 Euler vs. Timoshenko
The previous results are for Timoshenko beam theory. It is of interest to determine for which
parameter regimes Euler-Bernoulli beam theory gives results similar to Timoshenko beam theory.
Qualitatively, we would expect that for a large slenderness ratio such as B  120 , Euler would
give reasonably good results. However, as shown previously, as the spring stiffness increases,
the Euler impact force grows without bounds. Figure 12 shows the nondimensional time-force

relations for different values of k . It can be seen that for a soft spring stiffness in which k  1 ,
Euler-Bernoulli beam theory gives close results to Timoshenko beam theory, because the shear
force is relatively small and the impact is absorbed by the bending action of the beam and the

spring. On the other hand, for large k , the impact force increases unboundedly using Euler-

Bernoulli beam theory. For instance, for the case of k  1200 , the force calculated using the
Euler-Bernoulli beam theory is almost three times the force calculated using the Timoshenko
beam theory, and the duration is much shorter. The impact force is larger in the Euler-Bernoulli
beam model because the main mechanism of energy transfer at the start of impact, namely, the
transfer of energy from the translational kinetic energy to shear strain potential energy, is lacking
in the Euler model. This shear strain is replaced by an unphysical flexural strain in the Euler-
Bernoulli model that over predicts the impact force. The shorter duration, however, is due to
differences in dispersion at short and long time scales and in particular it can affect the duration
of the impact.

As it is well known (see e.g. H.Kolsky (1963), Y.C.Fung (1965) and S.H.Crandall (1968)),
the group speed increases without bound in the Euler-Bernoulli model, whereas the Timoshenko
model predicts two different branches for the group speed. As the number increases, one of the
modes asymptotically converges toward the shear speed, while the second mode converges
toward the longitudinal wave speed (for convenience, we plot the group speeds for the different
models in Figure 11). Thus, during impact against a stiff spring, a broadband wave whose wave
groups at the high wave-number end of the spectrum can move much faster in the Euler-
Bernoulli model than in the Timoshenko beam. These distinct dispersion characteristics of the
two beam models explain why the nondimensional time duration is close to one for the impact on
a stiff spring for the Timoshenko model, and why the impact ends prematurely in the Euler-

35
Bernoulli model under the same conditions. Indeed, because of the unbounded nature of the
group speed in the Euler-Bernoulli model, the trend in this model is for decreasingly shorter
duration as stiffness of the spring increases.

Figure 11: Branches of dimensionless group speed 𝑐̅𝑔 = 𝑐𝑔∗ /𝑐𝑠 as a function of the dimensionless
wave number 𝜉 = 𝜉 ∗ 𝑟 for 𝜏 = 10

36
 0.3

Timoshenko
0.25
Euler

Nondimensional force
0.2

0.15

0.1

0.05

0
0 0.5 1 1.5 2 2.5
Nondimensional time

(a) ̅𝑘 = 1

37
 0.4
Timoshenko
Euler

Nondimensional force 0.3

0.2

0.1

0
0 0.5 1 1.5 2
Nondimensional time

b) ̅𝑘 = 5

38
 0.6

0.5

Timoshenko
0.4
Nondimensional force

Euler

0.3

0.2

0.1

0
0 0.2 0.4 0.6 0.8 1 1.2

Nondimensional time

c) ̅𝑘 = 10

39
 1.2

Timoshenko
0.8
Nondimensional force
Euler

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
Nondimensional time

d) ̅𝑘 = 100

40
 3

2.5

Nondimensional force
2

Timoshenko
1.5 Euler

0.5

0
0 0.2 0.4 0.6 0.8 1
Nondimensional time

e) ̅𝑘 = 1200

Figure 12: Comparison of impact forces between Timoshenko and Euler-Bernoulli beam theories
for different ̅𝑘, with 𝜆𝐵 = 120, 𝜏 = 10, 𝜂2 = 1

2.5.7 Energy breakdown and beam vibration

The energy exchange gives an insight on the beam vibration and in particular into generation
and behavior of the impact force. Figure 13 shows the time evolution for the potential and kinetic
energy components. It should be noted that the results are for a stiff spring, where the maximum
impact force is dominated by the shear wave and the spring energy is very small: theoretically,
for rigid impact, the spring energy vanishes. The maximum impact force occurs when the spring
is at maximum compression and the speed of the point of contact is zero. The initial peak in the
force is dominated by shear and after the initial impact the force is dominated by the exchange
between translational kinetic energy and potential bending energy. Indeed, without rotation or
pre-strain, all initial energy is translational kinetic energy. Locally, the kinetic energy density,
and by integration the kinetic energy, exchange energy with the spring and shear potential energy

41
(Eq. (2.40) and Eq. (2.41)). Thus, initially, the beam’s initial energy is transformed into potential
shear energy and potential spring energy. This explains the sharp rise in the shear and spring
energies and the dominance of shear at the initial impact. Potential bending energy, on the other
hand, exchanges energy with kinetic rotational energy and depends on the build-up of curvature
and rotation (Eq. (2.42) and Eq. (2.43)). As such it is slower to respond to the impact; see Figure
13b. In the Euler-Bernoulli model, the absence of the neglected shear strain causes the kinetic
energy to exchange energy directly with flexural strain (Eq.(2.47)). The results show an
unrealistic increase in the impact peak force; see Figure 12.

0.1
Nondimensional Energy

0.01

0.001

0.0001
Total
-5 Kinetic-w
10
Kinetic-
-6 Bending
10 Shear
Spring
-7
10
0 0.2 0.4 0.6 0.8 1
Nondimensional time

(a) All energy components over the entire impact duration

42
 0.1

0.01

Nondimensional Energy
0.001

0.0001
Kinetic-w
-5 Kinetic-
10
Bending
-6 Shear
10
Spring
-7
10
0 0.02 0.04 0.06 0.08 0.1
Nondimensional time

(b) Close up of all energy components around the initial peak force

Figure 13: Time evolution of potential and kinetic energy components (log scale) for 𝜆𝐵 =
120, 𝜏 = 10, 𝜂2 = 1 and 𝑘̅ = 1200

Figure 14 shows the potential energy densities as a function of space and time. It is apparent
that at the time of impact, there is a large spike in the shear strain energy as the impact is
absorbed via shearing and spring deformation. This is then quickly transferred to bending strain
energy around the impact as curvature is established and the Euler-Bernoulli hypothesis of zero
shear for low wave number is approached. The shear wave propagates down the bar until it is
reflected at around td* / 2  L* / 2cs ( t  1/ 2 ). Of interest is the ensuing motion after reflection,
which displays two modes of dispersion: one corresponding to the flexural mode and traveling at

group speed cg  1 and a second, shear, quicker mode, at group speed cg   . The second

mode is not as distinct as the dispersion of the flexural waves that is noticeable in Figure 14, but
is clearly apparent in the dispersion of the rotational kinetic energy in Figure 15. Also evident in
this figure is a peak in the rotational kinetic energy, or “whipping” effect, as the latter is reflected.
The high frequency oscillation at the center of the beam near the end of impact explains the high

43
harmonics of the impact force before separation. Note that this second mode of dispersion is
fully absent in the Euler-Bernoulli model; see Figure 12.

(a) Shear energy density

(b) Bending energy density

Figure 14: Space-time plot of potential energy density for 𝜆𝐵 = 120, 𝜏 = 10, 𝜂2 = 1 and 𝑘̅ =
1200

44
 (a) Translational kinetic energy density

(b) Rotational kinetic energy density

Figure 15: Space-time plot of kinetic energy density for B  120 ,   10 ,   1 , k  1200
2

2.6 Summary
An analytical solution for the linear response of a Timoshenko beam impacting a stop
modeled as a spring has been presented. The solution, which is based on modal superposition,

45
admits contact at an arbitrary point within the beam span, as well as beam rotational as well as
translational initial velocities. As such, the present solution is a generalization as compared to
previous solutions. It has been shown that the initial impact is dominated by shear behavior, and
that for stiff stops, Euler-Bernoulli beam theory can significantly overestimate the initial impact
force and underestimate the contact duration. Indeed, based on Euler-Bernoulli beam theory the
impact force grows without bounds as the stop stiffness increases, whereas Timoshenko beam
theory shows the impact force is bounded. As a general rule, it is recommended that Timoshenko
beam theory rather than Euler theory be used for impact studies. Parameter studies have shown
that the impact force time history is controlled primarily by the ratio of the stop stiffness to the
beam shear stiffness. For homogeneous beams, the sectional rotational inertia has little effect on
the impact forces.

46
 Chapter 3
Beam Response to Longitudinal Impact by a Pole

3.1 Introduction
This chapter investigates a pole striking a column longitudinally. The model involves a beam
governed by Timoshenko beam theory and a pole governed by the one-dimensional wave
equation. The discussion is restricted to elastic impact and linear response because of the low
speeds involved in debris impact. For the case of a free-free beam hitting a stop, the beam, which
has initial translational and rotational velocities, can strike the stop anywhere along the beam.
The stop is flexible and it is modeled as a rod with wave propagation. For the case of axial
impact of a pole against a column, any combination of column end conditions are allowed, and
again contact can occur anywhere along the column. Although the pole-column impact is of
primary interest, the model is also valid for the case of a free-free beam, which has initial
translational and rotational velocities, hitting a stop anywhere along the beam. The stop is
flexible and it is modeled as a rod with wave propagation.

The performance of Euler beam theory is again investigated in this chapter. In addition, the
effect on the impact force of ignoring the rotary inertia in Timoshenko beam theory is
investigated; this has practical significance because some finite element codes include shear
deformation but ignore rotary inertia. Finally, a detailed discussion of the transfer of energy
between kinetic energies and the potential (strain) energies is provided. The analysis of energy
exchange examines the contributions of the different energy components to the impact dynamics.

This chapter is organized as follows. First, the governing equations and the chosen
nondimensionalization of the variables are presented. Then the analytical solutions, which are
based on superposition of the normal modes of vibration, are developed. Results of a study on
impact response for parameters within a range that might be seen in debris impact are presented
and interpreted. The final section contains conclusions from the present study.

47
 3.2 Physical System
A schematic of the system is shown in Figure 16. The figure shows a pole with an initial
velocity striking a column. A spring between the pole and the column represents a local contact
stiffness, k s . In this scenario, the column would likely have displacement boundary conditions
(e.g., fixed-fixed or pinned-pinned) and would have zero initial velocity. In a second scenario,
the beam could be traveling with an initial velocity v0 (at x1  0 ) and angular rotation 0 and
impacts a ‘pole’ or flexible ‘stop’. The pole may have a zero-displacement boundary condition at
the right end. In both cases, the origin of the coordinates x1  x2 is located at the point of impact.
For generality of the discussion, the beam in the second scenario also will be referred to herein as
a column, unless the discussion is specific to a translating beam.

The column and the pole are assumed to be elastic and homogeneous, with constant cross-
sectional properties. The pole has Young’s modulus E p ; sectional area Ap ; mass density  p ;

and length L p . The column is modeled by Timoshenko beam theory. It has Young’s modulus E ;

shear modulus G ; mass density ; sectional area moment of inertia I ; sectional area A ;
sectional rotary inertia I m  2  I , where  2 is defined as the ratio of I m to  I ; shear

coefficient  ; and total length L  L1  L2 .

For these systems, the fundamental kinematic variables, which are functions of time t, are the
axial displacement in the pole, u  x2 , t  , and the transverse displacement w  x1 , t  and cross-

sectional rotation   x1 , t  of the column. The latter is taken as clockwise positive.

It is convenient to work with dimensionless variables. For Timoshenko beam theory it is

c  E 
common to non-dimensionalize the length scale by r  I A and time by c0 r , where 0
(Boley and Chao 1955; Xing et al. 2002). Adopting this for the entire system results in the
following non-dimensional variables:

x1 w x u c
x1  ; w  ; x2  2 ; u  ; and t  t 0 (3.1)
r r r r r

48
Additional non-dimensional variables include

Li Lp k E c02 c0 p
li  , i = 1,2; l p  ; k  s ;   2;  (3.2)
r r EA r  G cs c0

cs   G 
in which and c0 p  E p  p .

Figure 16: Schematic of beam/column and pole

3.3 Equation of Motion

The equations of motion for the column are

1   2 w   2 w
    k 
 w  0, t   u  0, t  
   0 (3.3)
  x12 x1  t 2

49
  2 1  w  2  
2
       0 (3.4)
x12   x1  t 2

 w 
 in Eq. (3.3) is the Dirac delta. Note that the positive definition of shear strain,       , is
 x1 
such that positive shear force acts in the +x2 direction on a positive x1 face. To deal with the
discontinuity in shear at x1  0 , let w  w for x1  0 and w  w for x1  0 , and similarly for  .

The equation of motion for the pole is the one-dimensional wave equation:

 2u 1  2u
 0 (3.5)
x22  2  t 2

As mentioned, a number of different boundary conditions are possible. For the column these
are either zero displacement or shear and zero rotation or moment at x1  l1 and x1  l2 , and for

the pole they are zero displacement or zero force at x2  l p . Initial displacements are zero and

options for initial velocities are as described earlier.

The compatibility equations at x1  0 are equal displacement, rotation, and curvature in the
column, as well as the jump condition for shear

 w   w 
           k  w  u   0 (3.6)
 x1   x1 

The compatibility between the pole and the column results in

u 1
 k w  u   0 (3.7)
x2  m  2

in which  m   p Ap  A .

50
 3.4 Analytical Solution

3.4.1 Mode Shapes

A standard approach of separation of variables is adopted. The column mode shapes on each
side of the discontinuity at x1  0 are the usual Timoshenko beam mode shapes, i.e.,

 
 
  A1 cos  x1  A2 sin  x1  A3 e x1  A4 e  x1 


     x1   x1 
(3.8)
  ( A1 sin  x1  A2 cos  x1 )   ( A3 e  A4 e ) 
 
 
   

with

 2 ( 2   )    2 ( 2   )2  4  2 ( 2   )    2 ( 2   )2  4
 ,
2 2

   ( 2   2 ) /  ,   ( 2   2 ) / 

and  is the nondimensional frequency. Note that  is always real, but  may be real or
imaginary. Two special mode shapes must be considered. The first are the rigid body modes (if
the boundary conditions allow such) when w = 0 . The second is the special case of   1   
, a ® 0, in which the mode shapes can be obtained by taking the limit of Eq. (3.8). These
eigenfunctions are consistent with Xing et al. (2002), who considered only the case  = 1. For
the pole, the modes shapes are  p  C1 cos  x2  C2 sin  x2 where     .

