Vulnerability Assessment of Reinforced Concrete Columns ...eprints.qut.edu.au/43693/1/Herath_Thilakarathna_Thesis.pdf · Vulnerability assessment of reinforced concrete columns subjected

Vulnerability assessment of reinforced concrete columns subjected to vehicular impacts

i

Vulnerability Assessment of Reinforced Concrete

Columns Subjected to Vehicular Impacts

By

HMI Thilakarathna. MSc, BSc (Hons.)

A THESIS SUBMITTED TO THE SCHOOL OF URBAN DEVELOPMENT QUEENSLAND UNIVERSITY OF TECHNOLOGY IN PARTIAL

FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

December 2010


iii

DedicationDedicationDedicationDedication

To my parentsTo my parentsTo my parentsTo my parents with lovewith lovewith lovewith love


v

Abstract

Columns are one of the key load bearing elements that are highly susceptible to vehicle

impacts. The resulting severe damages to columns may leads to failures of the

supporting structure that are catastrophic in nature. However, the columns in existing

structures are seldom designed for impact due to inadequacies of design guidelines.

The impact behaviour of columns designed for gravity loads and actions other than

impact is, therefore, of an interest.

A comprehensive investigation is conducted on reinforced concrete column with a

particular focus on investigating the vulnerability of the exposed columns and to

implement mitigation techniques under low to medium velocity car and truck impacts.

The investigation is based on non-linear explicit computer simulations of impacted

columns followed by a comprehensive validation process. The impact is simulated

using force pulses generated from full scale vehicle impact tests. A material model

capable of simulating triaxial loading conditions is used in the analyses. Circular

columns adequate in capacity for five to twenty story buildings, designed according to

Australian standards are considered in the investigation. The crucial parameters

associated with the routine column designs and the different load combinations

applied at the serviceability stage on the typical columns are considered in detail.

Axially loaded columns are examined at the initial stage and the investigation is

extended to analyse the impact behaviour under single axis bending and biaxial

bending. The impact capacity reduction under varying axial loads is also investigated.

Effects of the various load combinations are quantified and residual capacity of the

impacted columns based on the status of the damage and mitigation techniques are

also presented. In addition, the contribution of the individual parameter to the failure

load is scrutinized and analytical equations are developed to identify the critical

impulses in terms of the geometrical and material properties of the impacted column.

In particular, an innovative technique was developed and introduced to improve the

accuracy of the equations where the other techniques are failed due to the shape of the

error distribution.


vi

Above all, the equations can be used to quantify the critical impulse for three

consecutive points (load combinations) located on the interaction diagram for one

particular column. Consequently, linear interpolation can be used to quantify the

critical impulse for the loading points that are located in-between on the interaction

diagram. Having provided a known force and impulse pair for an average impact

duration, this method can be extended to assess the vulnerability of columns for a

general vehicle population based on an analytical method that can be used to quantify

the critical peak forces under different impact durations. Therefore the contribution of

this research is not only limited to produce simplified yet rational design guidelines

and equations, but also provides a comprehensive solution to quantify the impact

capacity while delivering new insight to the scientific community for dealing with

impacts.


vii

Keywords

Dynamic analysis; Numerical simulation; Lateral impact; Circular column; Eccentric

loading; Bi-axial bending; Residual capacity; Analytical equations; Vulnerability

assessment.


ix

TABLE OF CONTENTS Abstract.........................................................................................................................v

Keywords....................................................................................................................vii

Table of contents..........................................................................................................ix

List of abbreviations....................................................................................................xv

List of symbols...........................................................................................................xvi

List of tables..............................................................................................................xxii

List of figures............................................................................................................xxii

Publications..............................................................................................................xxix

Statement of original authorship..............................................................................xxxi

Acknowledgements...............................................................................................xxxiii

1. INTRODUCTION ................................................................................................ 1-1

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

1.2 Aims and objectives ...................................................................................... 1-5

1.3 Hypotheses and research problems ............................................................... 1-6

1.4 Methodology ................................................................................................. 1-8

1.5 Thesis outline .............................................................................................. 1-12

2. LITERATURE REVIEW ................................................................................... 2-15

2.1 Characteristics of impact pulses .................................................................. 2-15

2.2 Behaviour of structural elements under impact loading .............................. 2-16

2.3 Dynamic impact tests on reinforced concrete columns ............................... 2-17

2.3.1 Columns subjected to soft impact ......................................................... 2-17

2.3.2 Columns subjected to hard impact ........................................................ 2-18

2.3.3 Columns subjected to axial impact ....................................................... 2-19

2.3.4 Shortcomings of the individual column tests ........................................ 2-20

2.4 Dynamic tests on reinforced concrete beams .............................................. 2-20

2.5 Behaviour of concrete under impact loads .................................................. 2-24

2.6 Dynamic properties of concrete and steel ................................................... 2-27

2.6.1 CEB-FIP specifications for concrete ..................................................... 2-27

2.6.2 Dynamic properties of steel .................................................................. 2-34


x

2.7 Interaction between reinforcement and concrete ......................................... 2-36

2.7.1 Static bond slip analysis ........................................................................ 2-36

2.7.2 Dynamic bond slip analysis................................................................... 2-38

2.8 Factors affecting ductility of concrete columns .......................................... 2-39

2.8.1 Effects of confinement on enhancement of the ductility and strength .. 2-39

2.8.2 Effects of concrete cover ....................................................................... 2-42

2.8.3 Compressive axial load level................................................................. 2-43

2.8.4 Combined effects of axial load and flexure........................................... 2-43

2.9 Shear capacity calculations and effects of axial load .................................. 2-45

2.10 Energy absorption characteristics under impact loads .......................... 2-47

2.11 Design practices and provisions of RF in critical sections .................... 2-49

2.11.1 Influence of the various parameters on confinement ............................ 2-52

2.11.2 Theoretical stress strain curves for confined concrete by transverse RF2-55

2.12 Effects of impact induced torsion in eccentrically loaded columns ...... 2-57

2.13 Impact reconstruction ............................................................................ 2-58

2.13.1 Application to accident reconstructions ................................................ 2-60

2.14 Design guidelines .................................................................................. 2-61

2.14.1 Dynamic design for impact ................................................................... 2-63

2.15 Knowledge gaps and literature review findings .................................... 2-64

3. FE MODELLING OF SHORT RC COLUMNS UNDER LATERAL IMPACT3-67

3.1 Introduction ................................................................................................. 3-67

3.2 Finite element modelling for impact problems ........................................... 3-68

3.3 Finite element modelling and selection of material models ........................ 3-69

3.3.1 Evaluation of Constitutive material models in LS-DYNA ................... 3-69

3.3.2 Theory on Mat_Concrete_Damage model ............................................ 3-70

3.3.3 Definition of compressive and tensile meridians at p < fc/3 ................. 3-72

3.3.4 Pressure cut-off and softening ............................................................... 3-73

3.3.5 Strain rate effect .................................................................................... 3-74

3.3.6 Equation of state .................................................................................... 3-75

3.3.7 Evaluation of LS-DYNA material model Mat_Brittle_Damage ........... 3-75

3.4 Development and validation of a numerical model of a RC column .......... 3-77


xi

3.4.1 Introduction ........................................................................................... 3-77

3.4.2 Experimental test set up ........................................................................ 3-78

3.4.3 Numerical simulation of the physical testing ........................................ 3-79

3.4.4 Convergence study and mesh discritization .......................................... 3-80

3.4.5 Contact algorithm and prevention of initial penetration ....................... 3-82

3.4.6 Validation of the finite element model using Mat_Concrete_Damage . 3-84

3.4.7 Material properties of steel.................................................................... 3-85

3.4.8 Load simulation for a dynamic system ................................................. 3-86

3.4.9 Hourglass energy and damping effects ................................................. 3-87

3.4.10 Procedure for axial load application ..................................................... 3-88

3.4.11 Confinement effects under strain gradient ............................................ 3-90

3.4.12 Numerical and experimental results for Mat_Brittle_Damage ............. 3-91

3.4.13 Comparison of numerical and experimental results for Mat_Concrete 3-92

_Damage .......................................................................................................... 3-92

3.5 Conclusions ................................................................................................. 3-95

4. IMPACT RECONSTRUCTION AND PARAMETRIC STUDIES ................... 4-97

4.1 Introduction ................................................................................................. 4-97

4.2 Impact reconstruction by using crash test data ............................................ 4-98

4.2.1 Vehicle-Column Interaction .................................................................. 4-99

4.2.2 Effects of Impact Pulse Parameters ..................................................... 4-100

4.2.3 Impact pulse modelling and effects of the impact pulse parameters .. 4-103

4.2.4 Simulation of impact of axially loaded columns in medium rise

buildings .............................................................................................. 4-104

4.3 Impact behaviour of the columns and possible damage modes ................ 4-106

4.4 Vulnerability prediction ............................................................................. 4-107

4.5 Bending moment and resultant shear due to impact ................................. 4-110

4.6 Effects of the diameter of the column, concrete grade and steel ratio ...... 4-111

4.7 Effects on the slenderness ratio ................................................................. 4-112

4.8 Energy absorption due to the impact ......................................................... 4-113

4.9 Effects of impact duration ......................................................................... 4-113

4.10 Conclusions ......................................................................................... 4-115


xii

5. CAPACITY OF THE AXIALLY LOADED COLUMNS UNDER LATERAL

IMPACTS .................................................................................................. 5-117

5.1 Introduction ............................................................................................... 5-117

5.2 Design against accidental loads ................................................................. 5-118

5.3 Parametric studies of impacted columns ................................................... 5-119

5.3.1 Finite element analysis of confined circular columns ......................... 5-119

5.3.2 Effects of the confinement .................................................................. 5-119

5.4 Effectiveness of confinement under impact loads ..................................... 5-120

5.5 Effects of the unconfined cover and use of external wrapping ................. 5-124

5.6 Effects of the slenderness ratio on capacity enhancement ........................ 5-125

5.7 Comparison of the dynamic and static shear capacities ............................ 5-126

5.8 Impact capacity of partially loaded circular columns ................................ 5-127

5.8.1 Introduction ......................................................................................... 5-127

5.8.2 Damage criterion ................................................................................. 5-128

5.8.3 Effect of axial load on the duration of the impact ............................... 5-129

5.8.4 Simplified method to investigate the residual capacity of columns .... 5-129

5.8.5 Effects of transverse reinforcement on capacity enhancement ........... 5-133

5.8.6 Effects of longitudinal reinforcement ratio ......................................... 5-138

5.8.7 Effects of the slenderness ratio ............................................................ 5-139

5.8.8 Anomalous behaviour of columns under post impact loading ............ 5-140

5.8.9 Buckling of reinforcement under impact ............................................ 5-144

5.9 Derivation of empirical relationships to predict critical impulse .............. 5-145

5.10 Derivation of simple linear regression equations ................................ 5-145

5.10.1 Descriptions of the outputs .................................................................. 5-145

5.10.2 Pearson Correlation ............................................................................. 5-146

5.10.3 Coefficient of Determination and Analysis of Variance ...................... 5-147

5.10.4 Interpretation of partial (regression) plots ........................................... 5-148

5.10.5 Regression coefficients and derivation of the linear equations ........... 5-150

5.11 Derivation of Polynomial equations .................................................... 5-152

5.12 Conclusions ......................................................................................... 5-155


xiii

6. IMPACT ON COLUMNS UNDER UNIAXIAL BENDING .......................... 6-159

6.1 Introduction ............................................................................................... 6-159

6.2 Behaviour of the impacted columns under single axis bending ................ 6-160

6.2.1 The load application procedure ........................................................... 6-161

6.2.2 Material models and mesh generation................................................. 6-162

6.2.3 Axial load and eccentric load applications in an explicit environment6-163

6.3 Deflection profiles and resultant bending moment ................................... 6-165

6.4 Deformation characteristics of the impacted column ................................ 6-167

6.5 Impact behaviour of the eccentrically loaded column .............................. 6-167

6.6 Behaviour of eccentrically loaded confined columns under impact ......... 6-169

6.7 Selection of the load combinations ........................................................... 6-170

6.8 Parametric studies and discussion of the finite element results ................ 6-171

6.8.1 Impact behaviour of eccentrically loaded columns under maximum

allowable capacity ............................................................................... 6-172

6.8.2 Impact behaviour under reduced load eccentricities ........................... 6-173

6.9 Impact behaviour of columns under positive eccentric loading ................ 6-174

6.9.1 Impact response under positive eccentric moments ............................ 6-176

6.10 Confinement effects on eccentrically loaded columns under impact .. 6-177

6.11 Effects of the longitudinal steel ratio on the impact behaviour of columns

.............................................................................................................6-178

6.12 Confinement effects on the impacted columns with high steel ratio .. 6-179

6.13 A comparison of the confined columns with different steel ratios ...... 6-180

6.14 Effects of the slenderness ratio and intensity of loading on impact capacity

.............................................................................................................6-181

6.15 Strain rate sensitivity of eccentrically loaded columns ....................... 6-182

6.16 Linear equations for 20% and 50% loaded columns ........................... 6-183

6.16.1 Linear equations for 20% loaded columns .......................................... 6-184

6.16.2 Linear equations for 50% loaded columns .......................................... 6-184

6.17 Conclusions ......................................................................................... 6-185

7. IMPACT ON COLUMNS UNDER BIAXIAL BENDING ............................. 7-187

7.1 Introduction ............................................................................................... 7-187


xiv

7.2 Numerical simulation of biaxial loaded columns ...................................... 7-189

7.3 Characteristics of the simulated columns .................................................. 7-191

7.4 Selection of load combinations ................................................................. 7-192

7.4.1 Impact capacity of columns under biaxial bending ............................. 7-192

7.5 Effects of the direction of the impact ........................................................ 7-195

7.6 Effect of reduced axial load on biaxial bending ........................................ 7-196

7.7 Effects of longitudinal steel ratio on biaxial bending ................................ 7-198

7.8 Effects of biaxial bending on 20% loaded impacted columns with a 4% steel

ratio. ................................................................................................................... 7-199

7.9 Damage mitigation of the impacted columns under single axis bending .. 7-201

7.10 Effects of the confinement on biaxial bending .................................... 7-201

7.10.1 Impact behaviour of 50% loaded columns with 4% steel under biaxial

bending ................................................................................................ 7-201

7.11 Behaviour of 20% loaded confined columns with 4% steel under biaxial

bending.. ............................................................................................................ 7-203

7.12 Impact behaviour of 50% loaded columns with 1% steel under biaxial

bending.. ............................................................................................................ 7-203


bending.. ............................................................................................................ 7-204

7.14 Effects of the steel grade and diameter of the hoops ........................... 7-204

7.15 Effects of slenderness ratio on impact capacity of columns under biaxial

bending.. ............................................................................................................ 7-208

7.16 Effects of the concrete grade on impact behaviour of columns .......... 7-210

7.17 Development of equations for biaxially loaded columns under lateral

impact.... ............................................................................................................ 7-212

7.17.1 Linear equations for 50% loaded columns (Impact angle 0o to 90o) ... 7-213

7.17.2 Linear equations for 20% loaded columns (Impact direction 0 to 90o)7-215

7.18 Conclusions ......................................................................................... 7-218

8. CONCLUSIONS AND FURTHER DEVELOPMENTS ................................. 8-221

8.1 Introduction ............................................................................................... 8-221

8.2 Main contributions of the thesis ................................................................ 8-221


xv

8.3 Practical significance ................................................................................. 8-222

8.4 Recommendations for future work ............................................................ 8-225

LIST OF ABBREVIATIONS

ALS - Accidental Limit State

ACI - American Concrete Institute

AJI - Architectural Institute of Japan

BS - British Standards

BM - Bending moment

CPU - Central Processing Unit

CS - Cross Section

CEB-FIP -European Committee for Concrete-International Federation for Prestressing

DIF - Dynamic Increasing Factor

DTEI - Department for Transport, Energy and Infrastructure

EAF - Energy of Approach Factor

EAFo - On set Energy of the vehicle

EoS - Equation of State

FE - Finite Element

EFA - Finite Element Analysis

EFM - Finite Element Model

HSC - High strength concrete

HSRC - High Strength Reinforced Concrete

LSC -Low strength concrete

NHTSA - National Highway Traffic Safety Administration

N.A. - Neutral axis

RC - Reinforced Concrete

REL - Release (version)

SDoF - Single degrees of freedom

SLS - Serviceability Limit State

S/R - Selective Reduced solid

NZS - New Zealand Standards


xvi

ULS - Ultimate Limit State

VC - Viscous Friction

VDC - Viscous Damping Coefficient

LIST OF SYMBOLS

LOWER CASE LATIN CHARACTORS

b - web width of a section

c/c - centre to centre

b1,b2,b3-user defined scale multipliers

bc - breadth of the core concrete for square/circular column

b” - width of the confined core

c - concrete cover thickness

cs - sound speed

d - effective depth of a section / diameter of a hoop

ds - distance between bar centres

ex - load eccentricity along the X axis

ey - load eccentricity along the Y axis

f ’ cc - compressive strength of confined concrete

fc,imp - dynamic compressive strength (mean)

fcm - mean compressive strength of concrete

f ’ co - compressive strength of unconfined concrete

fct,imp- dynamic tensile strength

fctm - tensile strength of concrete

fc - compressive stress

f ’ c - concrete compressive cylinder strength

fl - confining stress

kif α - hourglass resisting force vector

'lf - effective lateral confining stress

ft -tensile stress

fy - yield stress


xvii

'syf - yield strength of confined reinforcement

fy - yield strength of the transverse reinforcement

fyh - maximum strength of transverse steel

h - smaller dimension of the rectangular section

αih - nodal coordinates

k - spring stiffness

kd -internal scale multiplier

ke -confinement effectiveness coefficient

m - gross mass of a vehicle

n - modular ratio

p - mean stress in the tri-axial compression failure test/pressure

q - strain rate sensitivity of the material

r - pearson correlation

r f -modification factor to account for dynamic strength of concrete

s - spacing of transverse steel

sl - spacing between laterally confined longitudinal bars

t - duration

w - crushed width

x - crushed length / variable

v - velocity of the vehicle

v& - rate of deformation

vc - reduced shear stress

ev - element volume

vr - object velocity at impact

UPPER CASE LATIN CHARACTORS

A - shape function

Ab - area of the transverse reinforcements

Ac - net area of a concrete core

Ag - gross concrete area of a section


xviii

Ai - cross sectional area of an equivalent impacting object

Ash - transverse steel area across a section

Ash,c - ACI code provision of transverse reinforcement content

Asw - transverse reinforcement area within a spacing

Aws - area of the longitudinal steel in the section

B - shape function

C - a material constant

Co, C1- bulk viscosity coefficients

D - diameter

Dc - characteristic strain rate

Di - Damage index

Do - diameter of circular section

L - length of the element

Li - length of the impacting object

Le - element length

E - modulus of elasticity

Ec,imp- mean impact modulus of elasticity (compression)

Eci - modulus of elasticity at 28 days

Ec - reduced modulus of elasticity

Et - tangent moduli

Eo - absorbed energy by a crushed vehicle

Em - mesh density

F - impact or breaking force

Fo - plastic strength of a structure

G - shear modulus

G f - fracture energy of concrete

H - effective height

Hs -softening modules

I - impulse

Io - critical impulse

Ic - critical impulse (corrected)

Ie - confinement index


xix

I1 - first stress invariant of a stress tensor

J2 - second invariant of deviatoric stress tensor

J3 - third stress invariant

Li - length of the impacting object

M - moment

Mx - moment on a column corresponding to dxP axial load )10( << x

Mxs - moment about X axis

Mys - moment about Y axis

P - axial load / amplitude of an impact pulse

Pa - the axial capacity

Pc - critical impact force (corrected)

Pd - design axial load capacity

Pe - confining pressure

Po - critical impact force

Pr - residual axial load carrying capacity

R2 - coefficient of determination

Rd - design resistance

Rk - characteristic resistance

Sd - design load effects

Sk - characteristic load effect

Sy.x - standard error of estimate

βS -standard error of the slope

Tlimit-tensile limit

Vc - shear capacity of concrete

Vd - dynamic shear capacity

Vs - static shear capacity

Vn - shear strength corresponding to the maximum moment capacity

Vr - volumetric ratio of confined reinforcement

Vs - shear capacity of steel

Xi - independent variable

Yi - dependent/response variable

Yo - deformation capacity of a crushed vehicle


xx

X,Y - two orthogonal (principle) axes

GREEK CHARACTORS

β - shear retention factor

hβ - kinematic hardening parameter

σ - stress

cσ& - stress rate (compression)

ctσ& - stress rate (tension)

'dσ - dynamic flow stress

sσ - static flow stress

cε& - strain rate (compression)

ε& - strain rate

cε - compresive strain

sε - strain in steel

εv - volumetric strain

εv,yield - volumetric strain at yield

pdε -effective plastic strain

pijdε -plastic strain increment tensor

λ -damage function

σd - stress under dynamic conditions

σs - stress under static conditions

η - yield scale factor/viscosity

fγ - partial factor for loads

mγ - material factor

φ - diameter

φ’ - strength reduction factor

α - a factor dependent on the tie configuration


xxi

α’ - tangent of a crack angle

αs - shear-bending capacity ratio

µφ - curvature ductility factor

ρ - mass density / poisons ratio

ρcc - ratio of area of longitudinal reinforcement to area of core of section

sρ - ratio of volume of transverse reinforcement to volume of concrete core

ρtotal - steel ratio of the section

wρ - transverse reinforcement ratio

vρ - longitudinal steel ratio

kαΓ - nodal velocity

∆ - Impact angle in degrees

l∆ - deceleration length

∆te - critical time step size

∆t - duration of the plus

∆t - small time interval

λ∆ -volumetric plastic strain

σ∆ - stress difference

cyσ∆ - compressive meridian of the initial yield surface

cmσ∆ - maximum failure surface

cmσ∆ - residual surface (compression)

tmσ∆ - tensile meridian

Lσ∆ -loading surface after yielding

pfσ∆ - post failure surface

ABBRIVIATIONS OF UNITS

kg kilogram

km/h kilometres per hour

kN kilo Newton


xxii

m Metres

mm Millimetres

MPa mega Pascal

ms Milliseconds

mph Miles per hour

s Seconds

t tonne

cm centimetre

LIST OF TABLES

Table 2.1: Tangent moduli and reduced moduli of elasticity .................................. 2-30

Table 2.2: Mass and stiffness coefficient for impact reconstructions ..................... 2-61

Table 3.1: Material properties used for the concrete ............................................... 3-77

Table 3.2: Characteristics of the Feyerabend’s test specimens (Feyerabend 1988) 3-79

Table 3.3: Material properties used for the main reinforcement ............................. 3-86

Table 3.4: Material properties used for the Rigid Body .......................................... 3-86

Table 5.1: Descriptive Statistics ............................................................................ 5-146

Table 5.2: Pearson correlations ............................................................................. 5-147

Table 5.3: Coefficient of determination of the equation ....................................... 5-147

Table 5.4: Analysis of Variance ............................................................................. 5-148

Table 5.5: Regression Coefficients for linear equations ........................................ 5-151

Table 7.1: Biaxial load combinations on the 300mm column under 50% axial load

............................................................................................................................... 7-192

LIST OF FIGURES

Figure 1.1: Severely damaged columns due to a vehicle impact .............................. 1-1

Figure 2.1: Sequence of an impact (El-Tawil et al. 2005) ....................................... 2-15

Figure 2.2:Impact force vs. time histories for Chevy truck at various speeds ........ 2-16

Figure 2.3: Comparison between different modes of vibration with equal potential

energies (Hughes and Speirs 1982) ......................................................................... 2-21

Figure 2.4: Strain-rate sensitivity for concrete in compression, tension and flexure


xxiii

................................................................................................................................. 2-26

Figure 2.5: Model for the strain rate dependency of concrete elastic modulus ...... 2-29

Figure 2.6: Strain-rate influence on the compressive strength of concrete ............. 2-31

Figure 2.7: Strain-rate influence on the tensile strength of concrete ...................... 2-32

Figure 2.8: Model for the strain rate dependency of concrete in compression and

tension according to the CEB-FIP model code (CEB 1993) and with the modified

model according to Malvar and Crawford (1998-a) ............................................... 2-33

Figure 2.9: Distribution of confining pressure produced by various shapes of ...... 2-41

Figure 2.10: Stress-strain model for confined concrete proposed by Mander et al.