3.4.2 Characteristic Equation

The natural frequencies and the coefficients of the mode shapes are determined by imposing
the five compatibility conditions and five boundary conditions. The ten equations result in a
system of equations that determines the natural frequencies  . For the case of a free-free column
and fixed end pole, the characteristic equation is obtained from the determinant:

51
 0 0 0 0 a15 a16 a17 a18 0 0

a21 a22 a23 a24 0 0 0 0 0 0

0 0 0 0 a35 a36 a37 a38 0 0

a41 a42 a43 a44 0 0 0 0 0 0

1 0 1 1 1 0 1 1 0 0
0
0      0     0 0

  0        0     0 0

k k k k k k
          k 0
2 2 2 2 2 2

0 0 0 0 0 0 0 0 cos  l p sin  l p

k k k k k k k
0 0  
2  m
2
2  m
2
2  m2
2  m
2
2  m2
2  m
2
 m
2

with a15        sin  l1 ; a16        cos  l1 ; a17      exp   l1  ;

a18       exp  l1  ; a21         sin  l2 ; a22        cos  l2 ; a23      exp  l2 

; a24       exp   l2  ; a35    cos  l1 ; a36    sin  l1 ; a37   exp   l1  ;

a38   exp  l1  ; a41    cos  l2 ; a42    sin  l2 ; a43   exp  l2  ; and

a44   exp   l2  . For other boundary conditions, rows 1, 2, 3, 4, and 9 in the determinant

must be modified accordingly. As noted earlier,  can be either real or complex. In the latter
case, the characteristic equation is also complex.

Mathematica (Wolfram 2010) was used to obtain the natural frequencies herein. Starting
from zero, a very small increment in  was used to search for a segment in which a root existed,
and then Newton-Raphson was used to find the root. As an example, an increment of 1/100,000

52
was used to find 1000 modes. It was observed that if a mode is missed, the force-time curve is
not smooth and local peaks and valleys appear along the curve, and if insufficient modes are
used, sharp spikes appear around the initial peak of the force.

Once the natural frequencies and modes are determined, the response for initial velocity
conditions is obtained as

 w  x1 , t    n   x1  
     
  x1 , t     Bn  n  x1   sin n t (3.9)
   
 u  x2 , t    pn  x2  

given that
l2 lp

 n v0    n0 dx1   m   pn v0 p dx2

2

 l1
Bn 
0
(3.10)
 l2 lp

n   n    n dx1   m   pn
2 2 2 2
dx2 
 l 
 1 0 

in which v0 , 0 , and v0 p are the initial translational velocity of the column, the initial rotational

velocity of the column, and the initial translational velocity of the pole, as appropriate. For
example, a stationary column can be hit by a translating pole, or a fixed or free pole can be hit by
a translating and rotating beam (except only one dimensional motion of the pole is considered).

Note that the above equations related to the column can be easily modified for a simplified
model of a beam hitting a massless stop, i.e., a massless spring, by dropping the terms containing
u and p .

3.4.3 Initial Peak Impact Force

The initial peak impact force can be estimated as follows. Figure 17 shows a schematic of the
deflected shape of the system at time t (t << 1) after impact. The time is sufficiently small such
that the effective beam length, 2cs t , is so small that the deformation is dominated by shear. The

shear strain in the column is   2 cs t , in which  2 is the displacement of the column at the
impact point relative to the as-yet unaffected centerline. Accordingly, the force required to cause

53
the shear deformation is Fc  2 GA . Similarly, the effective length of the pole is c0 p t , and the

strain is  p   1  v0t  c0 pt , in which 1 is the displacement of the impacted end of the pole. The

corresponding force is Fp  E p Ap p . The force in the contact spring is just Fs  ks  2  1  .

Equilibrium of the massless spring requires Fc  Fp   Fs , resulting in the impact force

FcR FpR
Fc  Fp  (3.11)
F F cc
FcR  Fpr  cR 2pR s 0 p
v0 ks t

in which FpR   E p Ap c0 p  v0 and FcR   2 GA cs  v0 .

Δ2

cop t cst c

-Fs
vo t Δ1

Figure 17: Impact zone at time t

Two known special cases can be recovered easily from Eq. (3.11). The instantaneous impact
force of a pole hitting a rigid column, Fc  FpR , is obtained from Eq.(3.11) by taking the limit as

ks   and k GA® ¥ . This force can also be written as (Paczkowski et al. 2012)

FpR  k p mp v0 (3.12)

54
in which k p  E p Ap Lp and mp   p Ap Lp . The practical applicability of Eq.(3.12) for debris

impact has been shown based on full-scale tests of logs and shipping containers impacting a stiff
target (Piran Aghl et al. 2014). Similarly, the instantaneous impact force of a beam hitting a rigid
stop, Fc  FcR (Riggs et al. 2013; Xing et al. 2002) is obtained from the limit as ks   and
E p Ap   . This force can also be written as

FcR  2 ksh mv0 (3.13)

in which ksh   GA L and m   AL . Eqs. (3.12) and (3.13) are exact for rigid stops; note also the
similarity in their forms. These forms involving stiffness and total mass may be most convenient
for structural designers.

Because of uncertainty in the local contact stiffness, in this study ks is taken to be very large.
Thus, ks can be interpreted as a penalty parameter to couple the beam and the column. The large
but finite stiffness ks produces a steep ramp at impact with slope proportional to its magnitude,
instead of a jump in the force associated with an impact with infinite ks. This ramp is not only
physically more realistic, but it also prevents the occurrence of the Gibbs phenomenon that
would otherwise occur with a modal superposition solution for an infinite ks.

The limit of Eq. (3.11) as ks   results in the instantaneous peak impact force

FcR FpR
F0  (3.14)
FcR  Fpr

which can also be written as

1 2
F0  k p m p vo  ksh mvo (3.15)
1 2

in which 1  1   m   2 , and  2  1  2  m    . In the results section, the impact forces are

nondimensionalized by the values predicted by Eq.(3.15), which will demonstrate its

applicability. Note that if 1 is close to 1, the peak impact force is dominated by the axial

response of the pole, whereas when  2 is close to 1, the force is dominated by the shear response

55
of the column.

Eq. (3.14) also applies to the impact of two orthogonal beams if FpR is replaced by FcR of the

second beam. Eq.(3.14) has the same form as the formula for longitudinal impact of two bars,
although the derivation here is quite different (Graff 1991; Harrison and Nettleton 1997).

3.4.4 Mechanical Energy

The energy exchange during impact provides insight on the mechanics of impact force
generation and development. To simplify notation,   w t and    t are the transverse

velocity and the section angular velocity of the column, respectively;  p  u t is the velocity in

the pole;  B   x is the beam curvature, and  p  u x is the pole strain. The

nondimensional energy densities (per unit length) are defined as follows: w   2 / 2 (column

translational kinetic);    22 / 2 (rotational kinetic); u   m v2p / 2 (pole translational kinetic);

   2 / 2 (shear); k   B2 / 2 (bending) and u   m  2 p2 / 2 (pole strain). The spatial integration

of the energy densities results in the total energy for each component. The sum of these six
components plus the potential energy in the contact spring results in the total nondimensional
mechanical energy. Note that the energies are nondimensionalized by (EAr). The model does not
include any energy dissipation mechanisms and therefore the mechanical energy is conserved
during impact.

The exchange of energy is locally dependent on the evolution of the energy density for each
kinetic and potential energy component. The governing equations for each energy density can be
obtained directly from the model equations and kinematic considerations. For that purpose,
consider the following form of the Timoshenko model:

 1    B 
  k  w  u   and  2   (3.16)
t  x t x 

Also, taking time derivatives of the kinematic relations for the shear strain and the curvature, and
assuming sufficient regularity of the solution to allow for the interchanging of time and space
derivatives, one obtains

56
    B 
   and  (3.17)
 t x t x

The governing equation for the energy density for each component can be derived by multiplying
each of Eqs. (3.16)–(3.17) by its evolving variable. For instance, to obtain the governing
equation for the translational kinetic energy density w , multiply the first of Eq. (3.16) by  to
obtain

 w v 
  k ( w  u )v (3.18)
t  x

Similarly,

     u  p
 B   v p m  2 (3.19)
t x  t x

       k 
   B (3.20)
t  x  t x

 u  p
  m  2 p (3.21)
t x

From equilibrium at the contact point, k  w(0, t )  u (0, t )     0 , t     0 , t  , and it
1

follows that the spring energy changes at a rate proportional to the jump in the shear power:

dVk  (0, t)  p (0, t)

   0 , t     0 , t  (3.22)
dt   

The energy exchange between the other components is more complex. Away from the origin,

w exchanges energy with  through the local shear power term

 w   1    1    dVk
  ( )   p     
t t  x  t  x   dt

1 
The term ( ) represents a local source of power from the shear strain. When integrated,
 x
this term accounts for the power generated at the beam ends. Besides this exchange with w , the

57
shear energy density  also directly trades local specific energy with the rotational kinetic

energy  through   . Finally, in addition to the latter exchange,  trades energy with the

potential-energy density  from bending through the rotational local source

1   
  B    B   B
 x x x

3.5 Results

3.5.1 Parameter Range

As stated previously, the primary motivation for this work is to determine the force from
debris impact. The ranges for the nondimensional parameters investigated are based on possible
values for debris impact. A common U.S. design guideline (ASCE 2010) suggests that a
prototypical debris object during flooding may be a 1,000 lb log. Wood properties vary
considerably, but a nominal 9 m (29.5 ft) log with a diameter of 0.305 m (1 ft) are assumed here.
For the column, the materials chosen are concrete, steel, and wood. For the concrete column, a
square cross-section 0.6 m x 0.6 m is used (Mikhaylov 2009). For the steel column, a wide
flange beam W24X104 is used; it was chosen so that it has nearly the same flexural rigidity (EI)
as the concrete column. For the wood column, a diameter of 0.305 m is used, which is the same
as the wood pole. The dimensional properties of the pole and columns are given in Table 1, and
the corresponding nondimensional parameters are given in Table 2. For an isotropic material,
  2 1   /  , in which  is the Poisson ratio, and a reasonable range of  is from 20/9 at the
low end to 5 or 6 at the high end. However, wood is not an isotropic material, and wood
properties vary widely. A value of   10 is used here for the wood pole, which is at the low end
of values suggested in Yoshihara et al. (1998).

The results presented here are nondimensionalized, but it is useful to obtain an understanding
of the levels of force and duration that might be anticipated. An impact velocity of 10 km/h (6.2
mph) is physically reasonable during a tsunami. If the wood pole specified above were to hit a
rigid stop axially, the impact force given by Eq.(3.12), based on E = 9.3 GPa (1,350 ksi) and τ
=10, is 443 kN (99,600 lb). For a 9 m pole, the approximate duration is 4.22 ms. If it were to hit

58
a rigid stop transversely, the impact force given by Eq.(3.13) is 280 kN (62,950 lb) with an
approximate duration of 6.7 ms.

For the columns, two lengths have been chosen: 3.6 m and 6 m. For the wood pole, two
lengths have been chosen: 3 m and 9 m. For convenience in describing the results and for labels
in the plots, the notation CiPi is defined such that C1 = 3.6 m column, C2 = 6 m column, P1 = 3
m pole, and P2 = 9 m pole. For example, a curve indicated by C1P2 shows the results of impact
between a 3.6 m column and a 9 m pole. The length ratios Lp L for the different cases are

shown in Table 3.

The results are organized as follows. First, the case of a pole hitting fixed-fixed and pinned-
pinned columns will be presented. Second, results from Timoshenko and Euler beam theories are
compared. This is followed by an examination of the effect of rotary inertia. Finally, the energy
propagation and energy transfer is examined.

3.5.2 Axial Impact of a Pole against a Column

As mentioned previously, the spring between the column and the pole represents localized
effects, which are difficult to quantify in practice and which will be effectively ignored herein.
Packzkowski (2012) investigated the axial impact of a pole against a flexible, massless stop.
They found that for ratios of spring stiffness to pole stiffness kr  ks / ( E p Ap / Lp ) of 40 or larger,

the response was very nearly the same as for a rigid stop. To be conservative, herein kr = 100 is
used, and a parameter study has shown that the results presented are insensitive to higher values.

The impact force is nondimensionalized by Eq.(3.15). The time is nondimensionalized by

either tax  2Lp / coP , which is the time it takes for the impact axial wave in the pole to return back

to the impact point, or tsh  L / cs , the time it takes for a shear wave in the column to return to the
impact point. All impact occurs at the center of the column.

59
 Table 1: Dimensional Properties

E (GPa) G (GPa) ρ (kg/m3) κ

Wood Pole 9.3 1.033 512.5 0.9

Wood Column 9.3 1.033 512.5 0.9

Concrete Column 24.682 10.284 2400 0.833

Steel Column 200 76.923 7800 0.393

Table 2: Nondimensional Parameters

τ μ αm α1 α2

Pole-Concrete column 2.88 1.328 0.043 1.049 21.472

Pole-Steel column 6.624 0.841 0.242 1.262 4.811

Pole-Wood column 10 1 1 2.581 1.632

Table 3: Length Ratios

Case Lp / L
C1P1 0.833
C1P2 2.5
C2P1 0.5
C2P2 1.5

Figure 18 shows the impact force time history of the wood log hitting the fixed-fixed
concrete columns. Time is nondimensionalized by tax . For C1P1 and C2P1, tsh tax is 1.352 and
2.254, respectively. As a result, the impact force is dominated by the axial wave propagation, and
separation occurs when t  tax . However, for C1P2 and C2P2, tsh / tax is 0.451 and 0.751,
respectively. As a result, the returning shear wave in the column can be detected in the impact
force at these times. Note that the time history of a pole hitting a rigid stop would be a

60
rectangular pulse of magnitude 1 and duration 1. C1P2 shows an increase above the initial
impact force as a result of the vibration in the column.

Figure 19 shows the corresponding time histories for the steel columns. Time is again
nondimensionalized by tax . For C1P1 and C2P1, tsh / tax is 1.299 and 2.165, respectively, and

again the response is dominated by the axial wave. For C1P2 and C2P2, tsh / tax is 0.433 and
0.722, respectively. It is clear the returning shear wave results in a bigger disturbance in these
cases, with a substantial amplification in the impact force from column vibration in the case of
C1P2.

Considered together, Figure 18 and Figure 19 indicate that when a pole hits a stiff column,
the duration of loading is determined by the time it takes the longitudinal wave to return to the
impact point. However, if the shear wave in the column returns prior to this, the impact force can
be increased substantially over the initial force.

The impact force for the pole hitting a wood column is shown in Figure 20. Time is
nondimensionalized by t sh . The sharp drop-off in the initial impact force evidenced in this
figure is characteristic of a beam hitting a rigid stop. Hence, in all cases the impact force is
dominated by the wave propagation in the column. However, if the longitudinal wave in the pole
returns prior to the return of the shear wave, separation will occur. For C1P2, tax / tsh is 1.581,
and the impact duration is determined by the shear wave propagation in the column, a result that
is unattainable with Euler-Bernoulli beam theory (see section 3.5.3). Note also the sharp increase
in impact force near the end of impact, which is a result of the dynamics of the column. For the
other three cases, the impact duration is nearly equal to the time taken for the axial wave in the
pole to return to the impact point, where tax / tsh = 0.527, 0.316, and 0.949 for C1P1, C2P1, and
C2P2 respectively.

Figure 21 compares the force-time histories of a wood pole hitting fixed-fixed and pinned-
pinned steel columns for C2P2 (cf. Figure 19). Time is nondimensionalized by tax . The curves

for both cases are identical until time L / c0  0.333 tax . At one-half this time, the pinned support

of the column shows rotation. That is, the wavefront in the column is propagating at c0 , ahead of

61
the shear wave propagating at approximately cs . This wave causes rotation at the free end, then
causes a slight change in impact force when the wave returns to the impact location. The
wavefront propagating at c0 is associated with the propagation of bending moment and angular
velocity discontinuities (Boley and Chao 1955; Graff 1991). Note, however, that the impact force
has a stronger dependence on the wave propagating at nearly cs ; these waves cause first a
decrease and then an increase in the impact force. As long as the longitudinal wave in the pole
has not returned prior to the waves in the column, the boundary conditions of the column can
cause a significant difference in impact forces.