(1984) ...................................................................................................................... 2-55

Figure 2.11: Confining stress provided by the transverse reinforcements .............. 2-56

Figure 3.1: Failure surfaces in Mat_Comcrete_Damage_REL3 (Malvar et al. 1997)

................................................................................................................................. 3-71

Figure 3.2: The test set-up by Feyerabend (1988) .................................................. 3-78

Figure 3.3: The simplified test set-up and the cross section of specimen ............... 3-79

Figure 3.4: Convergence of the numerical model ................................................... 3-80

Figure 3.5: Mesh generation for the impacted column & rigid body ...................... 3-81

Figure 3.6: Single element under uni-axial tensile test ........................................... 3-84

Figure 3.7: Lateral pressure distribution and the resultant strain gradient .............. 3-90

Figure 3.8: Comparison of the resultant deflections ................................................3-91

Figure 3.9: Interface forces during the impact ........................................................ 3-92

Figure 3.10: Comparison of the resultant deflections ............................................. 3-93

Figure 3.11: Crack Propagation of the impacted column and numerical simulation

................................................................................................................................. 3-93

Figure 3.12: Comparison of the resultant impact force...........................................3-94

Figure 3.13: Comparison of the resultant reactions ................................................ 3-94

Figure 4.1: Front and side views of an impacted column (NHTSA) ...................... 4-99

Figure 4.2: Lateral pressure distribution across the diameter of the 300mm column

............................................................................................................................... 4-100

Figure 4.3: Force Time histories of full scale crash tests (NHTSA 1997) ............ 4-101

Figure 4.4: Iso-damage pulses............................................................................... 4-103

Figure 4.5: Effects of the strain rate or frontal stiffness ........................................ 4-103


xxiv

Figure 4.6: Cross sectional areas of the circular concrete columns........................4-105

Figure 4.7: Support conditions and external load applications ............................. 4-105

Figure 4.8: Plan view of the half models .............................................................. 4-106

Figure 4.9: Comparison of impact capacities of columns with full scale crash tests

(NHTSA) ............................................................................................................... 4-108

Figure 4.10: Honda Accord in a frontal collision at a speed of 48.3km/h ........... 4-108

Figure 4.11: Ultimate capacities of columns..........................................................4-109

Figure 4.12: Support reaction and Impact pulse ................................................... 4-109

Figure 4.13: Resultant bending moments...............................................................4-110

Figure 4.14: Damaged column under vehicle impact ........................................... 4-110

Figure 4.15: Effects of the diameter of the column.................................................4-111

Figure 4.16: Effects of the concrete grade ............................................................ 4-111

Figure 4.17: Effect of the slenderness ratio............................................................4-112

Figure 4.18: 500mm column with 4% steel .......................................................... 4-112

Figure 4.19: Equivalent impulse diagrams.............................................................4-113

Figure 4.20: Iso-damage pulses for 600mm column .............................................4-113

Figure 4.21: Force vs reaction histories for pulses with different durations ......... 4-114

Figure 4.22: A typical Pressure impulse curve......................................................4-114

Figure 4.23: Iso-damage contours for impact ....................................................... 4-114

Figure 5.1: Effects of confinement under lateral impacts ..................................... 5-119

Figure 5.2: Stress difference at the cover-core interface ....................................... 5-120

Figure 5.3: Capacity enhancement for 30MPa concrete ....................................... 5-122

Figure 5.4: Capacity enhancement for 50MPa Concrete ...................................... 5-123

Figure 5.5: Confined strength for different concrete grades ................................. 5-124

Figure 5.6: Columns confined with 12mm links at 100mm spacing .................... 5-125

Figure 5.7: Comparison of the dynamic and static shear capacities ..................... 5-126

Figure 5.8: Rehabilitation of a bridge after catastrophic failure of a column ....... 5-127

Figure 5.9: Axial pressure application on 300mm diameter column .................... 5-130

Figure 5.10: Deflection characteristics of 450mm column ................................... 5-131

Figure 5.11: Axial load sensitivity of impacted columns ...................................... 5-132

Figure 5.12: Deformation characteristics of 300mm column ............................... 5-134

Figure 5.13: Impact capacity of 300mm column under varying axial loads ......... 5-135


xxv

Figure 5.14: Enhanced capacities for confined columns with different axial loading

............................................................................................................................... 5-136

Figure 5.15: Capacity reduction due to axial load ................................................ 5-137

Figure 5.16: Capacity enhancement due to confinement ...................................... 5-137

Figure 5.17: Impact capacity under varying axial load..........................................5-139

Figure 5.18: Capacity increment under varying axial load ................................... 5-139

Figure 5.19: Impact capacities of short columns .................................................. 5-139

Figure 5.20: Different failure characteristics of structural columns ..................... 5-140

Figure 5.21: Axial load sensitivity of the counterintuitive effect .......................... 5-141

Figure 5.22: Typical failure pattern of rectangular columns under eccentric loading

............................................................................................................................... 5-142

Figure 5.23: Failure of columns by concrete crushing .......................................... 5-144

Figure 5.24(a-f): Partial regression plots of each parameter against Log P .......... 5-149

Figure 5.25: Accuracy of the prediction by linear equations ................................ 5-151

Figure 5.26: Accuracy of the polynomial equations ............................................. 5-153

Figure 5.27 (a-g): Steps of the derivation of polynomial equations ..................... 5-153

Figure 6.1: Plan view of the column head (under uni-axial bending) ................... 6-160

Figure 6.2: The Bulk head of columns used for the application of moment ......... 6-162

Figure 6.3: Numerical simulation of eccentrically loaded columns ..................... 6-163

Figure 6.4: Interaction diagram for the 300mm......................................................6-164

Figure 6.5: Extreme strain in the 300mm ............................................................. 6-164

Figure 6.6: A typical Interaction diagram of a column ......................................... 6-165

Figure 6.7: Contours of effective plastic strain ..................................................... 6-166

Figure 6.8: Time histories of BM of 300mm eccentrically loaded half column with 1%

steel ....................................................................................................................... 6-166

Figure 6.9: Deflection of 300mm diameter column with 4% steel at the near collapse

stage....................................................................................................................... 6-167

Figure 6.10: Resultant bending moment at different locations on the 450mm column

............................................................................................................................... 6-167

Figure 6.11: Resultant shear forces at different locations on the 450mm column 6-168

Figure 6.12: Lateral pressure distribution and the corresponding stress-strain

relationships .......................................................................................................... 6-169


xxvi

Figure 6.13: Interaction diagrams for 450mm column according to AS3600 and ACI:

318 ......................................................................................................................... 6-170

Figure 6.14: locations of the selected loading points on the interaction diagrams6-171

Figure 6.15: Eccentrically loaded columns with 1% steel ratio ............................ 6-172

Figure 6.16: Peak force under different load combinations .................................. 6-173

Figure 6.17: Comparison of the Impact capacities under positive and negative

moments ................................................................................................................ 6-175

Figure 6.18: Cracks on 20% loaded 300mm column with 1% steel.......................6-176

Figure 6.19: Cracks on 20% loaded 450mm and 600mm columns with 1% steel 6-176

Figure 6.20: Resultant bending moments of the 20% loaded 300mm column ..... 6-177

Figure 6.21: Capacity of eccentrically loaded confined columns under impact ... 6-177

Figure 6.22: Effect of longitudinal steel ratio on impact capacity ........................ 6-179

Figure 6.23: Impact capacity enhancement due to confinement ........................... 6-180

Figure 6.24: Effects of the longitudinal steel ratio on impact capacity enhancement

............................................................................................................................... 6-181

Figure 6.25: Effects of the slenderness ratio on capacity enhancement for columns

............................................................................................................................... 6-181

Figure 6.26: Strain rate sensitivity of a ductile column ........................................ 6-182

Figure 6.27: Residuals of the predicted values (20%)............................................6-183

Figure 6.28: Residuals of the predicted values (50%) .......................................... 6-183

Figure 7.1: Impact capacity prediction for intermediate load combinations ......... 7-189

Figure 7.2: Typical interaction diagram for circular columns under biaxial bending

............................................................................................................................... 7-190

Figure 7.3: Numerical model of the column under biaxial bending ..................... 7-191

Figure 7.4: Impact capacities of the columns under biaxial bending .................... 7-193

Figure 7.5: Simulation of the effects of the direction of the impact ..................... 7-195

Figure 7.6: Failure characteristics of 50% loaded 300mm columns under biaxial

bending .................................................................................................................. 7-198

Figure 7.7: Peak force vs slenderness ratio for 4m high columns made of 50MPa

concrete ................................................................................................................. 7-201

Figure 7.8: Impact capacity of 20% loaded columns under varying hoop spacing

............................................................................................................................... 7-205


xxvii

Figure 7.9: Impact capacities of 20% and 50% loaded columns under varying hoop

spacing................................................................................................................... 7-206

Figure 7.10: Impact capacities of 20% and 50% loaded columns under varying hoop

diameter ................................................................................................................. 7-206

Figure 7.11: Impact capacities of 20 and 50% loaded columns under varying yield

strength .................................................................................................................. 7-207

Figure 7.12: Peak force vs. Slenderness ratio for columns of 50MPa concrete with 1%

steel ....................................................................................................................... 7-209


steel ....................................................................................................................... 7-209

Figure 7.14: Ultimate capacity of 3m columns made of 50MPa concrete ............ 7-210

Figure 7.15: Ultimate capacity of 2m columns made of 50MPa concrete ............ 7-210

Figure 7.16: Comparison of peak force of 4m high columns made of Grade 30 and 50

concrete ................................................................................................................. 7-210


steel ....................................................................................................................... 7-212


steel ....................................................................................................................... 7-212

Figure 7.19(a-h): Partial regression plots of each parameter against Log P ......... 7-214

Figure 7.20: Residual of the predicted values ....................................................... 7-215

Figure 7.21 (a-h): Partial regression plots of each parameter against Log P ........ 7-216

Figure 7.22: Accuracy of the predicted values ...................................................... 7-217


xxix

Publications Refereed International Journal Papers HMI Thilakarathna, DP Thambiratnam, M Dhanasekar, N Perera. Numerical simulation of axially loaded concrete columns under transverse impact and vulnerability assessment. International Journal of Impact Engineering 2010; 37(11): p. 1100-1112. HMI Thilakarathna, DP Thambiratnam, M Dhanasekar, N Perera. Impact response and vulnerability assessment of concrete columns under vehicle impacts. Advances in Structural Engineering 2010; (under review). HMI Thilakarathna, DP Thambiratnam, M Dhanasekar, N Perera. Capacity of biaxially loaded circular reinforced concrete columns under transverse impacts. Engineering structures 2010; (under review). International Conference Papers HMI Thilakarathna, DP Thambiratnam, M Dhanasekar, N Perera. Behaviour of Axially Loaded Concrete Columns Subjected to Transverse Impact Loads. 34th Conference on Our World in Concrete & Structures ‘Green Concrete’; 2009. V28. p 359-366. Park Royal Hotel on Kitchener Road, Singapore HMI Thilakarathna, DP Thambiratnam, M Dhanasekar, N Perera. Impact Response and Parametric Studies of Reinforced Concrete Circular Columns. 4th international conference on ‘Protection of structures against hazards’. October 2009, p 347-354. Tsinghua Unisplendour International Centre, Beijing, China Book Chapter Yigitcanlar, T., Ed. (2010). Sustainable Urban and Infrastructure Development: Management, Engineering and Design. Chapter 15. Infrastructure sustainability: vulnerability of axially loaded columns subjected to transverse impact loads. Brisbane. Local Conference Paper HMI Thilakarathna, DP Thambiratnam, M Dhanasekar, N Perera. Vulnerability of Axially Loaded Columns Subjected to Transverse Impact Loads. March 2009. The second infrastructure theme Postgraduate Conference. Gardens Point campus, QUT. p.22-34.


xxxi

Statement of original authorship The work contained in this thesis has not been previously submitted for a degree or

diploma at any other higher education institution. To the best of my knowledge and

belief, the thesis contains no material previously published or written by another

person except where due reference is made.

Indika Thilakarathna

6 December 2009


xxxiii

ACKNOWLEDGEMENTS

I wish to express my heartfelt gratitude to the principle supervisor Professor David

Thambiratnam for his motivation, patience and endless support throughout the period

of my PhD studies. I also want to express an especial thankyou to the associate

supervisors Professor Sekar Dhanasekar and Professor Nimal Perera for their

encouragement and professional guidance during the entire period.

My sincere thanks go to Mr. Mark Barry and the other staff members at the High

Performance Computer Unit for their assistance and cooperation during the research

and for enthusiastic responses to my numerous requests for assistance. I extend my

thanks to the LEAP support, the QUT software provider for their great effort to assist

me with the problems during the Finite Element Modelling. It is a true pleasure to

express my gratitude to staff of the Document Delivery Unit at QUT for their

continuous assistance and cooperation whenever needed. Many thanks also go to Dr.

Greg Nagel and Department of International Student Services at QUT for revising and

proof reading of the thesis.

I gratefully acknowledge the financial support granted to me from the IPRS

scholarship and the President’s Fund of Sri Lanka to conduct the research. I am also

grateful to the personnel and my fellow doctoral students at the Faculty of Built

Environment and Engineering and also to my colleagues at QUT for sharing

knowledge and for contributing to a friendly and fruitful atmosphere. Finally, I also

wish to express my deepest gratitude to my wife, parents and family members for their

patience and understanding during the completion of this research.

Queensland University of Technology,

Brisbane,

December – 2010.

Indika Thilakarathna.


1-1

1. INTRODUCTION 1.1 Background

Increased industrialisation has brought to the forefront the susceptibility of concrete

columns in both buildings and bridges to vehicle impacts. Catastrophic failure of a

bridge or a building, as a result of a vehicle collision, worsens the consequences and

requires special attention to design and detailing. For example, the Department for

Transport Energy and Infrastructure (DTEI, 2004) in their annual report state that

vehicle impact with road-side objects alone costs the Australian society $4.6 billion

per year. As a whole, these crashes represent 30% of the total fatal motor vehicle

crashes.

Figure 1.1: Severely damaged columns due to a vehicle impact

In fact, there are several techniques that can be used to mitigate the damage due to

impact. Crash barriers, Fencing and Bollards are the most common methods that can

be used to prevent the direct collusion with structural columns or to reduce the

approaching speed of the impacting vehicle. Additionally, key elements can be

avoided by providing alternative load paths. However, there may be restrictions

applied on such methods due to aesthetic reasons and limitations on space. For

instance, these options are hardly adopted or inadequate to prevent the vehicle impact

on bridge underpass (see Fig 1.1). In the absence of suitable passive resistance,

structural strengthening will be the most appropriate solution.


1-2

In fact, structural columns are seldom designed for vehicle impacts due to inadequacies

of design guidelines. The behaviour and vulnerability assessment of columns subjected

to lateral impact is, therefore of an interest. This is requires a knowledge on the

behaviour of concrete under extreme loading conditions, which is very complex. In

fact, there are few experimental investigations conducted on laterally impacted

columns that demonstrate the effects of strain rate and confinement, particularly under

mid span hard impact conditions (Louw et al. 1992). However, the hard impacts

usually represent the possible upper bound of the typical vehicle impacts (Tsang et al.

2005) and hence generate over conservative results due to the exaggerated strain rate

effects. In contrast, the energy based vulnerability assessment techniques presented in

the literature (Tsang et al. 2005) limit their application to columns that are fail in

flexure under mid span low velocity impacts where strain rate and inertia effects are

not predominant. The typical low elevation vehicle impacts initiate flexural-shear

failure, a condition which is substantially different to flexural failure events under mid

span impacts. Thus, research on shear critical RC columns under low elevation impact

remains unexplored. Furthermore, since the impact response of columns is associated

with higher modes of vibrations, strain rate effects, confinement effects, as well as

various other vehicle specific parameters, the analysis process becomes quite complex

(Hughes and Speirs 1982; Varat and Husher 2000). Consequently, the impact

reconstruction techniques based on the deformable body assumption (Campbell 1974;

Prasad 1990) are largely simplified while the rest of the methods are limited to simulate

an impact between a specific vehicle and a column (El Tawil et al. 2005). These

methods may not be reliable to assess the vulnerability of columns against a general

vehicle population or new generation vehicles. As a result, the use of numerical

methods for the vulnerability predictions of RC columns is exceptionally limited. This

is also reflected by the generic and limited design guidelines provided in the current

codes of practices (AASHTO-LRFD 1998; EN 1991-1-7:2006) which do not provide

adequate design information on impacted columns. Additionally, different design

codes specify significantly different magnitudes for the expected quasi-static impact

loading, which indicates a lack of understanding of the dynamic behaviour of both the

column and colliding vehicle (AASHTO-LRFD 1998; EN 1991-1-7:2006;

Vrouwenvelder 2000). In fact, the equivalent static force given in AASHTO-LRFD


1-3

(1998) is based on information resulting from early full-scale crash tests of barriers

used for redirecting tractor trailers and truck collisions. This indicates that the design

forces were not derived directly from head-on collision tests. At present there are no

comprehensive guidelines on how to detail a vulnerable column to ensure that it will

survive in a catastrophic impact situation (El Tawil 2005). Consequently, there is a

pressing need for the development of some simplified yet rational method to quantify

the effects of impact.

Another aspect of the present work was to determine the vulnerability of the structural

column during the construction process when the applied load is only a portion of the

design load and hence the shear capacity and the stiffness have not reached their full

potential (Abrams 1987; Zeinoddini 2008). Proper impact damage assessment is vital

to determine whether the column has to be replaced or can be repaired for further use.

Design guidelines developed on partially loaded columns subjected to earthquake

(Esmaeily and Xiao 2004) and blast (Shi et al. 2008) loading may not be adequate in

this circumstance where mode of failure, strain rate effects and inertia effects are

substantially deferent. Moreover, a decision on the portion of total load that can be

allowed during the rehabilitation process has to be made. Proper damage assessment

will be essential to minimise the risk to rescue workers who enter into the building

following an impact, or when the affected bridge structure has to be used as a vital

supply line.

This thesis addresses these questions by investigating the impact capacity of columns

particularly in low to medium raised buildings designed according to the Australian

standards. Numerical simulation techniques were used in the process and validation

was performed using experimental results published in the literature. The analysis

extended to investigate the influence of critical parameters that govern the

vulnerability of columns under lateral impact loads. Numerical simulations were

conducted using the Finite Element program LS-DYNA (2006), incorporating steel

reinforcement, confinement effects and strain rate effects. A simplified method based

on impact pulses generated from full scale impact tests was used for impact

reconstruction and the effects of the various pulse loading parameters were

investigated under low to medium (30-80km/h) velocity impacts. A constitutive


1-4

material model which can simulate failure under a tri-axial state of stress was used for

concrete. The sensitivity of the material model parameters used for the validation was

also scrutinised and numerical tests were performed to examine their suitability to

simulate the shear failure conditions. Columns made of 30 to 50MPa concrete with a

longitudinal steel ratio ranging from 1% to 4% under pure axial loading were analysed

in the first phase. The study was then extended to analyse columns subjected to single

axis and biaxial bending. The impact capacity reduction under varying axial loads was

also quantified. Suggestions on the residual capacity of the columns based on the

status of the damage and mitigation techniques are presented. Empirical equations

have been developed to quantify the critical impulses in terms of the geometrical and

material properties of the impacted columns. Equations are provided for explicit and

reliable vulnerability assessment. Moreover, a universal technique which can be

applicable to determine the vulnerability of the impacted columns against collision

with new generation vehicles under most common impact modes is proposed.

The investigation has confirmed that columns in medium storey buildings located in

urban areas are vulnerable to medium velocity car and truck impacts and hence these

columns need to be re-designed for retrofitting. The proposed equations can be used to

quantify the critical impact pulses for 300mm to 750mm diameter circular columns

under axial loading. This allows the impact capacity of the columns to be improved by

optimum use of the key design parameters without relying on external energy

absorbers or wrapping. According to the overall results, the vulnerability of the axially

loaded columns can be mitigated by reducing the slenderness ratio and concrete grade,

and by choosing the design option with a minimal amount of longitudinal steel.

Results also indicated that the ductility as well as the mode of failure under impact can

be changed with the volumetric ratio of lateral steel. Moreover, to increase the impact

capacity of the vulnerable columns, a higher confining stress is required. The general

provisions of current design codes do not sufficiently cover this aspect and hence this

research provides additional guidelines to overcome the inadequacies of code

provisions.

In addition, an extensive numerical simulation was conducted on uniaxially and

biaxially loaded columns. In particular, the analysis procedure became quite complex


1-5

under biaxial loading due to large number of load combinations involved in the

analysis process which includes the direction of the impact. Consequently, non-critical

cased excluded from the analyses. Analytical equations were derived for 300mm to

600mm eccentrically loaded columns that are valid in the range of 0o to 900 impact

angles. The remaining impact angles can be treated separately based on the

observation that columns under the Positive Eccentric Loading are non-critical. This

procedure allows defining critical impulses for three consecutive points on the

interaction diagram for a one particular column so that linear interpolation can be used

to quantify the critical impulses for the points in-between. Having provided a known

force and impulse pair for an average impact duration, this method can be extended to

assess the vulnerability of columns for a general vehicle population. Consequently,

this investigation delivered new insight and comprehensive technique to the scientific

community for dealing with impacts.

1.2 Aims and objectives

The overall aim of this research is to generate design guidelines and equations on the

impact capacity of axially and eccentrically loaded columns while allowing optimum

use of key design parameters without relying on energy absorbers or wrappings.

Additionally, the impact capacity of the columns during construction will be assessed

comprehensively where the applied axial load is only a portion of the total allowable

load on the column and hence the shear capacity may be crucial under low elevation

impacts. In particular, this thesis is concerned the global behaviour of the impacted

columns and the influence of the critical parameters on their performance in order to

mitigate damage. The research also has developed a numerical modelling technique to

lessen the need for full scale vehicle impact testing.

The aims of this research are achieved by addressing the following enabling

objectives:

a) Development of a finite element model for an impacted column, and validating

the model using existing experimental results reported in the literature.


1-6

b) Development of an impact simulation (reconstruction) technique which allow

assessing the vulnerability of the column for a general vehicle population over most

common modes of collisions under low to medium velocity impacts.

c) Quantification of the effects of column diameter, axial load and moment, impact

height, impact duration, strain rate effects, concrete grade, impact direction, support

conditions, confinement and lateral steel ratio. It also investigates the method of

enhancing the performance of the impacted columns by optimum usage of the critical

parameters.

d) Evaluation of the impact behaviour of reinforced concrete columns under single

axis bending by extending the validation process to simulate the impacted columns

under eccentric loading. Assess the vulnerability of biaxially loaded columns under

serviceability conditions by excluding the non-critical load combinations from the

analyses to generate conservative outcomes.

e) Developing equations and design information to assess the vulnerability of the

columns both under pure axial and eccentric loadings. Additionally, generate design

information to assess the vulnerability of structures under construction while

providing mitigation techniques for partially loaded columns through the enhancement

of confinement effects.

f) Examination of the static and dynamic shear capacity of the columns and

quantify its correlation with the static shear capacity of the columns so that it can be

used for approximate vulnerability assessments for impacted columns.

g) However effects of the surrounding temperature variations and columns under

submerged conditions will not be investigated due to the complexity of the simulation

process.