1.2

0.8

C1P1
0
F/F

0.6
C1P2
C2P1
C2P2
0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
t/t
ax

Figure 18: Impact force for a wood pole hitting fixed-fixed concrete columns

62
 2

C1P1
C1P2
C2P1
1.5 C2P2
0
F/F

0.5

0
0 0.2 0.4 0.6 0.8 1
t/t
ax

Figure 19: Impact force for a wood pole hitting fixed-fixed steel columns

63
 1

C1P1
0.8 C1P2
C2P1
C2P2

0.6
0
F/F

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
t/t
sh

Figure 20: Impact force for a wood pole hitting fixed-fixed wood columns

64
 1.2 0.005

Fixed supports
Pinned supports
1
0.004

0.8

End support rotation

0.003
0
F/F

0.6

0.002
0.4
Rotation

0.001
0.2

0 0
0 0.2 0.4 0.6 0.8 1
t/t
ax

Figure 21: Impact force for a wood pole hitting steel columns (left axis) and the end rotation of
the pinned column (right axis) for the case of C2P2

3.5.3 Comparison of Timoshenko versus Euler Models

The impact response from Timoshenko (T) and Euler (E) models for the column are
compared in this section. A similar solution methodology as described above was implemented
for the Euler model. Figure 22 shows the force-time histories for a wood pole hitting fixed-end
concrete (C), steel (S) and wood (W) columns, for case C1P2. For the concrete column (circular
symbols), both beam models give similar results. This is because the concrete column is quite
stiff relative to the wood pole, and the response is dominated in both magnitude and duration by
the axial response of the pole. The nearly constant impact force is characteristic of a pole hitting
a stiff, massless stop. For the steel column (hollow diamond symbols), again it is quite stiff and
the duration is also governed by the wave propagation in the pole. However, the reflected shear
wave at time approximately 0.44 causes a drop in the impact force, and then the dynamics in the

65
column cause an increase. This behavior is captured poorly by Euler. For the wood column (solid
diamond symbols), the force and duration are governed by the column. The initial peak followed
by an exponential decay in the force is characteristic of a beam hitting a stiff, massless stop, for
example. The Euler model substantially over predicts the initial peak force. In addition, the early
return of waves in the column lead to poor prediction of the duration, which is governed by the
shear wave in the column, as well as the intermediate response. These results indicate that Euler
beam theory should not be used for these kinds of impact problems if the beam response is
significant.

As observed in Figure 22 for the wood column, Euler can over predict the initial peak force,
and as shown previously the magnitude is unbounded as the stiffness of the impact increases.
Figure 23 shows the results for the wood pole hitting a wood column, C1P2, in which the contact
stiffness has been varied. As noted previously, the stiffness ratio kr = 100 has been used in all the
previous results. Figure 23 also includes the results for kr = 50 and 500. As can be seen, both
Timoshenko and Euler give converged results for all cases except for the initial peak force as
predicted by Euler. This force continues to increase as the spring stiffness increases, which is
consistent with the derivation above that illustrated the potential unboundedness of the impact
force for an Euler model. As noted previously, Euler also fails to predict well the time variation
in the force.

66
 2

1.5
0
F/F

1
C-T

C-E

S-T
0.5 S-E

W-T

W-E

0
0 0.2 0.4 0.6 0.8 1 1.2
t/t
ax

Figure 22: Impact force for a wood pole hitting fixed-end concrete (C), steel (S), and wood (W)
columns for case C1P2 based on Timoshenko (T) and Euler (E) beam theories

2.5
k =50-T
r
2 k =50-E
r
k =100-T
r
1.5 k =100-E
r
0
F/F

k =500-T
r
1
k =500-E
r

0.5

0
0 0.2 0.4 0.6 0.8 1
t/t
sh

Figure 23: Impact force for a wood pole hitting a fixed-end wood column for case C1P2 for
different ratios of the contact stiffness to the axial pole stiffness, for Timoshenko (T) and Euler
(E) theories

67
 3.5.4 Effect of Rotary Inertia
It is of interest to determine the influence of rotary inertia on the impact forces. Some
structural finite element codes include shear deformation but ignore the rotary inertia. For this
study, impact of a wood beam against a (nearly) rigid stop is considered. The beam has the same
cross-sectional properties as before, with L / r  120 ,   20 / 9 , and a shear coefficient   0.9 .
The rigid stop is modeled as a very stiff, massless spring, with a stiffness equal to 1200 times the
shear stiffness of the beam.

The formulation cannot treat zero rotary inertia ( 2  0) ; therefore, this case was

approximated by a beam with one-hundredth the normal rotary inertia ( 2  0.01) . The results for
impact with and without rotary inertia are shown in Figure 24. As can be seen, the rotary inertia
in this case of nearly rigid impact for a beam with relatively high shear stiffness (small ) has a
noticeable effect on the impact force near the end of impact, when the waves in the beam return
to the impact point. The small oscillations for the case with rotary inertia that initiate at about
time 0.67 result from the rotational kinetic energy propagating at c0 returning to the impact point.
With small (zero) rotary inertia, these waves have little (zero) energy. These results indicate that
ignoring rotary inertia may not be sufficiently accurate if high fidelity in the impact force time
history is desired.

68
 1

0.8


0.6 
cR
F/F

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1
t/t
sh

Figure 24: Effect of rotary inertia on impact force for L/r=120 and   20 / 9 with   1 and
2

 2  0.01

3.5.5 Energy Breakdown

The energy exchange gives an insight on the generation and behavior of the impact force.
Figure 25 compares the total shear strain energy and the bending strain energy in the column as a
function of time (nondimensionalized by t sh ) for the case of a wood pole hitting a fixed-fixed
wood column (C1P2). The maximum impact force, shown in Figure 20, is dominated by shear,
as can be seen by the sharp increase in shear strain energy and the much slower increase in
bending strain energy.

Figure 26 shows the time histories for all strain and kinetic energies. The initial energy is all
translational kinetic energy of the pole, but as the impact progresses a significant part of the
initial pole translational kinetic energy is exchanged with the translational kinetic energy of the
column, suggesting an important role for the inertia of the column on the impact dynamics. Also
noteworthy is the initial energy exchange between the pole kinetic energy and shear and bending
energy. The evolution of the bending energy requires the buildup of the curvature and a gradient
of rotation (Eq.(3.21)). The shear energy on the other hand exchanges energy directly with the

69
kinetic energy of the pole (Eq.(3.20)). As a result, the initial exchange of the pole translational
kinetic energy is mainly with the shear energy when compared with the bending energy, and the
exchange is “local” in nature. As times progresses, curvature is established and pole kinetic
energy is exchanged predominantly with translational kinetic energy of the column and bending
energy of the column. Figure 27 shows the evolution of the various energy densities as functions
of space and time during impact.

From Figure 27, it is apparent that at time of impact, there is a large and local spike in the
shear energy density, which shows that the impact is absorbed initially via shearing. This local
spike is then quickly transferred to bending energy as the spike in the shear energy density at the
impact point propagates away from the impact point down the column until it is reflected at
about time L/2cs (𝑡̅ = 1/2); see Figure 27 (a). Although there is some wave dispersion after
reflection, it is not as distinct as the dispersion of the flexural waves. Of particular interest, at the
point of impact, 𝑥̅ = 0, the shear energy density reaches its maximum at the start of impact, and
then drops to near zero. Away from the point of impact, the evolution of the rotational kinetic
energy displays the richest dynamics. In fact, despite being of a smaller magnitude for this case,
Figure 27 (d) shows that the disturbance in the rotational kinetic energy at the point of impact
has two modes of wavefront propagation: at the distortional speed, cs , and at the dilatational

speed, c0 (cf. Figure 21). The reflection of the waves for the rotational kinetic energy at the fixed
ends is also more complex with several wavefronts being generated with pronounced lower
frequency oscillations in their amplitudes. As predicted by theory (Boley and Chao 1955; Graff
1991), besides the rotational kinetic energy, bending energy also displays the second wavefront
moving at dilatational speed, Figure 27 (b). Figure 27 (c) shows the evolution of the
displacement kinetic energy of the column. It is apparent that a significant part of the initial
energy is absorbed by the column at the impact point in the form of displacement kinetic energy.
This energy propagates away from the impact point on the shear wave till the end of the column.
At the end of the column, it is reflected and intensified by the strong shear strain gradient in the
path of the shear wave. In addition, this figure shows that a simple mass spring model for the
column would be inadequate, because the propagation of the kinetic energy by shear would be
absent. Finally, Figure 27 (e) and Figure 27 (f) depict the passive role of the pole for this case.
The pole serves as a mere signal propagator of the axial strain from the impact into the pole. In

70
particular, the wave motion on the pole has no effect on the duration of the impact, which is
determined in this case by the return of the reflected shear wave from the column's ends.

0.06

0.05
Nondimensional Energy

0.04
Shear-Column
Bending-Column
0.03

0.02

0.01

0
0 0.05 0.1 0.15 0.2 0.25 0.3 0.35 0.4
t/t
sh

Figure 25: Shear and bending strain energy in the wood column for case C1P2

71
 1

0.1
Nondimensional Energy

0.01

0.001

Total

0.0001 Kinetic-Pole

Strain-Pole

Translation-Column

-5 Rotation-Column
10
Bending-Column

Shear-Column

Spring
-6
10
0 0.2 0.4 0.6 0.8 1
t/t
sh

Figure 26: Component energies and total energy for the case C1P2 with a wood column

(a) Shear energy of the column (b) Bending energy of the column

72
 (c)
Translational kinetic energy of the column (d) Rotational kinetic energy of the column

(e) Kinetic energy of the pole (f) Strain energy of the pole

Figure 27: Time variation of energy densities for case C1P2 with a wood column

3.6 Summary
An analytical solution for the linear impact response of a pole impacting a column, or a
beam impacting a flexible axial stop, has been presented. The solution is based on
Timoshenko beam theory and uses modal superposition. Contact can occur at an arbitrary
point within the column span. The following conclusions may be made.

1. The initial peak impact force can be predicted by a simple formula, Eq. (3.15). In the
case of stiff columns, the pole properties govern the force; in the case of a stiff pole, the
column properties govern. In no case is the total mass of the projectile significant to the
initial peak force.

73
2. The impact duration is governed by either the longitudinal wave speed in the pole (for
stiff columns) or the shear wave speed in the column.
3. For cases in which the column properties are significant, the initial response is
dominated by shear behavior, and Timoshenko beam theory rather than Euler-Bernoulli
beam theory should be used. Indeed, based on Euler-Bernoulli beam theory the initial
impact force is unbounded as the stop stiffness increases, whereas Timoshenko beam
theory has a clear limiting value for the initial impact force.
4. The rotary inertia does not affect the initial peak impact force but it can have an effect
on the impact force time history after waves in the beam return to the impact point.
5. The energy plots provide a vivid picture of the complex interactions and energy
exchange taking place during the impact. For instance, of direct relevance to the present
work, it explains the importance of the shear mode at the beginning of the impact to set
the impact force and also depicts the relevance of the various energy exchanges in
setting later spikes in the force as well as the impact duration. The energy exchange also
shows the importance of the inertia of the column in absorbing a significant part of the
initial kinetic energy of the pole.

74
 Chapter 4
Transverse Impact of a Horizontal Beam on a Vertical
Column

4.1 Introduction
Impact between a beam and a rod (i.e., a ‘stop’) or between two beams has various
applications, including piping systems, valves, heat exchangers, and arrays of offshore oil
production risers. However, the primary interest herein is the impact of water driven debris on
coastal structures, for example when a wood pole or log impacts a structural column of a
building or other structure.

Many of the studies on beam response during impact involve a mass impacting a beam rather
than one beam impacting another beam transversely, which is the problem of interest here. In
addition, many studies used Euler-Bernoulli beam theory, which gives inaccurate results for
many transverse impact cases; see Khowitar et al. (2014). Moreover, there seems to have been
little attention given to the energy exchange during multiple impacts of two beams.

In this chapter, Timoshenko beam theory is used to model the transverse impact of a pole on
a column. Because our interest is primarily on the impact of water driven debris on coastal
structures, which often involves small impact speeds (~3 m/s), only elastic behavior is
considered. One important consequence of this assumption is that mechanical energy is
conserved.

The pole is moving initially with a constant velocity and hits laterally the column at the
midspan. A spring is placed between the pole and the column at the impact location. Although in
principle such a spring could be used to represent a local, linear contact stiffness, in practice such
stiffness is difficult to quantify. As in Khowitar et al. (2014), a very stiff spring is used such that
the effect of the contact spring is negligible; it functions essentially as a penalty parameter to
couple the pole and the column. Even a stiff contact spring as assumed here results in an impact
force time-history that has a finite loading rate, which is more realistic than the discontinuous

75
time history that would result from an infinite contact stiffness. In addition, it eliminates the
Gibbs phenomenon in impact simulations.

The primary focus in this chapter is on the contact force during multiple impacts and the
energy exchange and behavior during each contact and separation phase. The chapter is
organized as follows. First, the physical model is described and the governing equations and the
nondimensionalization of the variables are presented. Then, the analytical solutions, which are
based on superposition of normal modes of vibration, are developed. Results of a study on
impact response for parameters within a range that might be seen in woody debris impact are
presented and interpreted. Finally, the main findings of this study are summarized in the
conclusions.

4.2 Physical System

Figure 28 shows a schematic of a pole hitting laterally a column. The pole moves towards the

column with a constant translational velocity v0 (at x1  0 ) and angular velocity 0 and hits
*

anywhere along the pole. As mentioned, a contact spring with stiffness k* is placed between the
pole and the column at the impact points. Multiple impacts are considered, and the whole
collision event, from first contact to final separation, is divided into multiple contact phases and
separation phases.

The column has fixed-fixed boundary conditions and zero initial conditions. The origin of the

coordinate systems 𝑥1∗ and 𝑥2∗ ( x1  0 and x2  0 ) are located at the point of impact. The column
* *

and the pole are considered elastic and homogeneous, with constant cross-sectional properties.

The pole has Young’s modulus E p , shear modulus G p , mass density  p , sectional area

moment of inertia I p , sectional area Ap , sectional rotary inertia I pc   p  p I p , shear coefficient

2

 p , and total length Lp  L1 p  L2 p . A subscript “c” is used for the analogous column properties.

The speeds of sound are c0 p  E p /  p and c0c  Ec / c . The speeds of shear waves are given

by csp   pGp /  p and csc   cGc / c .

76
 For this system, the fundamental kinematic variables, which are functions of time t*, are the

transverse displacement of the pole wp ( x1 , t ) , cross-sectional rotation of the pole  p ( x1 , t ) ,

* * * * *

* * *
transverse displacement of the column wc ( x2 , t ) , and cross-sectional rotation of the column

c ( x2* , t * ) .