1.3 Hypotheses and research problems

Damage due to lateral impact loads applied on columns can be minimised if there is a


1-7

satisfactory amount of research information on their impact performance. To

strengthen this argument, the following hypotheses will be tested throughout this

research project:

I. The lateral impact capacity increases substantially with the axial load.

Consequently, the vulnerability of the structural column during the construction

process may be high where the applied load is only a portion of the design load and

hence the shear capacity and the stiffness have not reached their full potential. On

the other hand, shear strength enhancement, resulting from the increase of axial

load during impact, can effectively increase the lateral load carrying capacity of

short columns.

II. The impact capacity substantially changes with the bending moments due to initial

deformation present in the columns prior to the impact. In particular, columns under

mild tension and reduced compression have the tendency to reduce the shear

capacity under lateral impacts.

III. The impact capacity of the reinforced concrete columns can be increased by

improving their ductile characteristics. In general, the ductile capacity depends on

the amount and distribution of transverse reinforcement within the plastic hinge

region and this concept was particular effective under earthquake loading.

IV. Concrete grade, steel area, diameter of the column, hoop spacing, area of the hoops

and yield strength and effective height are the key design parameters which

determine the vulnerability of columns under shear critical quasi-static loading.

The relation between these parameters can be been expressed in terms of static and

dynamic shear capacities of the columns.

V. Impact induced torsional moments may significantly change the internal stress and

deformation capacity of structural columns by changing the failure mode

particularly when the impact force is applied perpendicular to the direction of

eccentric loading. There are no analytical models developed combining the effects

of shear-flexure-torsion interaction for vulnerability assessment of the impacted

columns.


1-8

VI. A strength enhancement of 30% could be expected in structural beams under mid

span impact loads, particularly when the strain rate effects accompanied by higher

flexural strains (Louw et al. 1992). This enhancement would be undesirable for

columns susceptible to fail in shear under vehicle impacts.

On the other hand, most designers tend to adopt oversimplified procedures in

vulnerability assessment and impact reconstructions. These procedures may not be

rational and hence, underestimate or overestimate the consequences. For instance,

Tsang et al. (2005) quantified the capacity of reinforced concrete columns under

mid-span head-on collisions by using a displacement based failure criteria. However

this energy based method limits its application to columns that fail in flexure under

low velocity impacts where strain rate and inertia effects are not predominant.

Alternatively, El Tawil et al. (2005) conducted a dynamic impact simulation to

demonstrate the inadequacies of the AASHTO-LRFD (1998) code provisions by using

a validated numerical model of a Chevy truck. The drawback of such methods is that

the outcomes cannot be applicable to a general vehicle population or new generation

vehicles. In fact, the equivalent static force given in AASHTO-LRFD (1998) is based

on information resulting from early full-scale crash tests of barriers used for

redirecting tractor trailers and truck collisions. This indicates that the design forces

were not derived directly from head-on collision tests. At present there are no

comprehensive guidelines on how to detail a vulnerable column to ensure that it will

survive in a catastrophic impact situation (Tsang et al. 2005). Consequently, there is a

pressing need for the development of some simplified yet rational method to quantify

the effects of impact.

1.4 Methodology

The finite element method has been widely used to perform numerical analysis as

physical testing can be expensive and time consuming. Once validated, the numerical

models allow detailed investigation of the stresses, strain and failure modes by

changing the parameters where physical testing may be difficult due to practical

limitations.


1-9

In general, structural columns can be conveniently subdivided into two categories;

axially loaded columns and columns with eccentric loading. Validation of the

numerical models for each category of columns is performed at the beginning by using

existing experimental testing and existing theoretical and numerical results. All

numerical modelling in this research project was conducted using LS-DYNA (2006);

using the pre-processor MSc Patran. LS-PREPOST was used to visualise the compiled

results. The finite element code LS-DYNA is capable of demonstrating and simulating

the nonlinear deformation of the impacted columns with the added advantage of

simulating contact between the impacting bodies where two materials with different

stiffness interact under high velocities.

The selection of proper material models is also important in the validation process as

the reliability of the result mainly depends on the capacity of the material formulation

to simulate the observable behaviour of the impact. Two material models were selected

for the validation process by considering their capacity to simulate the failure modes

and their reliability of predicting the properties of different concrete grades in the

parametric studies. After the validation, the most reliable model was used for the rest

of the analyses. In addition, simulating the impacting mass in a realistic manner in the

validation process allows comparison of the interface forces generated during the

impact so that boundary conditions can be suitably adjusted to simulate the actual

constraint applied on the column.

The impact reconstruction process also plays an important role. Producing realistic

finite element models of the impacting vehicle is very time consuming process and the

conclusions drawn for a specific vehicle cannot be extended to a general vehicle

population. Thus it was decided to use impact pulses generated from full scale vehicle

impact tests to simulate vehicle collision. Having observed that the strain rate effects

and the pulse shape (with same peak and duration) have negligible effects on the

vulnerability of the columns, triangular pulses were used in the impact reconstruction

process. The average duration of the typical vehicle impact was observed to be 100ms.

The advantage of this method is that it can be applied to determine the vulnerability of

the columns to new generation vehicles.


1-10

Circular reinforced columns of 2m to 4m in height made of 30MPa to 50MPa

concretes were considered in the analysis. Columns that are sufficient in capacity for 5

to 20 storey buildings designed as per AS3600 (2007) were considered in the first

phase of the analysis under pure axial loading. Having observed that the typical 20

storey building columns are less vulnerable to medium velocity (80km/h) vehicle

impacts, the second phase of the analysis was limited to the 15 storey building

columns where eccentric loads were taken into account. In the process, priority was

given to optimise the column design using key design parameters such as the

longitudinal and vertical steel ratios, concrete grade, effective height and support

conditions without relying on external energy absorbers or wrappings. Confinement

effects were introduced to the model by using equations by Mander (1988).

A parametric analysis was conducted on axially loaded columns in the first phase by

varying the key design parameters. The analysis was also extended to investigate

partially loaded columns to assure their safety during construction. The Damage Index

Di was used to identify the capacity degradation of the impacted columns. Dependency

of the duration of the impact on the enhanced stiffness characteristics of the columns

due to axial load variation was neglected in the analysis. The investigation was

conducted by reducing the axial load and then restoring the load at post impact stage. A

simplified method was introduced to apply the post impact loading by avoiding

complex explicit-implicit transformation based methods presented in the literature.

The collapse of the impacted columns is investigated by varying the parameters so that

there are substantial warnings before collapse following the post impact loading. The

hoop spacing was of main interest as it was successful under earthquake loading.

Consequently it was concluded that it would be more appropriate to replace the

impacted columns rather than repair them for further use. In addition, a set of

equations for explicitly determining the critical impact force (Pc) and critical impulse

(Ic) that has been developed using the results from the parametric study and further

simulations. In particular, an innovative technique was developed and introduced to

make sure the accuracy of the equations developed for predicting the critical impact

force and impulse where the other techniques are failed due to the shape of the error

distribution.


1-11

The analysis was extended to investigate the impact under uniaxial bending by

applying the impact force in the plane of bending under full service load. Numerical

simulation of half of the column was used for axially loaded columns where loads are

symmetrical about the direction of the impact. The columns that deflect against the

direction of the impact force are always safe compared to their counterpart. Thus

further analysis was conducted by excluding such conservative load combinations.

Having observed that the capacity drop can be around 30% when the load eccentricity

reaches the ‘balanced failure’ point under uniaxial loading (compared to the axially

loaded columns), further analyses have been limited to the 50% to 20% loaded

columns with corresponding moments. Consequently design guidelines are generated

to quantify the impact capacities of 50% and 20% loaded columns.

The impact angle was taken into account in the third phase of the analysis. A full

column was used in third phase where the biaxial moment application lead to

unsymmetrical loading. Load combinations were selected based on the interaction

diagrams for columns of grade 30to 50MPa concrete with 1% to 4% longitudinal steel.

It was observed that three load combinations are sufficient for the vulnerability

analyses of circular columns under biaxial bending. Vehicle impacts in the direction of

the resultant moment were excluded from the analyses. Conservative results were

generated by considering the columns with maximum (allowable) load eccentricities.

A software program based on the least square method was used to generate simplified

linear equations for all the three phases of the investigation. Polynomial equations

were developed to quantify the critical impact pulses where the percentage error of the

linear equation exceeded the acceptable limit. In particular, polynomial relationships

were developed in stages by varying one parameter at a time. These parameters were

then combined to produce an equation based on the least square method, which could

be used to quantify the peak force and the associated impulse at the near collapse stage

for fully loaded columns. In fact, the main aim was to define three consecutive points

on the interaction diagram so that linear interpolation can be used to quantify the

critical impulses for points in-between. In the process, effects of longitudinal and

lateral steel ratios, concrete grade, direction of the impact, strain rate sensitivity of

columns and slenderness ratio were also quantified.


1-12

1.5 Thesis outline

This section briefly summarizes the format of this thesis.

Chapter 2 provides a literature review on analytical and experimental research to-date

related to laterally impacted columns.

Chapter 3 describe the numerical modelling and simulations of reinforced concrete

columns under axial loading subjected to lateral impact. As the validation process

mainly focuses on the vulnerability assessment of axially loaded columns it is

extended to eccentrically loaded columns by using some of the investigation reported

in literature. This chapter also describe an experimental setup reported in the literature

the data from which have formed the basis for the validating the Finite Element (FE)

model.

Chapter 4 presents the methodology used for impact reconstruction with the aim of

developing a comprehensive method for vulnerability assessment that can be applied

to a general vehicle population including new generation vehicles. Impact pulses

generated from full scale impact tests are used for impact reconstruction. The dynamic

response of the impacted column and method for vulnerability prediction are also

discussed in this chapter.

Chapter 5 investigate the effects of enclose lateral confining steel reinforcement in

columns subjected to axial load and lateral impact loads. The confining model

proposed by Mander (1988) is used as the basis for the study. This chapter also

describe sensitivity analyse of key design parameters to lateral impact of axially

loaded columns. Based on the sensitivity analyses, analytical equations are developed

to quantify the critical impact pulse parameters for columns of specific diameters and

concrete grades. Investigation extended to quantify the residual capacity of the

partially loaded columns to make a decision whether or not the impacted column has to

be replaced or repaired for further use. Damage index D is used to identify the capacity

degradation of the impacted columns. The collapse of the impacted columns is

investigated by varying the parameters so that there are substantial warnings before

collapse following the post impact loading.


1-13

Chapter 6 describes the severity of eccentrically loaded columns subjected to

transverse impacts. The eccentrically loaded columns are divided into two groups

based on the direction of the impact in the plane of bending. To facilitate application of

the eccentric load the column is provided with a bulk head; the impact behaviour is

then investigated in detail.

Chapter 7 reports the impact behaviour of RC columns under biaxial bending and

extend the parametric analysis to columns under biaxial bending. Analytical equations

are derived for the bi-axially loaded columns; single axis bending is treated as special

case of a biaxial bending with zero moment about one axis. This chapter deals

exclusively with the orientation of the impact for 50% and 20% fully loaded columns

with the aim of defining three consecutive points on the interaction diagram for one

particular column so that linear interpolation can be used to predict the critical impulse

for points in-between. These equations are exclusively valid between 0o to 90o impact

angles. In fact, having provided that columns under positive eccentric loads can be

treated as non-critical, the equations can be extended to account for the other impact

angles.

Chapter 8 Summarises the main conclusions that have emerged from this thesis along

with their practical implications.


2-15

2. LITERATURE REVIEW

2.1 Characteristics of impact pulses

The duration of the typical vehicle frontal impact is around 100ms. Separation occurs

soon after the impact if there is no post collision speed or further impact. However, in

the real world, depending upon the circumstances of the impact and the type of vehicle

involved, the duration of the impact can vary. If the crash is off-set, that means if the

vehicle hit at an angle, then the duration of the impact will be longer. Depending upon

the angle of crash, offset collisions can result in approximately a 200ms impulse. If the

vehicle has a stiff front, the duration of the impact will be less because of smaller

deformation. An example of an impact sequence of a vehicle (truck) is shown in

Figure 2.1, where the total duration of the impact is around 0.18 seconds. The duration

and the force generated from an impact can also be taken as a measure of the

vulnerability caused by the impact.

Figure 2.1: Sequence of an impact (El-Tawil et al. 2005)

The force versus time response generated by the transverse impact of a Chevy truck for

various approaching speeds for two different pier configurations namely pier I and II,

is shown in Figures 2.2(a) & 2.2(b). The impact force versus time functions records

contain several small amplitudes followed by a large spike irrespective of the approach

velocities of the vehicle. The sharp peaks occur when the stiff and heavier components

such as chassis or engine block reach the pier and interact with it. As the approach


2-16

speed increases the first peak occurs earlier in time. It is also important to note that pier

I and II has two different pier configurations, and hence, different stiffness.

Comparatively higher peak force is observed in Figure 2.2(a), as the pier ‘I’ has the

higher level of stiffness.

Figure 2.2:Impact force vs. time histories for Chevy truck at various speeds

(El-Tawil et al. 2005) According to El-Tawil et al. (2005) the peak force generated at the impact is not

representative of the design structural demand, as the structures do not have enough

time to respond to a rapid change of loading. Chopra (2001) suggested that the

equivalent static force, which is defined as the static force necessary to produce the

same deflection at the point of impact as that produced by dynamic event, is a more

appropriate measure of the design structural demand. However, this displacement

based criteria for failure is distinguished from the traditional strength or strain based

criteria at the threshold of damage, and applicability of these criteria to assess column

response under high velocity impact may be questionable and requires needs careful

further considerations (Tsang et al. 2005).

2.2 Behaviour of structural elements under impact loading

Behaviour of structural elements under dynamic loading conditions is quite different

from the behaviour under static loading conditions. Dynamic loads, such as impact

loads, give rise to accelerations of the structural elements and kinetic energy and the

inertia effects must be considered in the analysis. Structural element subjected to

dynamic loading conditions must have higher energy absorption capacity, and

therefore, should be designed to allow for plastic deformations. The plastic

deformation capacity will improve the ductility of the element and hence prevent the


2-17

occurrence of a brittle failure mode under severe impact conditions. In addition, it is

important to utilize more flexible elements, which can undergo larger deformations

and thereby, obtain a larger energy absorption capacity.

Reinforced concrete (RC) elements subjected to dynamic loading conditions will

exhibit two forms of failure modes namely, flexure and shear (Johnny 2003). Flexural

failure will often result after the formation of plastic hinges at the locations where the

ultimate moment capacity is reached. This failure mode is characterised by initial

cracking of the concrete, subsequent yielding of the tensile reinforcement and

ultimately compression failure of the concrete. Also, this failure mode is rather ductile

and absorbs energy during impacts. Contrary to the flexural failure mode, the shear

failure mode is catastrophic and brittle in nature, which severely hampers the energy

absorption capacity of the element. The ultimate moment capacity of an element

cannot be obtained and diagonal tension cracks will form close to the supports,

followed by initial tensional cracks that develop at the points where the maximum

moment is reached. Premature failure will result.

2.3 Dynamic impact tests on reinforced concrete columns

Previous research on columns has mainly focused on improving the axial load carrying

capacity and stiffness, while improvement of impact resistance has been largely

unexplored. The few investigations conducted on laterally impacted columns highly

emphasised the importance of the stain rate effects. Some of the test results indicated

that the increased structural resistance is somewhat greater than the commonly

accepted maximum increase of 30% of the static resistance (Louw et al. 1992). Strain

rate effects, as well as the behaviour of the vehicle during the impact, are of primary

importance as far as the structural response is concerned (Prasad 1990). Thus the

impact is classified as soft or hard, based on the way that impact energy absorbs during

an impact. Generally, in a soft impact the striker absorbs most of the kinetic energy

through plastic deformation, while the structure experiences minor deformations.

2.3.1 Columns subjected to soft impact Leodolft (1989) tested thirty-nine 350x150x1600 mm reinforced concrete columns


2-18

under soft impact conditions. The soft impact condition was achieved by inserting a

pipe buffer system in between the pendulum and the column. The columns were

axially preloaded with 100 kN and 200 kN forces and subjected to an impact velocity

of about 7ms-1. The applied loads were sufficient to permanently damage the impacted

columns.

In this experiment, the peak load occurred later and is more likely to influence the

flexural shear resistance of the element, than its pure shear and inertia stiffness. During

the first 10 ms, the buffer system was subjected to elastic-plastic deformations. By that

time, substantial energy had been transferred to the column, which deflected

significantly. The generated axial load from the impact was increased as the column

increases in length and subsequently decreases and remained compressive. In addition,

the strain rate of up to 10-2 was generated at the rear surface of the column. It was

observed that the partially damaged columns exhibited the same static lateral capacity

as the undamaged columns. Moreover, the impacted columns were subjected to a

series of peak shear and corresponding moments and peak moments and

corresponding shear. According to the test results, it was concluded that the

dynamically loaded slender columns are considerably stronger than the ultimate load

predicted by the modified ACI equation for the slenderness ratio.

2.3.2 Columns subjected to hard impact However, during a hard impact, the kinetic energy of the striker is mostly absorbed by

the structure and the striker itself suffers small deformations. Fererabend (1988),

conducted an experimental investigation on 300x300x4000mm reinforced concrete

columns subjected to lateral impact at mid span. The columns were tested in a

horizontal position, where one end was restrained using a 20t mass to simulate the

inertial restraint provided by a bridge deck. The axial load was applied by pulling the

free sliding end using external prestressing bars towards the stationary end. The

impact load was generated by dropping a 1.14t mass onto the column at mid span and

the shear reinforcements were provided to ensure a flexural failure of column. An

important feature of the impact behaviour of that column was the initial increase in

axial force as the column lengthened along its centre line. The authors also observed

that the initial peak of the applied impact load depended on the inertial characteristics


2-19

of the column and the boundary conditions. Under the effects of the impact load, the

column experienced shear deformations while local deformations occurred at the point

of impact. Even though these deformations were relatively small, the initial impact

force had a high initial peak. The initial force was opposed primarily by the inertia

forces of the element. The shear stiffness of the column was the main parameter that

controlled its response. As the shock wave progressed through the cross sections of the

elements, they were subjected to fluctuating moments, shear forces and axial loads.

After observing these responses the author has emphasised the impotency of the all

these forces in determining the critical section.

By assuming a 10% increment of the material properties due to strain rate effects, the

dynamic moment capacity of the tested column exhibited 20% increment compared to

that of its static value. On the other hand, observed dynamic shear capacity of the

column was substantially greater than the ultimate static shear capacity of the column.

Therefore, it was concluded that the initial peak shear force generated during hard

impact, is not an indication of the ultimate structural resistance of the column when

adequately reinforced in shear. In addition, under the hard impact condition the

moment-shear combination moves from a low moment high shear value to a higher

moment much lower shear value. Therefore there is a possibility to generate initial

shear cracks in a section which probably diminishes the flexural resistance that

follows.

2.3.3 Columns subjected to axial impact The dynamic buckling response of columns under axial impact loads has been

subjected to extensive investigation over the past decades. The interest was mainly

focused on the behaviour of the short and slender columns under eccentric loads. In

addition, parametric studies have been conducted on element aspect ratio, element

formulation, boundary conditions and geometric imperfections. Most of the

researchers selected low velocity hard impact conditions with fairly large masses.

Kenneth et al. (1964) conducted research on a total number of 205 plain and RC

columns under concentric and eccentric loading conditions. In addition, two

foundation conditions were also simulated by using rigid pads and rubber supports. In

this experiment, the rise time was defined as the time between the commencement of


2-20

the load and the ultimate load. The average rise time in all the dynamic tests was about

30ms. Longitudinal inertial forces were neglected because the longitudinal natural

period of the columns was less than 4.0 ms for all the specimens. The results of this test

showed that even though the ultimate load on statically loaded columns agreed well

with the ACI 318-02 (2002) code specifications, dynamically loaded columns

exhibited a 30 to 40% increment compared to the statically loaded columns. In

addition, columns on elastic foundations, such as rubber or soil, were stronger when

loaded dynamically than similar columns on rigid supports. The test results further

indicated that, the ultimate strength of short dynamically loaded columns can be

computed from the ACI equation, after being modified to account for the strength

increment of the material due to strain rates. However this was valid only for the strain

rates in which the inertial forces are negligible.

2.3.4 Shortcomings of the individual column tests There are several disadvantages associated with individual column tests (Gebbeken et

al. 2007). The main disadvantage being the idealized boundary conditions. The

flexibility of the realistic support conditions was not taken into account in these tests.

This factor can shift the location of the plastic hinge and consequently, the failure

mechanism would be different from the usual fixed assumption. In addition, the effects

due the wave reflection at the boundaries cannot be neglected. At free boundaries, the

compressive wave is reflected as a tensile wave, while at fixed boundaries, the

reflected wave becomes a compressive wave. The small models are the ones that suffer

most due to the boundary conditions (Gebbeken et al. 2007). Even though the shear

cracks, spalling of the concrete cover, and confinement failure are the ideal failure

modes for the individual columns, the effects of the global structural configuration

cannot be neglected.

2.4 Dynamic tests on reinforced concrete beams

Depending on the nature of the transient dynamic load, an element can be subjected to

higher modes of vibrations. Even though the amplitudes of the higher modes are

relatively small, they can give rise to larger shear forces in the element. Hughes and

Speirs (1982) performed a theoretical and experimental investigation on the behaviour


2-21

of concrete beams subjected to mid span impact by a falling mass. They mainly

focused on the static and the dynamic vibration modes of the beams under the

pin-ended conditions. By comparing the static mode with first and third free vibration

modes under equal potential energies, a relationship between the mode of vibration

and the mode of failure was derived. The mode profiles considered are illustrated in

the Figure 2.3.

The figure shows reduction in displacement and bending moments for the first and

third modes, compared to that of the static mode. However, the shear forces are greater

in the dynamic modes and as a consequence, beam could fail in flexure under static

loading, whereas an identical beam might fail in shear under dynamic loading. The

experimental results of Takeda et al. (1977) support these arguments. They have

demonstrated that the reinforced concrete beams with high shear could fail in a ductile

manner under static loading, but in a brittle manner under dynamic loading with some

reduction in shear strength of concrete.

Figure 2.3: Comparison between different modes of vibration with equal potential

energies (Hughes and Speirs 1982) Niklasson (1994) investigated the impact behaviour of simply supported reinforced

concrete beams with different amounts of reinforcement and different concrete grades.

The impact loads were applied as symmetrical point loads on either side of the

mid-span. In this experiment, the rise time and the amplitude of the applied loads were

controlled by placing rubber pads at the interface between the beam and the impacted

mass. This diminished the magnitude of the applied load, while increasing the rise


2-22

time. The increased rise time led to low impact frequencies, compared to a short rise

time. Therefore, the number of modes of vibration excited by the loads differed from

each other, which led to a different failure mode. In addition to the rise time, the

natural frequency of vibration may also influence the response of the structural

elements (Niklasson 1994). Therefore, in order to determine the impact response of a

beam, the correlation among the load duration, rise time of the load and the natural

period of vibration must be taken in to account.

When the loading pulse has a short duration and a high peak value, the probability of

shear failure increases (Kishi et al. 2002). Other than the load intensity, Kishi et al.

(2002) also showed that beam stiffness played an important role in the change of

failure mode. By conducting research on concrete beams without stirrups, the authors

concluded that beams with a low amount of reinforcement may fail in flexure, whereas

beams fail in shear when the lateral reinforcement amount was doubled. However,

when the velocity of the impact increased further, beams carrying a low amount of

lateral reinforcement also failed in shear. Based on similar experimental results, Palm

(1989) also made similar conclusions. Ansell (2005) points out that a concrete

structure, designed to fail in flexure under static loads, may fail in shear when loaded

dynamically.