A similar nondimensionalization as in Boley (1955), Xing et al. (2002), and Khowitar et al.

(2014) is used herein. The length scale is nondimensionalized by rp  I p / Ap and time by

rp / c0 p . The following nondimensional variables result:

x* w*p x2* w* c r
x1  1 wp  x2  wc  c t  t * 0 p k  p k *
rp rp rp rp rp E p Ap
, , , , ,

The ratios of modulus of elasticity to effective shear modulus for the pole and column are

 p  E p /  pGp and  c  Ec /  cGc . The ratio of the speed of sound in the two materials is

c  c0c / c0 p . r  rc / rp is the ratio of the two radii of gyration, and m  c Ac /  p Ap is the

ratio of the masses per unit length.

77
 Figure 28: Schematic of pole and column

4.3 Equations of Motion

The equations of motion of the pole are

1   wp  p   2 wp
2

    k  w
 p  0, t   w  0, t  
   0 (4.1)
 p  x12 x1 
c
 t 2

 2 p1  w   2

  p   p    p2 2p  0 (4.2)
x1  p  x1
2
 t

while the equations of motion of the column are

 2 wc c  c  2 wc

x22 x2
 k  p   c  
 w 0, t  w 0, t   
c2 t 2
0 (4.3)

78
  2c 1  wc  c3  2c
    c  0 (4.4)
x22 r2 c  x2  c t
2 2

in which  is the Dirac delta distribution. In the contact phase, there is a discontinuity in shear at
 
x1 , x2  0 . Therefore the displacements are denoted as wp  wp for x1  0 , wp  wp for x1  0 ,

wc  wc for x2  0 , and wc  wc for x2  0 . Similar notation is used for  .

The compatibility equations at x1  0 for the pole and at x2  0 for the column are equal
displacement, rotation, and curvature. In addition, we have the jump conditions for shear:

wp  0, t  wp  0, t 
   p k  wp  0, t   wc  0, t    0 (4.5)
x1 x1

wc  0, t  wc  0, t  
  2 c k  wp  0, t   wc  0, t    0 (4.6)
x2 x2 c  m

4.4 Analytical Solution

4.4.1 Separation of Variables

The method of separation of variables is adopted in which the kinematic variables can be
written as

 wp ( x1 , t )    pn

( x1 ) 
     
  p ( x1 , t )   pn ( x1 ) 
     qn  t     (4.7)
 wc ( x2 , t )  n  cn ( x2 ) 
 c ( x2 , t )   cn ( x2 ) 

The mode shapes of the pole and the column are the usual Timoshenko beam mode shapes. For
the pole, the mode shapes can be written as (Khowitar et al. 2014)

79
   p   A1 cos 1 x1  A2 sin 1 x1  A3 e1x1  A4 e 1x1 
 
 p     1  A1 sin 1 x1  A2 cos 1 x1    1  A3 e  A4 e 
     1 x1  1 x1

   (4.8)
 c  
  2 x2   2 x2
B1 cos  2 x2  B2 sin  2 x2  B3 e  B4 e
 

 c   B  sin  x  B  cos  x   B  e 2 x1  B  e  2 x 
 2  1 2 2 2 2 2 2  3 4 
with

 2  p2   p     2  p2   p   4
2

1 
2

 2  p2   p     2  p2   p   4
2

1 
2

  1   p 2   2  /  ,  1   p 2   2  / 


 2 c2   c   r2 2 c2   c   4c2
2

r
2 
2 c


 2 c2   c   r2 2 c2   c   4c2
2

r
2 
2 c

 c   c 
 2    2   22  /  2 ,  2    2   22  /  2
 
2 2
c  c 

where    rp / c0 p is the nondimensional natural frequency and  is the dimensional

* *

frequency. It should be noted that 1,2 are always real, but 1,2 can be real or imaginary. The

natural frequencies and the coefficients of the mode shapes are determined by imposing the
compatibility and boundary conditions.

The generalized time function is

80
 qn  t   An cos nt  Bn sin nt (4.9)

In transitioning between contact and separation phases, the ‘initial’ conditions at transition time

t t must be transferred to the new phase. The general formulas for the coefficients An and Bn
can be determined using the initial conditions and the property of orthogonality of the mode
shapes from

Iv
An  I w cos ntt  sin ntt
n
(4.10)
Iv
Bn  I w sin ntt  cos ntt
n

where

 w     dx1  m  wtccn  c r  tc cn dx2

2 2
tp pn p tp pn
Iw 
    dx1  m  cn2  c r   cn2 dx2
2 2 2 2
pn p pn

v    p2tp pn dx1  m  vtccn  c r  tc cn dx2

2
tp pn
Iv 
   p2 pn dx1  m  cn2  c r   cn2 dx2
2 2 2
pn

wt ,  t , vt , t are the transverse displacement, cross-section rotation, transverse velocity, and

cross-section rotational velocity, respectively, at transition time tt .

Two special mode shapes must be considered. The first are the rigid-body modes when
n  0 , in which the mode shapes  ,  are  x,1 for rotational motion and 1,0  for
translational motion, and the generalized time function is

qn  Ant  Bn (4.11)

1
with An  I v and B n  I w  I v t t . The second is when   , which means 1  0 , where
p  p
the mode shapes can be obtained by taking the limit of Eq.(4.8).

81
 4.4.2 Initial Peak Impact Force
For a very short time after the first impact, the deformation is dominated by shear, and the
initial peak force can be derived the same way as in (Khowitar et al. 2014) for axial impact by
satisfying the force equilibrium at the impact point. For a stiff contact spring, the initial peak
force is

FpR FcR
F0  (4.12)
FpR  FcR

in which FpR  2 p csp v0 Ap , FcR  2c csc v0 Ac , and csp and csc are the shear wave speeds in the

pole and the column respectively. FpR and FcR are the initial impact forces for the pole and

column, respectively, hitting a rigid stop; see Khowitar et al. (2014) for details.

It may be deduced from Eq. (4.12) that the initial peak force is dominated by local shear
response and is independent of both pole and column lengths, and hence the total initial
momentum and total mass. For design purposes, it is more convenient to write the impact force

in terms of the shear stiffness of the pole, k p   pGp Ap / Lp , or of the column, kc   cGc Ac / Lc

and the respective total mass. Using the expressions for FpR and FcR in Eq. (4.12) results in

F0  1 k p mp v0 (4.13)

with

2
1 
1
1
cs m

or equivalently

F0  2 kc mc v0 (4.14)

with

82
 2
2 
1  cs m

where cs  csc / csp  c  p /  c

From Eq.(4.13), it is clear that the peak impact force occurs when a (given) pole hits a rigid

stop (i.e. c approaches infinity) and it is given by F0 = 2 k p mp v0 = FpR .

4.4.3 Mechanical Energy

The energy exchange is useful to understand the behavior of the impact force during multiple

impacts. For convenience, all energies are nondimensionalized by E p Ap rp , and the

nondimensional energies are

1  wp   p2   p 
L2 p 2 L2 p 2

TKEp     dx1 RKEp     dx1

 L1 p
2  t   L1 p
2  t 

1   p  1  wp
L2 p 2 L2 p 2

BEp     dx1 SEp      p  dx1
 L1 p
2  x   L1 p
2 p  x 

  w   2   2   
L2 c 2 L2 c 2

TKEc   m  c  dx2 RKEc   c m r  c  dx2

 L1c
2  t   L1c
2  t 

  2  2      2  w
L2 c 2 L2 c 2

BEc   m c r  c  dx2 SEc   m c  c  c  dx2
 L1c
2  x   L1c
2 c  x 

TKE is the translational kinetic energy, RKE is the rotational kinetic energy, BE is the
bending strain energy, SE is the shear strain energy, and “p” and “c” refer to pole and column,
respectively.

To write the relationships between different energy components, it is convenient to represent

the velocities as v p  wp / t , vc  wc / t ,  p   p / t , and c  c / t ; the curvatures as

83
 wp
 p   p / x1 and  c  c / x2 ; and the shear strains as  p    p and  c  wc  c . The
x1 x2
nondimensional equations of motion for the pole then become

1  p v
 k  wp  0, t   wc  0, t     p  0 (4.15)
 p x1 t

 p 1  p
  p   p2 0 (4.16)
x1 p t

and for the column

 c  v
 k  wp  0, t   wc  0, t     c2 c  0 (4.17)
x2 c t

 c 1  3 c
 2  c  c2 0 (4.18)
x2 r  c c t

The nondimensional energy densities (per unit length) are as follows: wp  v p / 2 is the
2

translational kinetic energy density of the pole, wc  mvc / 2 is the translational kinetic energy
2

density of the column,  p   p  p / 2 is the rotational kinetic energy density of the pole,
2 2

 c  c2 m r2c2 / 2 is the rotational kinetic energy density of the column, Bp   p2 / 2 is the

bending strain energy density of the pole, Bc  m c r  c / 2 is the bending strain energy density
2 2 2

of the column,  p   p / 2 p is the shear strain energy density of the pole, and
2

 c  m c2 c2 / 2 c is the shear strain energy density of the column. Using Eqs. (4.15) and (4.16)
one can obtain the governing equations for the pole in terms of the energy densities:

wp v p  p
  vk  wp  0, t   wc  0, t   (4.19)
t  p x1

84
  p  p  Bp p
   (4.20)
t  p x1 p p

 p  p v p  p
   (4.21)
t  p x1  p p

 Bp  p
p (4.22)
t x1

The governing equations for the column in terms of the energy densities can be obtained in a
similar manner using Eqs. (4.17) and (4.18).

As discussed in Khowitar et al. (2014), these equations describe the interaction of the various
energy densities and can be used to explain the energy transfer between the different energy
components.

4.5 Solution Methodology

The analytical solution is obtained by following the same procedure used in Khowitar et al.
(2014). The boundary and compatibility conditions can be written in terms of the mode shapes,
which results in 16 equations to obtain the natural frequencies. A numerical method was used to
obtain the natural frequencies of the system. The bisection method is used to narrow down the
search domain and then the secant method was used to obtain the frequency. As discussed in the
next section, six cases were considered here. A convergence study revealed that as a result of
high harmonics during the impact, 500 symmetric mode shapes were sufficient for all cases.
Some details regarding the solution are discussed in the appendix, as is a comparison with a
finite element model.

4.6 Numerical Results

4.6.1 Physical Dimensions

The results of a wood pole impacting a fixed-end column are considered here. The impact is
at midspan of both the column and the pole, which is an important scenario for design. The pole
has an initial translational velocity. The same material and dimensions are used for the pole and

85
the column as in the study conducted by (Khowitar et al. 2014). The pole has a weight of 342-kg
(754-lb), length of 9-m (29.5-ft), and a diameter of 0.305 m (1 ft). This is somewhat less than the
prototypical 1,000 lb wood pole suggested by a common U.S. design guideline (ASCE 2010).
The materials of the columns are concrete, steel, and wood. The lengths of the columns are 3.6 m
(C1) and 9 m (C2). For the concrete column, a 0.6 x 0.6 m cross-section is assumed. The steel
column is a wide flange I-beam, W24X104, which has a bending stiffness EI approximately the
same as the concrete column. The wood column has the same diameter and material properties as
the wood pole. These choices of columns vary from very flexible (long wood column) to very
stiff (short concrete column). Therefore, the results reported herein should bound many useful
practical cases.

Relevant dimensional properties and nondimensional parameters are shown in Table 4 and
Table 5, respectively. A value of   10 for wood is used, which is at the low end of the values
suggested in Yoshihara et al. (1998). A convergence study showed that the results do not change

if the stiffness of the spring is 700 times the shear stiffness of the pole, k p ; therefore this was the

minimum spring stiffness used.

The nondimensional time introduced previously is not physically very meaningful. Therefore,
for presentation of the results time is nondimensionalized by the time the shear wave in the pole

takes to return to the impact point, t  t / t p , where t p  Lp / csp ≈ 0.00668 s.

*

Table 4: Dimensional Properties

Member E (GPa) G(GPa)  (kg / m3 ) 

Wood pole and column 9.3 1.033 512.5 0.9
Concrete column 24.682 10.284 2,400 0.833
Steel column 200 76.923 7,800 0.393

86
 Table 5: Nondimensional Parameters

Case c m r c 1 2
Pole-concrete column 2.88 23.074 2.272 0.753 1.891 0.109
Pole-steel column 6.624 4.123 3.347 1.189 1.661 0.339
Pole-wood column 10 1 1 1 1 1

The impact force is nondimensionalized by the instantaneous peak force obtained from Eq.
(4.13). The dimensional instantaneous impact forces, per meter per second, are 98 kN, 86.5 kN,
and 50.4 kN for the concrete, steel, and wood columns, respectively. It can be seen from Table 5
and Eq. (4.13) that the variation in these values depends mainly on the mass ratio, which ranges
between 1 and 23 and is the parameter that has the most variation for these cases.

4.6.2 Force-Time History

Figure 29 shows the impact force-time history for a 9 m wood pole hitting 3.6 m and 6m
concrete, steel, and wood columns. Note that in all cases the initial peak force is predicted well
by Eq. (4.13), i.e., the initial nondimensional peak force is nearly 1. However, the maximum
impact force is up to 1.5 times the initial peak.

For the concrete columns, the duration of the first contact phase is 0.00668 s, which is the
time taken by the shear wave in the pole to return to the impact point. At that time, the velocity
along the pole away from the impact point is still for the most part the initial velocity. The
forward momentum leads the portion of the pole at the impact point to move and hit the column
again after a short time. For most of the impact cases, there is a high frequency pounding after a
fairly large separation phase. However, this is not the case for the most flexible (long wood)
column, in which the durations of all the contact and separation phases are small. When the shear
wave in the column returns to the impact point prior to separation, the force experiences a
sudden change and subsequent high harmonics during the first contact phase. For instance, the
shear wave in the column returns to the impact point at t  0.29 for the short concrete column
(Figure 29a), 0.48 for the long concrete column (Figure 29b), and 0.27 for the short steel column
(Figure 29c).

87
 It can be noticed from Figure 29 that for concrete and steel columns, the nondimensional
duration of the impact event is between 5.5 and 6. However, for the wood column, the durations
are 7.1 and 10.4. This is because the more flexible wood columns displace more, and the pole
and column move more together and final separation takes longer. The higher frequency of the
contact events results from the smaller relative displacement between the pole and the column
for these cases

1.8 1.8

1.6 1.6

1.4 1.4
Nondimensional force

Nondimensional force
1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Nondimensional time Nondimensional time

(a) 3.6 m concrete column (b) 6 m concrete column

1.8 1.8

1.6 1.6

1.4 1.4
Nondimensional force

Nondimensional force

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Nondimensional time Nondimensional time

(c) 3.6 m steel column (d) 6 m steel column

88
 1.8 1.8

1.6 1.6

1.4 1.4
Nondimensional force

1.2

Nondimensional force
1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 9 10 11
Nondimensional time Nondimensional time

(e) 3.6 m wood column (f) 6 m wood column

Figure 29: Impact force time histories

4.6.3 Energy-Time History

Figure 30 shows the time histories of energy for the six cases. The energy is
nondimensionalized by the initial kinetic energy of the pole. The rotational kinetic energies and
the spring energy are omitted from the plots because their values are very small relative to the
other energies.

For all cases, most of the energy is in bending strain and translational kinetic energies of the
pole. However, it can be seen that in a very short time after the first impact, the shear energy
dominates the bending energy, which confirms that the instantaneous impact force is dominated
by shear deformation either in the pole or in the column. The bending energy then dominates as
the effect of the impact propagates along the pole and column. Since the stiffness of the column
is relatively high for the case of the concrete and steel columns, the sum of the bending strain
energy and the kinetic translational energy in the pole constitutes almost 90% of the total energy
during the collision event. Moreover, these two energies are almost symmetric around a
horizontal line at 45% of the total energy. On the other hand, for the soft (wood) column, a
significant part of the total energy is absorbed by the bending strain energy and translational
kinetic energy of the column.

89
 For the concrete and steel columns, the kinetic translational energy of the pole decreases with
the increase of the bending strain energy until t  0.8 , at which time the impact force
experiences a sharp increase, the kinetic translational energy starts to increase and the bending
strain energy starts to decrease. At t  1.35 the translational kinetic energy of the pole reaches a
relative maximum and the bending energy reaches a relative minimum value, at which the
second contact phase almost starts. After that the kinetic energy decreases while the bending
energy increases. When t  3 , the bending strain energy of the pole reaches a relative maximum
value and the kinetic translational energy reaches a relative minimum value, and then the
bending energy gradually decreases with the increase of the kinetic energy until a local drop at
t  4 occurs in the bending energy at which the separation phase with the larger duration almost
ends. After the bending energy reaches a local minimum at t  4.2 , the bending energy of the
pole then starts to increase with the decrease of the kinetic translational energy until it reaches
another peak value at t  4.8 at which the translational kinetic energy has another minimum
value. This basically antisymmetric energy exchange between these two components is a
reflection of the very low energy in the other components of the pole and the stiffness of the
columns. It can be seen that the bending strain energy and the translational energy of the pole are
equal at t  2 and t  5.5 .

On the other hand, for the short wood column, the kinetic translational energy reaches its
minimum value at t  3.5 , at which the bending energy of the pole reaches its peak value. They
are equal at t  2.4 and t  6 . The bending energy of the column has peaks at t  1 and t  6 .
For the case of the long wood column, the kinetic translational energy of the pole attains its
minimum value at t  5 , at which point the bending strain energy of the pole attains its peak
value. They are equal at t  2.9 and t  6.8 . The bending energy of the column has peak values
at t  2 and t  8.5 .

It is interesting to note how much energy is transferred to the column at the final separation.
For the most rigid (short concrete) column, only 1.5% of the initial energy is transferred to the
column. For the most flexible (long wood) column, almost 7.2% of the initial energy is
transferred to the column. In this case, at the time of separation most of the energy transferred to
the column is absorbed by bending strain energy. In all cases there is relatively little net energy

90
transfer to the column. This means that the impact will have a similarly small impact on the
structure of which the column is a part. However, it should be noted that the fixed end conditions
of the columns likely reduce the net energy transfer.

1.2 1.2
Total Total
TKEp TKEp
BEp BEp
1 SEp 1 SEp
TKEc TKEc
BEc BEc
SEc SEc

Nondimensional energy
Nondimensional energy

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Nondimensional time Nondimensional time

(a) 3.6 m concrete column (b) 6 m concrete column

1.2 1.2
Total Total
TKEp TKEp
BEp BEp
1 SEp SEp
TKEc 1
TKEc
BEc BEc
SEc SEc
Nondimensional energy

Nondimensional energy

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Nondimensional time Nondimensional time

(c) 3.6 m steel column (d) 6 m steel column

91
 1.2 1.2
Total Total
TKEp TKEp
BEp BEp
1 SEp 1 SEp
TKEc TKEc
BEc BEc
SEc SEc
Nondimensional energy

Nondimensional energy
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 9 10 11
Nondimensional time Nondimensional time

(e) 3.6 m wood column (f) 6 m wood column

Figure 30: Component energies and total energy

4.6.4 Impulse-Time History

Figure 31 shows the impulse time histories for the six cases. The impulse is
nondimensionalized by the initial momentum of the pole. The plot shows that the final impulses
for the concrete and steel columns are almost 1.6 times the initial momentum of the pole. For the
wood column, the final impulse is almost 1.8 times the initial momentum of the pole. This
indicates that the final mass-averaged rebound velocity of the pole is almost 60% of the initial
velocity for the concrete and steel columns, and 80% of the initial velocity for the wood column.
These results are consistent with the energy plots in Figure 30.

92
 2 2

1.5 1.5
Nondimensional impulse

Nondimensional impulse
1 1

0.5 0.5

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Nondimensional time Nondimensional time

(a) 3.6 m concrete column (b) 6 m concrete column

2 2

1.5 1.5
Nondimensional impulse

Nondimensional impulse

1 1

0.5 0.5

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Nondimensional time Nondimensional time

(c) 3.6 m steel column (d) 6 m steel column

93
 2 2

1.5 1.5
Nondimensional impulse

Nondimensional impulse
1 1

0.5 0.5

0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 9 10 11
Nondimensional time Nondimensional time

(e) 3.6 m wood column (f) 6 m wood column

Figure 31: Impulse time histories a wood pole hitting columns of different lengths and materials

4.6.5 Energy Decomposition by Modes

The previous energy plots depict the evolution of the extensive energy quantities. They show
the bulk exchange between the energy components, but are blind to the local processes of this
exchange. This and the next section discuss these local processes.

Figure 32 shows the mode contribution of bending and shear energies of the pole for the
short concrete column (left) and long wood column (right). The nondimensional times are 5.5
and 10.2 for the short concrete column and the long wood column, respectively, which
correspond to the durations of the collision event.

These figures clearly show the sharp distinction between the shear and flexural contributions
to the beam vibrational motion. The evolution of the flexural energy, shown in Figure 32(a), is
dominated by the small wave number modes, and associated low period motion. These modes
aggregate most of the flexural energy regardless of the relative stiffness between pole and
column. In addition, their characteristic time scale is much larger than each individual contact
phase, which results in a time evolution that is less affected by the impact events.

94
 By contrast, the shear modes, shown in Figure 32(b), are characterized by short period
vibration and a broader energy spectrum. The increased energy contained in the higher wave
number modes reveal the local nature of the impact, and shows the dominance of the shear at the
early instants of the impact. Indeed, as shown in (Khowitar et al. 2014), it is the shear stress that
sets the magnitude of the peak force. The prevalence and locality of the shear forces at the initial
instants of the impact are also the fundamental assumptions underpinning Eq. (4.12). This
equation shows good agreement between the initial peak-forces obtained with the exact solution
of the Timoshenko beam equations and Eq. (4.12). And, moreover, Eq. (4.12) shows that the
incoming total momentum is immaterial to the computation of the peak force.

Lastly, but not less important, the modal analysis in Figure 32 shows a broadening of the
spectrum of the shear energy over time. The broadening is more pronounced for the impact of
the pole against the more compliant long wood-column and results from the enhanced energy
exchange and increased impact phases between the pole and column in this case.

(a) Bending energy by mode

95
 (b) Shear energy by mode

Figure 32: Modal contribution of energy of the pole for short concrete column (left) and long
wood column (right)

4.6.6 3D Energy Density:

Figure 33 shows the energy density for the short concrete column (on the left) and the long
wood column (on the right). These figures complement the modal analysis of the previous
section with spatiotemporal plots of the energy density for each of the participating energy
components in the pole. A salient feature of these plots is the irregular nature of the vibrational
motion after a series of impact events.

As shown in Figure 33, the two dominant energy components for the full extent of the impact
are the translational kinetic energy and the bending energy of the pole. The bulk of the former
and latter components travel at the shear speed cs, but in the case of bending a small portion of
the energy also travels at dilatational speed c0. In the observed vibrational motion, the bulk
motion of the translational kinetic energy represents the large modes that retain a degree of
organization over an appreciable extent of time, especially for the impact against the concrete
column. These large structures can still be seen at the initial stages of impact on the long wood,
but close to the end of all impacts seem to broadly mix along the half beam. The impact with the
long wood also displays an accumulation of kinetic energy at the free end of the beam that shows
a spatial complementarity with flexural energy, in addition to the temporal complementarity
shown in Figure 33 by the corresponding extensive quantities. Thus bending accumulates at the
impact point, where the beam curvature is largest. Concurrently with these regular mode

96
propagations, dispersion, reflection and interaction are also of import to the impact dynamics.
The latter is more frequent for the impact with the long wood, and together with dispersion and
reflection yields extraordinarily small time scale motions for the rotational kinetic energy and
potential shear energy at the pole. Rotational kinetic energy density is null at the impact point for
the impact at mid beam and grows towards the free end of the pole, where it displays a
“whipping” effect. The shear energy is fed from the column at the impact point and quickly
spreads to each half beam and becomes choppy throughout the beam. In fact, Figure 33(b) shows
an unexpectedly high amplitude low mode for shear energy in the impact against the long wood
that results from the nearly continuous mutual pounding between pole and column.

(a) Translational kinetic energy density

97
(b) Rotational kinetic energy density

(c) Bending energy density

98
 (d) Shear energy density

Figure 33: Time variation of energy densities of the pole for short concrete column (left) and
long wood column (right)

4.7 Computational Aspects

The characteristic equation of the natural frequency  was solved numerically to obtain the
roots using Mathematica 9.0. A frequency increment of 1/200,000, starting at zero, was used to
narrow down the search domain around the root, and then the secant method was used to obtain
the roots. Due to the high harmonics during the impact, high precision (350 decimal digits) was
used to obtain the roots and the response. For all cases, 500 symmetric mode shapes were used to
obtain the response. To decrease the analysis time, symmetric geometry was also utilized, i.e.,
integrations were carried out only over half the pole and column. In a desktop computer (3 GHz
Intel Core i5, with four cores and 12 GB RAM memory, running Windows 8.1), the time taken
for each phase is approximately 45 minutes, resulting in total times of the collision event
between 720 minutes (for 6 m steel column with 16 phases) to 6570 minutes (for 6 m wood
column with 146 phases).

In addition to the analytical solution described above, a finite element solution was also
utilized. The analytical solution was used to verify the finite element solution. For the finite

99
element model, both the pole and the column are modeled using a standard shear-deformable
beam element that uses the exact static stiffness. A lumped mass matrix was used. It has been
shown that the time history of the impact force can be influenced by the rotary inertia (Khowitar
et al. 2014), so lumped rotational inertia was included in the mass matrix.

Several runs were performed to assess the correctness of the FEM solution, as well as to
determine the mesh and time step requirements for accurate simulations. The converged mesh
consisted of 4000 elements for the pole and 3000 elements for the longest column. The Newmark
method was used for a direct integration of the equations of motion. For all cases, a
nondimensional time step of 1.5x10-4 was used. Figure 34 compares the analytical and finite
element force-time histories for a 6-m wood pole hitting a 3.6 m concrete column. The results
obtained from the two approaches match very well.

100
 1.5

Analytical
FEA

1
Nondimensional force

0.5

0
0 1 2 3 4 5 6
Nondimensional time

Figure 34: Force time history for 3.6 m concrete column obtained from analytical and FEA
solution

4.8 Summary
The linear response of a beam hitting transversely a fixed-fixed column is studied in this
paper. The results presented are for a 9 m wood pole hitting transversely concrete, steel, and
wood columns having lengths of 3.6 m and 6 m. The pole moves with an initial velocity and hits
laterally the column at midspan. Multiple impacts were considered and the whole collision event
is divided into contact and separation phases. Results from the analytical solution were used to
verify the finite element solution. In all cases excellent matches were obtained. The following
conclusions may be made.

101
1. The initial peak force is well-predicted by Eq. (4.13). However, the maximum impact
force can reach up to 1.5 times the initial peak force.

2. Variation in initial peak impact force depends mainly on the mass ratio of the column and
the pole.

3. The instantaneous impact force is dominated by shear deformation either in the pole or in
the column, or both.

4. For most of the cases, there is a high frequency pounding after a fairly large separation
phase. However, for the most flexible (long wood) column, the durations of all contact
and separation phases are small due to the high frequency pounding between the pole and
the column.

5. For concrete and steel columns, the sum of the bending strain energy and the kinetic
translational energy in the pole constitutes almost 90% of the total energy during the
collision event.

6. For the soft (wood) column, a significant part of the total energy is absorbed by the
bending strain energy and translational kinetic energy of the column during the collision
event.

7. For the most rigid (short concrete) column, only 1.5% of the initial energy is transferred
to the column. However, for the most flexible (long wood) column, almost 7.2% of the
initial energy is transferred to the column. In this case, at the time of separation most of
the energy transferred to the column is absorbed by bending strain energy at final
separation.

8. The total impulses for the concrete and steel columns are almost 1.6 times the initial
momentum of the pole. For the wood columns, the total impulse is almost 1.8 times the
initial momentum of the pole. This indicates that the final mass-averaged rebound
velocity of the pole is almost 60% of the initial velocity for the concrete and steel
columns, and 80% of the initial velocity for the wood columns.

102
 Chapter 5
Additional Scenarios

5.1 Introduction
In this chapter, additional scenarios for axial and transverse impacts that supplement the
results in the previous two chapters are considered. The axial impact simulations presented in
Chapter 3 only considered the first contact phase. In this chapter, the same scenarios are
simulated but through the entire impact event, capturing the multiple impacts, if any. The
maximum impact force, which normally does not occur during the first contact phases, is sought.
Moreover, the main differences between cases that exhibit single and multiple impacts will be
clarified from the energy exchange. For transverse impact, free-end and pin-end column
boundary conditions are considered to supplement the fixed-end boundary conditions in Chapter
4. The free-end condition, while not particularly relevant for a pole hitting a column, has some
interest in the transverse impact of two poles, for example.

In attempt to assess the values of the initial peak force obtained herein for design purposes, it
is useful to give an insight on whether applying this force as a static load can bound the dynamic
response. The values of the maximum shear force and bending moment are obtained by applying
a static load of a value equal to the initial peak force at mid-span of a fixed-end column. These
values are then compared to the values of the shear force and the bending moments from the
dynamic analyses.

As compared to the analytical results in the previous chapters, the results in this chapter were
obtained using a finite element program. As shown in Section 4.7, if modeled appropriately the
finite element solutions agree very well with the analytical solutions. For the long pole, 4000
elements were used and 3000 elements were used for the long column. The Newmark method
was used for direct integration of the equations of motion. For all cases, a time step of 10 -6 s was
used.

103
 5.2 Axial Impact
Axial impact will be discussed for a wood pole hitting concrete, steel, and wood columns.
For each column material there will be four cases C1P1, C1P2, C2P1, and C2P2. C1 indicates a
column of length 3.6 m, C2 indicates a column of length 6 m, P1 indicates a pole of 3 m, and P2
indicates a pole of 9 m. These are the same cases as considered in Chapter 3.

5.2.1 Wood pole hitting concrete column

For the axial impact of a pole hitting a concrete column, the collision event is characterized
by a single impact. The reason for this is that the high stiffness of the column relative to the pole
leads to very small displacements of the column at the impact point. Therefore, the force-time
plots are not repeated here because they are identical to those presented in Chapter 3, Figure 18.