Moment and shear waves generated by the impact loads will travel from the impacted

zone towards each support, as described by Hughes and Speirs (1982). Under these

circumstances, large bending moments and shear forces will generate along the beam,

which differs from the shear and moment distribution in a static load case. This will

result in local failure with the mass punching a cone out of the beam, due to the

concentration of large shear forces in that area, usually referred to as punching shear

failure (Anzell 2005).

Bentur et al. (1986) conducted a series of tests on conventionally reinforced concrete

beams under impact loading conditions. The impact load was generated by dropping

345kg mass from a 3m height onto the 1.4x0.1x0.125m beam with a spanning length

of 0.960m. According to their observations, concrete can withstand a higher bending

load under impact than under static conditions and can absorb more energy under


2-23

impact loading conditions. Further, steel deformation under the impact load was

confined to a region only few centimetres long, located beneath the point of impact.

The deformation caused by the impact exceeded the maximum strain capability of the

steel. This means that there was not enough time to develop extensive bond slip along

the length of the bar under impact conditions. Most of the time, the steel failed in a

ductile manner at the point of the impact. However, under quasi-static loading, beams

which were deflected to the same degree showed no evidence of reinforcement failure.

Instead, significant cracking and debonding of concrete along the reinforcing bar was

observed.

Kulkarni and Shah (1998) investigated the strain rate effects of concrete beams by

applying the impact force using hydraulic system. The beams were loaded at a rate of

0.00071 cm/s and 38 cm/s. The tests on seven pairs of singly reinforced simply

supported concrete beams (without shear reinforcements), showed some reduction of

the total number of cracks at high strain rates. Also, there wasn’t any sharp ‘Yield

Point’ or a ‘Yield Plateau’ in the load-deflection curves for beams failing in flexure at

the high strain rate. Standard sectional analysis using a rate-dependent constitutive

relationship did not adequately predict the shape of the high-rate load deflection curve,

and the localised yielding of steel at higher strain rate was believed to be one of the

reasons behind this observation. In addition, the final failure mode shifted from shear

failure at the static rate, to flexural failure at the high rate, contrary to what was found

by other researchers.

Small-scaled cantilever beams, having different shear span to depth ratios and stirrup

spacings, were tested by Chung and Shah (1989) to investigate the effect of loading

rate on bond in beam-column joints. The generated strain rates were in the range of

0.004-0.08s-1. Reduction of the cracks in the specimens was observed due to increased

tensile strength of the concrete and bond strength between concrete and steel. The

improved bond strength at high strain rates also led to a stress concentration in the

reinforcements. As a result, the steel yield earlier and lead to a lower ductility ratio at

failure. The stiffness of the beams also affected by the loading rate and different failure

modes were observed at different loadings rates. It is evident that the confinement

effects act as a governing factor of the failure modes. For stirrup spacing of s/2 the


2-24

specimens failed in brittle and ductile modes under dynamic and static loading

conditions respectively. For specimens with stirrup spacing of s/4, the loading rate

appeared to have no effect on the failure mode.

2.5 Behaviour of concrete under impact loads

The behaviour of concrete under the effects of high strain rate has been extensively

studied over the past few years. According to the Watstein (1953), the initial

investigations carried out by Jones and Richard (1936) and Granville (1938)

concluded that the compressive strength of concrete increased with the rate of loading.

Watstein (1953) suggested that there was an increase of over 80% in compressive

strength for concrete loaded at a strain rate of 10s-1.

Tests carried out on plain concrete specimens under uniaxial loading conditions

revealed that compressive impact strength can be as much as 85-100 % higher than the

static strength of concrete (Bischoff and Perry 1995). However, with respect to the

critical axial strain values at peak stress, there is no general agreement about changes

in the deformation. Material properties such as the compressive strength, the poison’s

ratio, the volumetric strain and the ductility, may increase as the strain rate increases

beyond the static value. Particularly, uniaxial compressive strength of plain concrete

increases linearly with the logarithmic increase in strain rate (Mainstone 1975; Suaris

and Shah 1982). The secant modulus of elasticity may also be changed under high

strain rates. But there is no evidence of change of elastic modulus or initial tangent

modulus due to the effects of strain rate (Bischoff and Perry 1995).

Concerning the energy absorption, it appears to depend to a large extent on the strain

capacity of concrete, which in turn is governed by the failure mode. This means that

the higher grades of concrete will show less absorption of energy at failure contrary to

what would be expected (Georgin 2003). In addition, Fu et al. (1991a) observed that

concrete fails in an explosive manner under very high strain rates. Theoretically,

energy absorbed per unit volume of a material during an impact, can be written as;

E22σ Eq. 2.1

where σ is the stress of the material and E is the modulus of elasticity of the material.

This means that small portion of a member, where the highly localised stress occurs,


2-25

absorbs an excessive amount of energy before the main portion of the member can be

stressed appreciably. As a result, the small portion where the localised stress occurs is

likely to be stressed above the yield stress of the material. Therefore, even if the

material is relatively ductile, the energy absorbed by the local area of the member may

be subjected to rupture. In fact, this may account for explosive manner under which

concrete fails in very high strain conditions.

Hughes et al. (1972) investigated the compressive strength and ultimate strain of

concrete by using the drop hammer test. They investigated the effects on compressive

strength, ultimate strain, energy absorption and deformation modes of concrete cubes

under impact loading. Trapezoidal load function was assumed for both, the hammer

impact pulse and cube impedance function and hypothetical load and strain records

were derived. Concrete with varying mix proportions and two different course

aggregates were examined and the resultant compressive strength and ultimate strain

were reported.

The compressive strength of cubes tested at a stress rate of less than 160 kPa/s and

rates of strain of less than 8s-1 was almost the same as the static compressive strength.

Increment of the strain rate beyond 8s-1 caused considerable effects on the

compressive strength. For example, an average increment of the compressive strength

of 28 days concrete cube was 11% greater when it was tested at a strain rate of 10s-1.

When the strain rate was increased to 14s-1, the strength was increased to 25%. From

these test results, it can be concluded that the concrete gains a substantial increase in

compressive strength under the effects of high rate loading. On the other hand, this

enhancement of strength may change the failure mode of concrete from ductile to

brittle. Therefore, omission of the loading rate may cause substantial errors in both,

prediction of failure mode and magnitude of impact response.

According to Fu et al. (1991b), early studies on split cylinder test indicate that tensile

strength of concrete increases with increasing stain rate. Comparing results between

dynamic and static tests, Cowell (1966) observed an increasing tensile strength of

18-65%. Tekeda and Techicawa (1971) obtained a 70% increment in the tensile

strength of concrete. These test data were obtained using different test set ups,


2-26

specimen dimensions and material properties, and hence direct comparison would be

difficult. However, the general perception is that the strength increment due to strain

rate effects is more pronounced in tension than in compression (Barpi 2004).

According to Figure 2.4 the considered strain rates are less than 1s-1. However the

difference is become even large beyond this strain limit (CEB-FIP 1990).

Figure 2.4: Strain-rate sensitivity for concrete in compression, tension and flexure

( Suaris and Surendra 1985)

Analytical stress-strain curves derived by Suaris and Surendra (1985) also indicated

the higher strain-rate sensitivity in tension compared to the compression. The

graphical representation of the suggested curves is shown in the Figure 2.4. Test results

on compressive, tensile and flexural responses of concrete also indicated that the

amount of strength increase is the highest for concrete under tension and lowest for

concrete under compression (Tekeda et al. 1977). The strength increment under

flexure lies in-between the increments gained under tension and compression. This

means that concrete elements under flexure will exhibit higher strain rate sensitivity

than the elements under compression (Sukontasukkul and Mindess 2003).

The water/cement ratio has a considerable influence on concrete behaviour at high

strain rates. Kaplan (1980) tested a large number of concrete specimens to investigate

the relationship between the concrete strength and loading rate for concrete with

various moisture contents. He concluded that the moisture content in concrete is one of

the principle variables affecting the relationship between strength and the loading rate.

When concrete cubes with a low water/cement ratio were tested under high strain rate,


2-27

a considerable increment of the compressive strength was observed compared to that

of the cubes tested under static loading conditions (Cowel 1966). Decreasing the

water/cement ratio from 0.98 to 0.63 increased the static compressive strength only by

3.5 times the average standard deviation. However for the impact tests, it increased the

dynamic compressive strength by 14 times the average standard deviation. More

recently Reinhardt et al. (1990) extensively investigated the influence of the free water

content on the behaviour of micro-concrete, maximum aggregate size of which was

2mm, under strain rate from 0.25 to 1.25s-1. Dry specimens showed a small increase of

strength of 1.45, while wet specimens showed a larger increase of 4.1.

2.6 Dynamic properties of concrete and steel

2.6.1 CEB-FIP specifications for concrete Concrete is very strain rate sensitive. In the CEB-FIB Model Code (1990), there is a

relationship for DIF (Dynamic Increase Factor) for compression and tension at varying

strain rates. The DIF in the code is a design value, which means that the given strength

increments are lower than the values obtained from the experiment. For a given stress

rate, the compresive strength under high rates of loading may be estimated from the

following equations (CEB-FIP 1990).

2.6.1.1 Modified strain rate for concrete in compression

( ) s

coccmimpc ff ασσ && /, = for sMPac /106≤σ& , Eq. 2.2

( ) 3/1, / cocscmimpc ff σσβ &&= for sMPac /106>σ& , Eq. 2.3

cmocms ff610

1

+=α , Eq. 2.4

and 26log −= ss αβ , Eq. 2.5

where,

fc,imp is the mean impact compressive strength,

cσ& is the stress rate (MPa/s) valid in the range 1 MPa/s < cσ& < 107 MPa/s,

fcm is the mean concrete compressive strength,

fcmo = 10 Mpa,

sMpaco 1−=σ& .


2-28

Similarly, for a given strain rate the compressive strength may be estimated from,

( ) s

coccmimpc ff αεε 026.1, / &&= for 130 −≤ scε& , Eq. 2.6

( ) 3/1, / cocscmimpc ff εεγ &&= for 130 −> scε& , Eq. 2.7

215.6log −= ss αγ , Eq. 2.8

where,

cε& is the strain rate valid for 30x10-6 < cε& < 3x102 s-1,

coε& = -30x10-6 s-1, Eq. 2.9

cmocms ff610

1

+=α . Eq. 2.10

2.6.1.2 Modified strain rate for concrete in tension Compared to other materials, concrete exhibits higher strain rate sensitivity under

impact loading, due to scale size of the heterogeneity (Weerheijm and Doormaal 2007).

Especially, the tensile strength exhibits a strong increase beyond loading rates in the

order of 10MPa/s. For a given stress rate, the tensile strength under high rates of

loading may be estimated from the following equations.

( ) s

ctoctctmimpct ff δσσ && /, = for sMPact /106≤σ& , Eq. 2.11

( ) 3/1, / ctoctctmimpct ff σσ &&= for sMPact /106≥σ& , Eq. 2.12

with cmocm

s ff610

1

+=δ , Eq. 2.13

377log −= sδλ , Eq. 2.14

where,

fct,imp is the mean impact tensile strength,

ctσ& is the stress rate (MPa/s) valid for 0.1 MPa/s < ctσ& < 107 MPa/s,

fctm is the mean tensile strength,

=ctoσ& 0.1MPa/s,

fcmo = 10 MPa.

Similarly, for a given strain rate the tensile strength under high rate of loading may be


2-29

estimated from,

( ) s

ctoctctmimpct ffδ

εε 016.1, / &&= for 130 −≤ sctε& , Eq. 2.15

( ) 3/1, / ctoctsctmimpct ff εεβ &&= for 130 −> sctε& , Eq. 2.16

33.211.7log −= ss δβ , Eq. 2.17

where

ctε& is the strain rate (s-1) valid for 1216 103103 −−− ×<<× ss ctε& ,

octε& = 3×10-6 s-1.