Figure 35 shows the time evolution of the energy components of the pole and the column.
The time is nondimensionalized with respect to tax  2Lp / coP , which is the time it takes for the

axial wave in the pole to return back to the impact point. The energy is nondimensionalized with
respect to the initial kinetic energy of the pole. The symbols in the plots are: TE; total energy,
TKEp; translational kinetic energy of the pole, ASEp; axial strain energy of the pole, TKEc;
translational kinetic energy of the column, RKEc; rotational kinetic energy of the column, BEc;
bending energy of the column, SEc; shear energy of the column, and SprE; strain energy of the
spring. All impacts occur at the center of the column.

It can be seen from Figure 35 that most of the kinetic energy of the pole is transferred into
axial strain energy in the pole. This is due to the relatively high stiffness of the column, which
leads to higher deformation in the pole relative to the column. It is noticed also that the bending
energy of the column is smaller than the translational kinetic energy of the column for all cases
except for the case C1P2, in which the shear wave in the column reaches the impact point at 0.45
tax and causes the impact force to increase. As stated previously, shear energy of the column is

much larger than the bending energy of the column for a very short time after the first impact,
which indicates that the initial peak impact force is dominated by the local shear deformation of
the column at the impact point. In addition, kinetic and axial energy of the pole reaches their
minimum and maximum values, respectively, when the axial wave in the pole reaches its far end.

104
 1.2 1.2
TE TE
TKEp TKEp
ASEp ASEp
1 TKEc 1 TKEc
RKEc RKEc
BEc BEc
SEc SEc
SprE SprE
0.8 0.8

Nondimensional energy
Nondimensional energy

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Nondimensional time Nondimensional time

(a) C1P1 (b) C1P2

1.2 1.2
TE TE
TKEp TKEp
ASEp ASEp
1 TKEc TKEc
RKEc 1
RKEc
BEc BEc
SEc SEc
SprE SprE
0.8
Nondimensional energy

0.8
Nondimensional energy

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Nondimensional time Nondimensional time

(c) C2P1 (d) C2P2

Figure 35: Component energies and total energy for a wood pole hitting axially a concrete
column

Figure 36 shows the impulse-time history. The impulse is nondimensionalized with respect to
the initial momentum of the pole. It can be seen that the total impulse at the final separation is
approximately 1.8 times the initial momentum of the pole. This implies the mass-averaged
rebound velocity of the pole is around 80% of the initial velocity of the pole.

105
 As discussed in the previous section, most of the energy is transferred into axial strain energy
of the pole (almost 80%), and since the impact is elastic, the pole returns to its original shape
after separation occurs, releasing the strain energy in the form of kinetic energy. Thus, the
rebound velocity of the pole is around 80% of the initial velocity of the pole.

2 2

1.8 1.8

1.6 1.6

1.4 1.4
Nondimensional impulse

Nondimensional impulse
1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Nondimensional time Nondimensional time

(a) C1P1 (b) C1P2

2 2

1.8 1.8

1.6 1.6

1.4 1.4
Nondimensional impulse
Nondimensional impulse

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Nondimensional time Nondimensional time

(c) C2P1 (d) C2P2

Figure 36: Impulse for a wood pole hitting axially a concrete column

106
 5.2.2 Wood pole hitting steel column
For the scenario of a wood pole hitting a steel column, the only case that exhibits multiple
impacts is C2P2. In this case, the flexibility of the steel column is sufficient to allow the column
to vibrate with larger amplitudes. Also, the long pole has a bigger mass, increasing the inertia of
the pole. The force-time plots for the cases of C1P1, C1P2, and C2P1 are shown in Chapter 3,
Figure 19.

Figure 37 shows the force-time history of the case C2P2. The force is nondimensionalized
with respect to the maximum instantaneous force obtained from Eq. (3.15) and the time is
nondimensionalized with respect to tax  2Lp / coP . It can be seen that in the first contact phase,

there is a sharp decrease and then increase in the impact force at the time when the shear wave in
the column returns to the impact point. The plot shows also that the maximum impact force
occurs in the second contact phase and it is approximately 1.6 times the initial peak force.

1.8

1.6

1.4

1.2
Nondimensional force

0.8

0.6

0.4

0.2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Nondimensional time

Figure 37: Force-time history for case C2P2 of a wood pole hitting axially a steel column

Figure 38 shows the energy components versus time. It can be seen that for the cases C1P1
and C2P1, the pole is short and separation occurs before the shear wave in the column returns to
the impact point. For these two cases, the impact does not cause much bending deformation in

107
the column due to the relatively small mass of the pole. Hence, the translational kinetic energy of
the column is larger than the bending energy of the column throughout the entire collision event.
It is interesting also to note that the shear energy for these two cases exceeds the bending energy,
which shows that most of the deformation in the column is due to shear rather than bending.
However, for the case of C1P2, the bending energy of the column exceeds the translational
kinetic energy of the column. Also, when the shear wave in the column returns to the impact
point, the impact force experiences a sharp increase. Moreover, the shear energy of the column is
always larger than the bending energy during the collision event. For the first three cases, there is
only a single impact and most of the initial kinetic energy of the pole is transferred into strain
energy of the pole. For the last case C2P2, at the beginning of the impact, the axial strain energy
of the pole absorbs most of the kinetic energy of the pole, and reaches its peak value at t  0.5 .
It can be seen that the bending energy of the column has a small value, but it grows gradually
until it reaches its peak value at t  1 . The shear energy exceeds the bending energy shortly after
the first impact, which emphasizes the dependence of the initial peak force on the local shear
deformation of the column. The translational kinetic energy of the column attains its maximum
value at the beginning of the second contact phase, and then decays rapidly until the final
separation.

1.2 1.2
TE TE
TKEp TKEp
ASEp ASEp
1 TKEc 1 TKEc
RKEc RKEc
BEc BEc
SEc SEc
SprE SprE
0.8 0.8
Nondimensional energy
Nondimensional energy

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Nondimensional time Nondimensional time

(a) C1P1 (b) C1P2

108
 1.2 1.2
TE TE
TKEp TKEp
ASEp ASEp
1 TKEc 1 TKEc
RKEc RKEc
BEc BEc
SEc SEc
SprE SprE
0.8 0.8
Nondimensional energy

Nondimensional energy
0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Nondimensional time Nondimensional time

(c) C2P1 (d) C2P2

Figure 38: Energy components and total energy for a wood pole hitting axially a steel column

Figure 39 shows the time evolution of the impulse. It can be seen that for cases C1P1 and
C2P1, the final impulse is approximately 1.5 times the initial momentum of the pole. However,
for the case C1P2, the final impulse is around 1.8 times the initial momentum of the pole. In this
case, the slope of the curve increases when the shear wave returns to the impact point and the
force starts to increase sharply. Figure 39 (d) for the case C2P2 shows that the final impulse is
approximately 1.7 times the initial momentum of the pole, which indicates that the mass-
averaged rebound velocity of the pole is approximately 70% of the initial velocity.

109
 2 2

1.8 1.8

1.6 1.6

1.4 1.4

Nondimensional impulse
Nondimensional impulse

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
Nondimensional time Nondimensional time

(a) C1P1 (b) C1P2

2 2

1.8 1.8

1.6 1.6

1.4 1.4
Nondimensional impulse

Nodimensional impulse

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Nondimensional time Nondimensional time

(c) C2P1 (d) C2P2

Figure 39: Impulse for a wood pole hitting axially a steel column

5.2.3 Wood pole hitting wood column

All cases of axial impact of a pole hitting a wood column involve multiple impacts due to the
relatively small stiffness of the column. Figure 40 shows the force-time histories. The time is
nondimensionalized with respect to tsh  L / cs , the time it takes for a shear wave in the column to

110
return to the impact point. The results show that the maximum impact force always occurs near
the end of the collision event and can reach up to 1.9 times the initial peak force. It can be
noticed that for the case of a short pole, C1P1, and C2P1, the duration of the entire collision
event is smaller than that for the long pole cases C1P2 and C2P2, which indicates that the time of
the final separation depends on the wave propagation in the pole.

2 2

1.8 1.8

1.6 1.6

1.4 1.4

Nondimensional force
Nondimensional force

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 6 7
Nondimensional time Nondimensional time

(a) C1P1 (b) C1P2

2 2

1.8 1.8

1.6 1.6

1.4 1.4
Nondimensional force

Nondimensional force

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 6 7 8
Nondimensional time Nondimensional time

(c) C2P1 (d) C2P2

Figure 40: Impact force for a wood pole hitting axially a wood column

111
 Figure 41 shows the time evolution of the energy components. It can be seen that the axial
strain energy of the pole is very small, and the translational kinetic energy of the column absorbs
most of the kinetic energy of the pole at the beginning of the impact. This indicates that the
displacement of the column at the impact point prevents much of the axial deformation of the
pole as they move together in the same direction. The shear energy of the pole at the very
beginning of the impact exceeds the bending since the deformation in the column is mostly due
to shear. The bending energy then builds up gradually and attains its local peak value
approximately around half of the final separation time.

1.2 1.2
TE TE
TKEp TKEp
ASEp ASEp
1 TKEc TKEc
1
RKEc RKEc
BEc BEc
SEc SEc
SprE SprE
0.8
Nondimensional energy

0.8
Nondimensional energy

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 6 7
Nondimensional time Nondimensional time

(a) C1P1 (b) C1P2

112
 1.2 1.2
TE TE
TKEp TKEp
ASEp ASEp
1 TKEc TKEc
1
RKEc RKEc
BEc BEc
SEc SEc
SprE SprE
0.8
Nondimensional energy

0.8

Nondimensional energy
0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 6 7 8
Nondimensional time Nondimensional time

(c) C2P1 (d) C2P2

Figure 41: Energy components and total energy for a wood pole hitting axially a wood column

Figure 42 shows the time evolution of the impulse. The final impulse ranges between 1.7-1.9
times the initial momentum of the pole, and thus the mass-averaged rebound velocity of the pole
is between 70% and 90% of the initial velocity.

2 2

1.8 1.8

1.6 1.6

1.4 1.4
Nondimensional impulse

Nondimensional impulse

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 6 7
Nondimensional time Nondimensional time

(a) C1P1 (b) C1P2

113
 2 2

1.8 1.8

1.6 1.6

1.4 1.4

Nondimensional impulse
Nondimensional impulse

1.2 1.2

1 1

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 0.5 1 1.5 2 2.5 3 3.5 4 0 1 2 3 4 5 6 7 8
Nondimensional time Nondimensional time

(c) C2P1 (d) C2P2

Figure 42: Impulse for a wood pole hitting axially a wood column

5.3 Transverse Impact

Multiple impacts are investigated for a 9 m wood pole hitting transversely a 6 m wood
column having free and pinned end boundary conditions. The time is nondimensionalized with
respect to tsh  Lp / csp , the time it takes for a shear wave in the pole to return to the impact point.

Figure 43 shows the force-time histories of a wood pole hitting a free-end wood column (left)
and pinned-end wood column (right). The force is nondimensionalized with respect to the initial
peak force obtained from Eq. (4.13). It can be seen from Figure 43 (a) that the maximum force
during the entire collision event almost equals the peak instantaneous force in the first contact
phase. This is because the free-end column vibrates with lower frequencies than the end-
restrained column. The entire column moves in the direction of the initial velocity while
vibrating with smaller amplitudes after each impact. The figure shows also a separation phase
with a long duration from t  3.4 to t  8 where the column moves with a constant velocity. Of
all the cases for different end conditions of the column, the free-end boundary condition gives
the shortest duration of the entire collision event.

114
 Figure 43 (b) shows the case of a pinned-end column. The maximum force is almost 1.48
times the initial peak force. The free rotations at the ends of the column allow the column to
vibrate with higher amplitudes, and hence the duration of the entire collision event is longer than
that for the fixed-end column. The maximum force has a smaller value than that for the fixed-end
column because of the lower frequencies and the bigger displacements that the pinned-end
column undergoes at the impact point.

1.2 1.8

1.6

1
1.4

0.8 1.2

Nondimensional force
Nondimensional force

1
0.6
0.8

0.4 0.6

0.4
0.2
0.2

0 0
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 10 11 12 13
Nondimensional time Nondimensional time

(a) free-end column (b) pinned-end column

Figure 43: Impact force for 9 m wood pole hitting transversely a 6 m wood column

Figure 44 shows the time evolution of the energy components for the free-end column (left)
and for pinned-end column (right). It can be seen from Figure 44 (a) that most of the kinetic
energy of the pole is transferred into translational kinetic energy of the column, which exceeds
the bending energy of the column throughout the whole collision event. It is noticed that between
t  3.4 and t  8 the column moves with constant velocity in the same direction of the initial
velocity. The bending energy of the column exceeds the bending energy of the pole for a short
time after the first impact, and then the bending energy of the pole exceeds the bending energy of
the column and reaches its peak value at half of the duration of the collision event. The shear
energy of the column is very small compared to the other components and the bending energy
becomes very small starting from the fifth contact phase at t  2.7 .

115
 Figure 44 (b) shows that the translational kinetic energy of the column exceeds other energy
components for a short time after the initial impact and reaches its maximum value at t  1 . It is
observed that most of the kinetic energy of the pole is transferred into bending energy of the
column. The shear energy of the column exceeds the bending energy of the column for a very
short time after the initial impact, then the bending energy builds up until it reaches its maximum
value almost at half of the impact duration.

1.2 1.2
TE TE
TKEp TKEp
BEp BEp
SEp 1 SEp
1 TKEc
TKEc
BEc BEc
SEc SEc
0.8

Nondimensional energy
0.8
Nondimensional energy

0.6 0.6

0.4 0.4

0.2 0.2

0 0
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 10 11 12 13
Nondimensional time Nondimensional time

(a) free-end column (b) pinned-end column

Figure 44: Energy components and total energy for a 9 m wood pole hitting transversely a 6 m
wood column

Figure 45 shows the impulse-time history for a free-end column (left) and pinned-end
column (right). The impulse is nondimensionalized with respect to the initial momentum of the
pole. It can be noticed from Figure 45 (a) that the final impulse is only approximately 0.6 times
the initial momentum of the pole. This indicates that the final mass-averaged velocity of the pole
is almost 60% of the initial velocity. It also indicates that the pole and the column are moving in
the same direction of the initial velocity throughout the entire collision event.

Figure 45 (b) shows that for the pinned-end column, the final impulse is almost 1.9 times the
initial momentum of the pole. This indicates that the mass-averaged rebound velocity is almost
90% of the initial velocity.

116
 0.6 2

1.8

0.5
1.6

1.4
Nondimensional impulse

Nondimensional impulse
0.4

1.2

0.3 1

0.8

0.2
0.6

0.4
0.1

0.2

0 0
0 1 2 3 4 5 6 7 8 9 0 1 2 3 4 5 6 7 8 9 10 11 12 13
Nondimensional time Nondimensional time

(a) free-end column (b) pinned-end column

Figure 45: Impulse for a 9 m wood pole hitting transversely a 6 m wood column

5.4 Maximum Shear and Bending Envelopes

In attempt to assess the values of the initial peak force obtained herein for design purposes, it
is useful to give an insight on whether applying this force as a static load bounds the dynamic
response. The values of the maximum shear force and bending moment are obtained by applying
a static load of a value equal to the initial peak force at mid-span of a fixed-end column. These
values are then compared to the values of the shear force and the bending moments from the
dynamic analyses. The initial peak force for the axial impact, F0 is obtained from Eq. (3.15) and

for transverse impact from Eq. (4.13). The dynamic amplification factors are Rs  SFD / SFS and

Rb  BM D / BM S
, where
SFD is the maximum dynamic shear force, BM D is the maximum

dynamic bending moment, SFS  F0 / 2 is the maximum static shear force, BM S  F0 Lc / 8 is the

maximum static bending moment, and Lc is the length of the column.

117
 5.4.1 Envelope versus time

5.4.1.1 Axial Impact

Figure 46 shows the maximum shear and bending moment envelopes with respect to the
nondimensional time for axial impact. The nondimensionalization of the time depends on the
impact case where tax , t sh and Ci Pi were defined previously.

It can be seen that the case C1P2 gives the maximum shear amplification factor for single
impact, and C2P2 for multiple impacts. For all cases, case C1P2 always gives the maximum
bending moment amplification factor. This is due to the high frequency vibrations of the column
that depend on the stiffness of the column and inertia of the pole. For a single impact, case C1P2
has the maximum bending and shear force amplification factors. For steel column, C1P2 has the
maximum moment, and C2P2 has the maximum shear due to the multiple impacts. For wood
column, C2P2 gives the maximum shear force, and C1P2 gives the maximum bending moment.
For all column materials, the dynamic shear force equals the static shear force shortly after the
first impact. For concrete and steel columns, there is a jump in the maximum shear force when
the shear wave returns to the impact point. For all cases, the maximum shear force amplification
factor is 3.87 (refer to Table 6 for maximum dynamic amplification factors) and it is attained for
the case of steel column C2P2.

3 2.5
C1P1 C1P1
C1P2 C1P2
C2P1 C2P1
2.5 C2P2 C2P2
2

1.5
s

1.5
R

1
1

0.5
0.5

0 0
0 0.2 0.4 0.6 0.8 1 1.2 0 0.2 0.4 0.6 0.8 1 1.2
t/t t/t
ax ax

(a) Concrete column

118
 4 2
C1P1 C1P1
C1P2 C1P2
3.5 C2P1 C2P1
C2P2 C2P2

3 1.5

2.5
s
R

b
2 1

R
1.5

1 0.5

0.5

0 0
0 0.5 1 1.5 2 2.5 0 0.5 1 1.5 2 2.5
t/t t/t
ax ax

(b) Steel column

3 1.4
C1P1 C1P1
C1P2 C1P2
C2P1 C2P1
C2P2 1.2 C2P2
2.5

1
2

0.8
s

1.5
R

0.6

1
0.4

0.5
0.2

0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
t/t t/t
sh sh

(c) Wood column

Figure 46: Maximum shear force (left) and bending moment (right) amplification factors for
axial impact

5.4.1.2 Transverse Impact

It can be seen from Figure 47 that short concrete and steel columns give the largest shear
force amplification factors. Largest dynamic shear force always occurs at later impacts. It can be
noticed that in all cases the maximum dynamic shear force is equal to the maximum static shear

119
force shortly after the first impact. The maximum values of the shear force amplification factor
are 4.71 and 4.15 for short concrete and steel columns, respectively. The short concrete and steel
columns also give the maximum bending moment amplification factor with values greater than 1.
However, for all wood column cases as well as the long concrete and steel columns, the values of
the bending moment amplification factor are less than 1.

5 2

4
1.5

3
s

b
R

2 R

0.5
1

0 0
0 1 2 3 4 5 6 0 1 2 3 4 5 6
Nondimensional time Nondimensional time

(a) 3.6 m concrete column

4 1

3.5

0.8
3

2.5
0.6
b
s

2
R
R

0.4
1.5

1
0.2

0.5

0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Nondimensional time Nondimensional time

(b) 6 m concrete column

120
 5 1.5

4 1.2

3 0.9
s

b
R

R
2 0.6

1 0.3

0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Nondimensional time Nondimensional time

(c) 3.6 m steel column

3 1

2.5
0.8

2
0.6
b
s

1.5
R
R

0.4

0.2
0.5

0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Nondimensional time Nondimensional time

(d) 6 m steel column

121
 3.5 1

3
0.8

2.5

0.6
2
s

b
R

R
1.5
0.4

0.2
0.5

0 0
0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8
Nondimensional time Nondimensional time

(e) 3.6 m wood column

2.5 0.6

0.5
2

0.4
1.5
s

b
R

0.3
R

0.2

0.5
0.1

0 0
0 1 2 3 4 5 6 7 8 9 10 11 0 2 4 6 8 10 12
Nondimensional time Nondimensional time

(f) 6 m wood column

Figure 47: Maximum shear force (left) and bending moment (right) amplification factors for
transverse impact

122
 5.4.2 Envelope versus space

5.4.2.1 Axial Impact

It can be seen from Figure 48 that for a concrete column, the maximum dynamic shear force
equals the maximum static shear force at the impact point. Case C1P2 gives the maximum shear
force amplification factor at the fixed supports. For a steel column, C1P1 and C2P1 (short pole)
give Rs  1 at the impact point. The shear force amplification factor for the steel column for case
C1P2 is greater than that for C2P2 up to 1/3 of the column length measured from the impact
point, after that it increases for case C2P2 and reaches its maximum value at the fixed support.
For all wood column cases, the dynamic shear force is greater than the static shear force at the
impact point, and C2P2 gives the maximum shear force amplification factor at the fixed end.

For bending moment, case C1P2 always gives the maximum moment at the fixed end. For
concrete column and steel column, in all cases except C2P2, the minimum value of the dynamic
bending moment occurs around 1/4 the column length and its value is around 0.25. This is
because the moment at 1/4 the column length vanishes if a concentrated load is applied at mid-
span of the column. However for wood column, the minimum value of Rb is around 0.2 due to
the complexity of harmonics. It can be concluded that the stiffer the column, the higher the
bending, and multiple impacts always magnify the dynamic shear force and bending moment at
later impacts.

123
 3 2.5
C1P1 C1P1
C1P2 C1P2
C2P1 C2P1
C2P2 C2P2
2.5
2

2
1.5

b
s

R
1.5
R

0.5
0.5

0 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
L/L L/L
c c

(a) Concrete column

4 2
C1P1 C1P1
C1P2 C1P2
C2P1 C2P1
3.5 C2P2 C2P2

1.5
3

2.5
s

1
R

1.5
0.5

0.5 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
L/L L/L
c c

(b) Steel column

124
 3 1.4
C1P1 C1P1
C1P2 C1P2
C2P1 C2P1
C2P2 1.2 C2P2
2.5

2
0.8
s

b
R

R
0.6
1.5

0.4

1
0.2

0.5 0
0 0.1 0.2 0.3 0.4 0.5 0 0.1 0.2 0.3 0.4 0.5
L/L L/L
c c

(c) Wood Column

Figure 48: Maximum shear force (left) and bending moment (right) amplification factors for
axial impact

5.4.2.2 Transverse Impacts

Figure 49 shows the amplification factor of maximum shear force and bending moment. The
symbols in legends are: cC1 is short concrete column, cC2 is long concrete column, sC1 is short
steel column, sC2 is long steel column, wC1 is short wood column, and wC2 is long wood
column.

It is clear from Figure 49 that stiffer columns always give higher dynamic amplification
factors for moment and shear. This is due to the high frequency vibration. In addition, for cases
involving multiple impacts, the dynamic amplification factor increases always at later impacts.
Short concrete columns always give the largest dynamic shear force and bending moment. For
all cases, the dynamic shear force is greater than the static shear force at the impact point.
Maximum shear force amplification factor is attained for short concrete column at the fixed
support.

In all cases except wood, the dynamic bending moment is less than the static bending
moment for long columns. Maximum bending occurs at the fixed end for short concrete column
with a value of 1.88. The minimum value is attained around 1/4 the column length and its value

125
ranges between 0.2 and 0.55. Table 6 to Table 9 show the dynamic amplification factors for
different impact scenarios.

5
cC1
cC2
sC1
sC2
4 wC1
wC2

3
s
R

0
0 0.1 0.2 0.3 0.4 0.5
L/L
c

(a) Shear force

126
 2
cC1
cC2
sC1
sC2
wC1
wC2
1.5

1
R

0.5

0
0 0.1 0.2 0.3 0.4 0.5
L/L
c

(b) Bending moment

Figure 49: Maximum amplification factors for transverse impact

Table 6: Maximum dynamic amplification factors for shear force for axial impact

Case C1P1 C1P2 C2P1 C2P2

Concrete Column 2.44 2.53 1.22 1.73
Steel Column 1.74 3.14 1.90 3.87
Wood Column 2.75 2.61 2.03 2.29

127
 Table 7: Maximum dynamic amplification factors for bending moment for axial impact

Case C1P1 C1P2 C2P1 C2P2

Concrete Column 0.73 2.11 0.44 0.93

Steel Column 1.46 1.99 0.64 1.82
Wood Column 0.69 1.32 0.45 0.71

Table 8: Maximum dynamic amplification factors for shear force for transverse impact

Case 3.6 m column 6 m column

Concrete Column 4.71 3.68

Steel Column 4.15 2.63
Wood Column 3.38 2.20

Table 9: Maximum dynamic amplification factors for bending moment for transverse impact

C1 C2

Concrete Column 1.88 0.89

Steel Column 1.46 0.88

Wood Column 0.82 0.59

5.5 Summary
Additional scenarios were investigated for multiple impacts. In the first part, the linear
response of a wood pole hitting axially a fixed-end column is studied. The material of the
column can be concrete, steel, or wood and its length is 3.6 m or 6 m. The length of the pole is 3
m or 6 m. The impact occurs at the mid-span of the column. In the second part, the linear
response of a 9 m wood pole hitting transversely a 6 m column is studied. The column end

128
boundary conditions can be free or pinned. The impact also occurs at the mid-span of the pole
and the column. All the results presented in this chapter are obtained using FEM solution. The
following conclusions may be made.

1. For a wood pole hitting axially a concrete column, the collision event is characterized by
a single impact for all cases. The reason for this is that the high stiffness of the column
causes the column to vibrate with high frequencies and small amplitudes at the impact
point. Most of the kinetic energy of the pole is transferred into axial strain energy of the
pole which attains its maximum value when the axial wave in the pole hits the far end of
the pole. The mass-averaged rebound velocity of the pole is approximately 80% of the
initial velocity.
2. For a wood pole hitting axially a steel column, only the case of a 9 m pole and a 6 m
column exhibits multiple impacts consisting of two contact phases. The maximum impact
force occurs in the second contact phase and is almost 1.6 times the instantaneous peak
force. For the other cases having a single impact, most of the deformation in the column
is due to shear and most of the kinetic energy of the pole is transferred into strain energy
of the pole. For the case of multiple impacts, most of the energy is transferred into axial
strain energy of the pole shortly after the first impact. After that, the bending energy and
the translational kinetic energy of the column absorb most of the kinetic energy of the
pole.
3. For the case of a wood pole hitting axially a wood column, all the cases experience
multiple impacts. The maximum force occurs near the end of the final separation and
reaches up to 1.9 times the instantaneous impact force. The axial strain energy of the pole
is very small throughout the entire collision event. Most of the kinetic energy of the pole
is transferred into bending energy of the column. The mass-averaged rebound velocity of
the pole ranges between 70% and 90% of the initial velocity.
4. For a 9 m wood pole hitting transversely a 6 m wood column, the maximum force for the
free-end column is almost equal to peak instantaneous impact force. However, for the
case of pinned-end column, the impact force reaches up to 1.48 times the peak
instantaneous force. For the case of a free-end column, most of the energy of the column
is translational kinetic energy and the column and the pole move in the same direction of

129
 the initial velocity throughout the entire collision event. On the other hand, most of the
kinetic energy of the pole is transferred into bending energy of the pole and the column.
The impulse at the final separation is 0.6 times the initial momentum of the pole in the
case of free-end column. However, it reaches up to 1.9 times the initial momentum of the
pole for the pinned-end column.
5. For axial and transverse impacts, cases involving stiffer columns and higher inertia of the
pole always give higher shear force and bending moment dynamic amplification factors.
In addition, for cases involving multiple impacts, the dynamic effect is magnified always
at later impacts. Hence, cases of short fixed-end column and long pole give the maximum
dynamic shear force for single impact and give the maximum bending moment for single
and multiple impacts. However, cases of long column and long pole give the maximum
dynamic shear force.

130
 Chapter 6
Conclusion
The elastic impact of a translating flexible pole is studied herein. Three scenarios are
considered: 1) transverse impact against a rigid stop, 2) longitudinal impact against a flexible
column, and 3) transverse impact against a flexible column. Based on Timoshenko beam theory,
an analytical solution method using mode superposition for the coupled spring-pole or column-
pole system is presented. Any physical set of boundary conditions can be accommodated for the
pole and the column.

For all cases involving axial impacts, the maximum initial impact force is governed by the
local shear deformation in the column and the axial deformation in the pole. However, for
transverse impacts, the maximum initial impact force is governed by the local shear deformation
in the pole and the column. A simple formula for the maximum initial force is derived and shown
to be quite accurate. In no case is the total mass of the pole significant to the initial peak force.
Indeed, based on Euler-Bernoulli beam theory the initial impact force is unbounded as the spring
stiffness increases whereas Timoshenko beam theory has a clear limiting value for the initial
impact force. The impact duration depends on the wave propagation in the pole or the column.

6.1 Response of a Beam Hitting Transversely a Stop

An analytical solution for the linear response of a Timoshenko beam impacting a stop
modeled as a spring has been presented. The solution, which is based on modal superposition,
admits contact at an arbitrary point within the beam span, as well as beam rotational as well as
translational initial velocities. As such, the present solution is a generalization as compared to
previous solutions. It has been shown that the initial impact is dominated by shear behavior, and
that for stiff stops, Euler-Bernoulli beam theory can significantly overestimate the initial impact
force and underestimate the contact duration. Indeed, based on Euler-Bernoulli beam theory the
impact force grows without bounds as the stop stiffness increases, whereas Timoshenko beam
theory shows the impact force is bounded. As a general rule, it is recommended that Timoshenko
beam theory rather than Euler theory be used for impact studies. Parameter studies have shown

131
that the impact force time history is controlled primarily by the ratio of the stop stiffness to the
beam shear stiffness. For homogeneous beams, the sectional rotational inertia has little effect on
the impact forces.

6.2 Beam Response to Longitudinal Impact by a Pole

An analytical solution for the linear impact response of a pole impacting a column, or a
beam impacting a flexible axial stop, has been presented. The solution is based on
Timoshenko beam theory and uses modal superposition. Contact can occur at an arbitrary
point within the column span. The following conclusions may be made.

1. The initial peak impact force can be predicted by a simple formula, Eq.(3.15). In the case
of stiff columns, the pole properties govern the force; in the case of a stiff pole, the
column properties govern. In no case is the total mass of the projectile significant to the
initial peak force.
2. The impact duration is governed by either the longitudinal wave speed in the pole (for
stiff columns) or the shear wave speed in the column.
3. For cases in which the column properties are significant, the initial response is dominated
by shear behavior, and Timoshenko beam theory rather than Euler-Bernoulli beam theory
should be used. Indeed, based on Euler-Bernoulli beam theory the initial impact force is
unbounded as the stop stiffness increases, whereas Timoshenko beam theory has a clear
limiting value for the initial impact force.
4. The rotary inertia does not affect the initial peak impact force but it can have an effect on
the impact force time history after waves in the beam return to the impact point.
5. The energy plots provide a vivid picture of the complex interactions and energy exchange
taking place during the impact. For instance, of direct relevance to the present work, it
explains the importance of the shear mode at the beginning of the impact to set the impact
force and also depicts the relevance of the various energy exchanges in setting later
spikes in the force as well as the impact duration. The energy exchange also shows the
importance of the inertia of the column in absorbing a significant part of the initial kinetic
energy of the pole.

132
 6.3 Transverse Impact of a Horizontal Beam on a Vertical Column
The linear response of a beam hitting transversely a fixed-fixed column is studied in this
paper. The results presented are for a 9 m wood pole hitting transversely concrete, steel, and
wood columns having lengths of 3.6 m and 6 m. The pole moves with an initial velocity and hits
laterally the column at midspan. Multiple impacts were considered and the whole collision event
is divided into contact and separation phases. Results from the analytical solution were used to
verify the finite element solution. In all cases excellent matches were obtained. The following
conclusions may be made.

1. The initial peak force is predicted well by Eq. (4.13). However, the maximum impact
force can reach up to 1.6 times the initial peak force.
2. Variation in initial peak impact force depends mainly on the mass ratio of the column and
the pole.
3. The instantaneous impact force is dominated by shear deformation either in the pole or in
the column, or both.
4. For most of the cases, there is a high frequency pounding after a fairly large separation
phase. However, for the most flexible (long wood) column, the durations of all contact
and separation phases are small due to the high frequency pounding between the pole and
the column.
5. For concrete and steel columns, the sum of the bending strain energy and the kinetic
translational energy in the pole constitutes almost 90% of the total energy during the
collision event.
6. For the soft (wood) column, a significant part of the total energy is absorbed by the
bending strain energy and translational kinetic energy of the column during the collision
event.
7. For the most rigid (short concrete) column, only 1.5% of the initial energy is transferred
to the column. However, for the most flexible (long wood) column, almost 7.2% of the
initial energy is transferred to the column. In this case, at the time of separation most of
the energy transferred to the column is absorbed by bending strain energy at final
separation.

133
 8. The total impulses for the concrete and steel columns are almost 1.6 times the initial
momentum of the pole. For the wood columns, the total impulse is almost 1.8 times the
initial momentum of the pole. This indicates that the final mass-averaged rebound
velocity of the pole is almost 60% of the initial velocity for the concrete and steel
columns, and 80% of the initial velocity for the wood columns.

6.4 Additional Scenarios

Additional scenarios were investigated for multiple impacts. In the first part, the linear
response of a wood pole hitting axially a fixed-end column is studied. The material of the
column can be concrete, steel, or wood and its length is 3.6 m or 6 m. The length of the pole is 3
m or 6 m. The impact occurs at the mid-span of the column. In the second part, the linear
response of a 9 m wood pole hitting transversely a 6 m column is studied. The column end
boundary conditions can be free or pinned. The impact also occurs at the mid-span of the pole
and the column. All the results presented in this chapter are obtained using FEM solution. The
following conclusions may be made.

1. For a wood pole hitting axially a concrete column, the collision event is characterized by
a single impact for all cases. The reason for this is that the high stiffness of the column
causes the column to vibrate with high frequencies and small amplitudes at the impact
point. Most of the kinetic energy of the pole is transferred into axial strain energy of the
pole which attains its maximum value when the axial wave in the pole reaches the far end
of the pole. The mass-averaged rebound velocity of the pole is approximately 80% of the
initial velocity.
2. For a wood pole hitting axially a steel column, only the case of a 9 m pole and a 6 m
column exhibits multiple impacts, and in that case there are two contact phases. The
maximum impact force occurs in the second contact phase and is almost 1.6 times the
instantaneous peak force. For the other cases having a single impact, most of the
deformation in the column is due to shear and most of the kinetic energy of the pole is
transferred into strain energy of the pole. For the case of multiple impacts, most of the
energy is transferred into axial strain energy of the pole shortly after the first impact.

134
 After that, the bending energy and the translational kinetic energy of the column absorb
most of the kinetic energy of the pole.
3. For the case of a wood pole hitting axially a wood pole, all the cases experience multiple
impacts. The maximum force occurs near the end of the final separation and reaches up to
1.9 times the instantaneous impact force. The axial strain energy of the pole is very small
throughout the entire collision event. Most of the kinetic energy of the pole is transferred
into bending energy of the column. The mass-averaged rebound velocity of the pole
ranges between 70% and 90% of the initial velocity.
4. For a 9 m wood pole hitting transversely a 6 m wood column, the maximum force for the
free-end column is almost equal to the peak instantaneous impact force. However, for the
case of a pin-end column, the impact force reaches up to 1.48 times the peak
instantaneous force. For the case of a free-end column, most of the energy of the column
is translational kinetic energy and the column and the pole move in the same direction of
the initial velocity throughout the entire collision event. On the other hand, most of the
kinetic energy of the pole is transferred into bending energy of the pole and the column.
The impulse at the final separation is 0.6 times the initial momentum of the pole in the
case of the free-end column. However, it reaches up to 1.9 times the initial momentum of
the pole for the pin-end column.
5. For axial and transverse impacts, cases involving stiffer columns and higher inertia of the
pole always give higher shear force and bending moment dynamic amplification factors.
In addition, for cases involving multiple impacts, the dynamic effect is magnified always
at later impacts. Hence, cases of short fixed-end column and long pole give the maximum
dynamic shear force for single impact and give the maximum bending moment for single
and multiple impacts. However, cases of long column and long pole give the maximum
dynamic shear force.

135
 List of References
Alaska (2013). “Tsunami Debris and Other Marine Debris in Alaska.”
<http://dec.alaska.gov/commish/tsunami-debris/>. (July 10, 2013).

Arroyo-Caraballo, J. R. and Ebeling, R. M. (2006). “Glancing-blow impact forces by a barge

train on a lock approach wall.” Journal of Infrastructure Systems, 12,(5), 135-143.

ASCE (2010). “Minimum design loads for buildings and other structures.” ASCE/SEI 7-10,
American Society of Civil Engineers.

Boley, B. A. (1955). “An approximate theory of lateral impact on beams.” Journal of applied
mechanics, 22, 69-76.

Boley, B. A. and Chao, C. C. (1955). “Some solutions of the Timoshenko beam equations.”
Journal of Applied Mechanics, 22, 579-586.

Christoforou, A. P. and Yigit, A. S. (1998). “Effect of flexibility on low velocity impact

response.” Journal of Sound and Vibration, 217,(3), 563-578.

Consolazio, G. R., Cook, R. A., McVay, M. C., Cowan, D., Biggs, A., and Bui, L. (2006).
“Barge impact testing of the St. George Island causeway bridge.” UF 00026868/FDOT
BC-354 RPWO 76, University of Florida, Gainesville, FL.

Consolazio, G. R. and Cowan, D. R. (2005). “Numerically efficient dynamic analysis of barge

collisions with bridge piers.” Journal of Structural Engineering, 131,(8), 1256-1266.

Consolazio, G. R., Getter, D. J., and Davidson, M. T. (2009). “A static analysis method for
barge-impact design of bridges with consideration of dynamic amplification.” UF Project
No. 00068901/FDOT BD-545-85, University of Florida, Gainsville, FL.

Cowper, G. R. (1966). “The Shear Coefficient in Timoshenko's Beam Theory.” Journal of

Applied Mechanics, 33, 335-340.

136
Dorogy, A. and Rittel, A. (2008). “Transverse impact of free-free square aluminum beams: an
experimental-numerical investigation.” Journal of Impact Engineering, 35,(6), 569-577.

Ervin, E. K. (2009). “Vibro-impact behavior of two orthogonal beams.” Journal of Engineering

Mechanics, 135,(6), 10.1061//asce/0733-9399/2009/135:6/529, 529-537.

Ervin, E. K. and Wickert, J. A. (2007). “Experiments on a beam-rigid body structure repetitively

impacting a rod.” Nonlinear Dynamics, 50,(3), 10.1007/s11071-006-9180-3, 701-716.

Evans, G. R., Jones, B. C., McMillan, A. J., and Darby, M. I. (1991). “A new numerical method
for the calculation of impact forces.” Journal of Physics D: Applied Physics, 24, 854-858.

FEMA (2012). “Guidelines for design of structures for vertical evacuation from tsunamis, second
edition.” FEMA P646, Federal Emergency Management Agency.

Goyder, H. G. D. and Teh, C. E. (1989). “A study of the impact dynamics of loosely supported
heat exchanger tubes.” Journal of Pressure Vessel Technology, 111, 394-401.

Graff, K. F. (1991). Wave Motion in Elastic Solids, Dover, New York.

H.Kolsky (1963). “Stress Waves in Solids.” Dover, New York.

Haehnel, R. B. and Daly, S. F. (2002). “Maximum impact force of woody debris on floodplain
structures.” ERDC/CRREL TR-02-2, US Army Corp of Engineers, Engineer Reserach
and Development Center.

Harrison, H. R. and Nettleton, T. (1997). Advanced Engineering Dynamics, Elsevier.

Khowitar, E., Riggs, H. R., and Kobayashi, M. H. (2014). “Beam response to longitudinal impact
by a pole.” Journal of Engineering Mechanics, 10.1061/(ASCE)EM.1943-7889.0000765,
04014045.

Kobayashi, M. H., Genest, R., Riggs, H. R., and Paczkowski, K. (2012). “Simple hydroelastic
impact models for water-borne debris.” Proceedings of the Institution of Mechanical

137
 Engineers, Part M: Journal of Engineering for the Maritime Environment, 226,(2),
10.1177/1475090212437627, 170-179.

Langley, R. S. (2012). “The analysis of impact forces in randomly vibrating elastic systems.”
Journal of Sound and Vibration, 331,(16), 10.1016/j.jsv.2012.03.020, 3738-3750.

Matsutomi, H. (2009). “Method for estimating collision force of driftwood accompanying

tsunami inundation flow.” Journal of Disaster Research, 4,(6), 435-440.

Mikhaylov, Y. (2009). “Evaluation of prototypical reinforced concrete building performance

when subjected to tsunami loading.” MS, University of Hawaii, Honolulu 111 pp.

NRC (2004). Preventing earthquake disasters: The grand challenge in earthquake engineering:
A Research Agenda for the Network for Earthquake Engineering Simulation (NEES), The
National Academies Press, Washington, D.C.

Packzkowski , K., Riggs, H. R., Naito, C. J., and Lehmann, A. (2012) (2012). “A one-
dimensional model for impact forces resulting from high mass, low velocity debris.”
Structural Engineering and Mechanics, 46,(6), 831-847.

Paczkowski, K., Riggs, H. R., Naito, C. J., and Lehmann, A. (2012). “A one-dimensional model
for impact forces resulting from high mass, low velocity debris.” Structural Engineering
and Structural Mechanics, 42,(6), 831-847.

Piran Aghl, P., Naito, C., and Riggs, H. R. (2014). “Full-scale experimental study of impact
demands resulting from high mass, low velocity debris.” Journal of Structural
Engineering, 10.1061/(ASCE)ST.1943-541X.0000948, 04014006.

Riggs, H. R., Cox, D. T., Naito, C. J., Kobayashi, M. H., Piran Aghl, P., Ko, H. T.-S., and
Khowitar, E. (2013). ‘Water-driven debris impact forces on structures: experimental and
theoretical program.” 32nd Intl. Conf. on Ocean, Offshore and Arctic Engineering,
Nantes, France, ASME, OMAE2013-11128.

138
Robertson, I. N., Carden, L., Riggs, H. R., Yim, S., Young, Y. L., Paczkowski, K., and Witt, D.
(2010). “Reconnaissance following the September 29, 2009 tsunami in Samoa.”
UHM/CEE10-01, University of Hawaii, Honolulu.

Robertson, I. N., Riggs, H. R., Yim, S. C. S., and Young, Y. L. (2007). “Lessons from Hurricane
Katrina storm surge on bridges and buildings.” Journal of waterway, port, coastal, and
ocean engineering, 133,(6), 463-483.

Rossetto, T., Peiris, N., Pomonis, A., Wilkinson, S. M., Del Re, D., Koo, R., and Gallocher, S.
(2006). “The Indian Ocean tsunami of December 26, 2004: observations in Sri Lanka and
Thailand.” Natural Hazards, 42,(1), 10.1007/s11069-006-9064-3, 105-124.

S.H.Crandall, D. C. K., E.F.Kurtz Jr., D.C.Pridmore-Brown (1968). “Dynamics of Mechanical

and Electromechanical Systems.” McGraw-Hill Book Company, New York.

Timoschenko, S. (1914). “Zur frage nach der wirkung eines stoßes auf einen balken.” Zeitschrift
für Mathematik und Physik, 62, 198-209.

Toro, R. (2012). “Debris from the Japanese Tsunami.” <http://www.livescience.com/21067-

japan-tsunami-debris-tracking.html>.

Vinayaravi, R., Kumaresan, D., Jayaraj, K., Asraff, A. K., and Muthukumar, R. (2013).
“Experimental investigation and theoretical modelling of an impact damper.” Journal of
Sound and Vibration, 332,(5), 10.1016/j.jsv.2012.10.032, 1324-1334.

Wagg, D. J. and Bishop, S. R. (2002). “Application of non-smooth modelling techniques to the

dynamics of a flexible impacting beam.” Journal of Sound and Vibration, 256,(5),
10.1006/jsvi.5020, 803-820.

Wang, C. and Kim, J. (1996). “New analysis method for a thin beam impacting against a stop
based on the full continuous model.” Journal of Sound and Vibration, 191,(5), 809-823.

Wolfram. 2010. Mathematica, version 8.0, Wolfram Research, Inc., Champaign, IL.

139
Xing, Y., Qiao, Y., Zhu, D., and Sun, G. (2002). “Elastic impact on finite Timoshenko beam.”
Acta Mechanica Sinica, 18,(3), 252-263.

Y.C.Fung (1965). “Foundations of Solid Mechanics.” Prentice-Hall, Inc, New Jersy.

Yin, X. C., Qin, Y., and Zou, H. (2007). “Transient responses of repeated impact of a beam
against a stop.” International Journal of Solids and Structures, 44,(22-23),
10.1016/j.ijsolstr.2007.04.009, 7323-7339.

Yoshihara, H., Kubojima, Y., Nagaoka, K., and Ohta, M. (1998). “Measurement of the shear
modulus of wood by static bending tests.” Journal of Wood Science, 44, 15-20.

Zimmermann, K. A. (2012). “Hurricane Katrina: Facts, Damage & Aftermath.”

<http://www.livescience.com/22522-hurricane-katrina-facts.html>.

140