2.6.1.3 Modulus of elasticity

The strain rate dependence of the elastic modulus is also included in the CEB-FIP

Model Code 1990, as presented in Figure 2.5 with the governing equations. The effect

of stress and strain rate on modulus of elasticity may be estimated from,

( ) 025.0, / cocciimpc EE σσ &&= , Eq. 2.18

( ) 026.0, / cocciimpc EE εε &&= , Eq. 2.19

where,

cσ& is the stress rate (MPa/s),

ε& is the strain rate (s-1),

Figure 2.5: Model for the strain rate dependency of concrete elastic modulus

according to the CEB-FIP Model Code 1990 (CEB 1993)


2-30

coσ& = -1.0 MPa/s and ε& co = -30×10-6 s-1 for compression,

ctoσ& = 0.1 MPa/s and ε& cto = 3×10-6 s-1 for tension,

Ec,imp is the impact modulus of elasticity,

Eci is the modulus of elasticity (MPa) at a concrete age of 28 days, obtained from the

following equations:

( )[ ] 3/1/' comccoci fffEE ∆+= , Eq. 2.20

Where,

f ’ c is the characteristic strength (MPa),

∆f = 8 MPa,

fcom = 10 MPa,

Eco= 2.15×104 MPa.

Similarly, when the actual compressive strength of concrete at an age of 28 days f ’ c is

known, Eci may be estimated from,

[ ] 3/1/ cmocmcoci ffEE = . Eq. 2.21

In the situations where only an elastic analysis is carried out for a structure, the

reduced modulus of elasticity Ec should be used in order to account for the initial

plastic strain. The reduced modulus of elasticity can be calculated as,

Ec= 0.85Eci Eq. 2.22

Values of the tangent moduli Eci and the reduced moduli Ec for different grades of

concrete are given in the Table 2.1.

Table 2.1: Tangent moduli and reduced moduli of elasticity

According to the CEB-FIP Model Code 1990 (CEB 1993), the strain rate of 3×10-5s-1 is

considered as the datum where the dynamic loading rates begin. As a comparison, a

strain rate of around 0.01s-1 can be expected for concrete columns subjected to soft

impact loading (Louw et al. 1992). Many researchers have comprehensively

investigated the compressive and tensile strength of concrete at different strain rates

(Magnusson 2007). In their works, it was observed that there is an increasingly large

scatter in the test results for increasing strain rates (see Fig. 2.6). The reason for this

Concrete grade C12 C20 C30 C40 C50 C60 C70 C80 Eci (GPa) 27 30 34 36 39 41 43 44 Ec(GPa) 23 26 29 31 33 35 36 38


2-31

relatively large scatter would have been the differences of the experimental techniques

and the method of analysis that they employed in the testing procedure (Bischoff and

Perry 1991). On the other hand, the results may have been influenced by factors such

as specimen size, geometry, aspect ratio and the moisture content in the tested

specimens (Magnusson 2007; Bischoff and Perry 1991). However, in contrast to this

scattered behaviour, there is a relatively sharp transition zone with different strain rates.

The sharp transition zone is common for both compressive and tensile strengths and

each portion can be represented by straight lines, which consist of a moderate

increment followed by a steep increment (see Fig. 2.6 & 2.7).

Figure 2.6: Strain-rate influence on the compressive strength of concrete

(Bischoff and Perry 1991) Many researchers commented on the first moderate increment in strength, as

summarised by the Johansson (2000). Many of them observed that wet concrete

specimens are more strain-rate sensitive than dry specimens during the first moderate

increment. Therefore, most of the explanations were based on the viscose effects

produced by the free water inside the micropores. For example, when the specimen is

loaded in compression, the free water in the specimen is forced to move inside, which

results in a build-up of internal pressure. This pressure improves the capacity of the

materials to resist external loads, delays the crack initiation and improves the

compressive strength. The water trapped inside the micropores not only helps to

increase the compressive strength of the concrete, the thin films of water trapped

between the particles aid the reduction of movement between aggregates and hence


2-32

increase the tensile strength.

Figure 2.7: Strain-rate influence on the tensile strength of concrete

(Malvar and Crawford 1998-a). As the strain rate exceeds the transition zone, both compressive and tensile strength

show a sharp increment (see Fig. 2.6 & 2.7). This is mainly caused by the inertia

effects and the lateral confinements of concrete. Weerheijm (1992) investigated the

effects of changes in the strain rate on tensile strength of concrete. They suggested that

changes in the stress and energy distributions due to inertia effects around the crack

tips are the cause of the sharp strength increment. However, under compressive

loading conditions the process is affected by the propagation of micro-cracks around

the crack-tips, and hence, the explanation may not be valid under compressive loading

conditions.

A reasonable explanation of the behaviour observed under the high compressive strain

rate conditions was given by Bischoff and Perry (1995). Their argument was that when

it comes to the crack-propagation, the crack-propagation velocity must have an upper

physical limit. When the stress rises faster than the time needed for propagation of

cracks, an apparent delay in the crack propagation occurs and hence an increase of

strength can be expected. They also considered the effects of lateral inertia

confinements. This effect can be described by the following hypothesis. An elastic

material subjected to axial compression should exhibit lateral expansion due to the

effects of Poison’s ratio. However due to inertial confinements, a cylindrical specimen

subjected to a rapid axial load increments is not able to expand in the radial direction at


2-33

an instant. This creates conditions similar to those in a specimen that is subjected to a

uniaxial stress conditions under increased lateral confinements. Even though this

condition prevails only until the material accelerates in the radial direction, a

substantial increase in compressive strength can be generated (Bischoff and Perry

1991).

Figure 2.8: Model for the strain rate dependency of concrete in compression and

tension according to the CEB-FIP model code (CEB 1993) and with the modified

model according to Malvar and Crawford (1998-a)

A model for the strain rate dependency of concrete in compression and tension was

presented in the CEB-FIP Model Code 1990 (CEB 1993) and is valid for strain rates of

up to 300s-1. According to the CEB-FIP code, the rapid increase in the material

properties for both, compressive and tensile loading starts at strain rates greater than

30s-1 (see Fig. 2.8(a) & (b)). However, based on the extensive amount of reliable

experimental results, Malvar and Crawford (1998-a) suggested that the rapid change

of the tensile strength should start around 1s-1, instead of 30s-1. Therefore, they

proposed a formulation similar to the CEB-FIP model, which was fit against the test

data given in Figure 2.7 and their results with the CEB-FIP specifications are shown in

Figure 2.8(b).


2-34

2.6.2 Dynamic properties of steel The stress-strain behaviour of steel is particularly sensitive to the loading rate and this

phenomenon is known as strain rate sensitivity. As far as energy absorption is

concerned, the strain rate sensitivity plays an equally important role to that of the

inertia effect of the material. It clearly reflects from the load-displacement curve of the

material, which was tested under various uniaxial compression strain rates (Marsh and

Campbell 1963).

Information from the existing literature on the effects of strain rate on the yield

strength of reinforcing steel has been summarised by Lu et al. (1991b). According to

Lu et al. (1991b), Cowell (1966) found that strength increment for steel with static

strength 351MPa when tested under strain rates of 0.03s-1, 0.1 s-1, 0.3s-1 and 1.0s-1

respectively was 10%, 13%, 17% and 19%. However, for steel with yield strength of

264 MPa loaded at similar strain rates, the corresponding increment rose to 25%, 33%,

38% and 53% respectively. Similar observations were also made by Norris et al. (1959)

concerning static yield strength of 330 MPa and 278 MPa, when tested under similar

conditions.

Wakabayashi et al. (1980) performed a tensile test on round and deformed steel bars

with a 13mm diameter. The measured stress-strain curves showed that with increasing

strain rate both, the upper and lower yield stress increased. Compared to the yield

strength at quasi-static rate, the average increase in lower yield strength was 7-8% at a

strain rate of 0.005s-1 and 16-18% at a strain rate of 0.1s-1. A similar increment for the

upper yield strength of the material was also observed.

To account for the strain-rate effects in numerical applications, it is more desirable to

have an explicit rate-dependent constitutive equation. Numerous attempts have been

made to generate an effective constitutive model to describe the strain-rate sensitivity

of the material. Cowper-Symond’s relation is the most widely employed

rate-dependent constitutive equation, applicable particularly for solving impact

problems. It was also found to be reliable for considering strain rate effects. This

relationship basically represents a perfectly plastic material with a dynamic yield or

flow stress that depends on strain rate. According to Reid and Reddy (1986), the


2-35

Cowper-Symond’s relationship can be widely used to account for strain rate effects in

dynamic structural plasticity problems. In addition, this equation gives dynamic flow

stresses, which agree well with the dynamic uniaxial tension and compression test

results on several materials (Jones and Wierzbicki 1983). Other than the factors

mentioned before, the major advantage of this constitutive equation is its portability

with finite element programs. This means that the data required to generate strain rate

effects can be directly fed in to the finite element program and hence be extensively

used for research in this thesis. The Cowper-Symond’s constitutive equation is noted

as follows:

( ) q

csd D1

1 εσσ &+= , Eq. 2.23

where, Dc represents a characteristic strain rate, q is a measure of the rate sensitivity of

the material, ε& is the strain rate and σd and σs represent the dynamic and static stress

of the material respectively.

Malver (1998) proposed another equation for steel reinforcing bars produced under

ASTM standards. The equation is particularly valid for the steel bars with yield stress

ranging from 290 MPa to 710 MPa and for strain rates between 10-4 and 10s-1. The

Dynamic Increasing Factor (DIF), which is defined as the ratio of the dynamic to static

yield stress, was used to represent the influence of strain rate on strength enhancement

under dynamic conditions. To derive these equations Malvar (1998) used several test

results available in the literature. It is evident that under dynamic loading for a strain

rate of up to 10s-1, the strength properties of the reinforcing bars increased up to 60 %.

For determining the yield strength and ultimate strength for reinforcing bars at

different strain rates, he proposed the following formulation of the DIF:

αε

= −410

&DIF , Eq. 2.24

where for yield stress α = αfy

−=

414

040.0074.0 y

fy

fα , Eq. 2.25

and for the ultimate stress α = αfu

−=

414

009.0019.0 y

fu

fα , Eq. 2.26

where fy is in MPa.


2-36

Based on the existing DIF data for both yield and ultimate strengths, it is concluded

that the DIF is inversely proportional to the yield strength. This means that the same

argument of higher strength gain for lower strength materials under dynamic loads

appears to be applicable in both concrete and steel. Generally, the yield strain, the

strain at which strain hardening begins, as well as the length of the yield plateau in the

stress-strain diagram of steel, will increase at higher strain rates (Lu et al. 1991b). But

there are no significant effects of the loading rate on the modulus of elasticity and

ultimate strain for steel (Wakabayashi et al. 1980; Malvar 1998).

2.7 Interaction between reinforcement and concrete

Chemical adhesion, frictional resistance and rib-bearing are the main components of

interaction between reinforcement and concrete. According to Luts et al. (1967),

chemical adhesion is the main resisting mechanism for very small values of bond

stress in order of 200psi (1.38MPa). When bond stresses increase further, chemical

adhesion is replaced by the wedging action of the ribs (Malvar 1992). Longitudinal

and radial cracks can be generated under the influence of the wedging action of the ribs

and if adequate confinement is not provided, bond failure occurs soon after the cracks

propagate to outer layers of the concrete. If proper confinement is provided, bond

stress reaches a maximum of 3cf ′ and meanwhile the frictional type of forces are

involved gradually to provide the required bond.

2.7.1 Static bond slip analysis Menzel (1939) conducted a series of tests to investigate the effects of surface

conditions, cement ratio, embedment length and position of the bar relative to

placement direction of the concrete. Menzel identified the marked effects of surface

conditions on bond resistance and concluded that increased cement ratio or increased

embedment resulted in increased bond resistance.

Experiments conducted by Ferguson and Thompson (1962) revealed that bond

between reinforcement and concrete was a function of development length, and not

bar size. They emphasised the adequacy of bar cover to acquire the required

development length and found that ultimate bond stress is proportional to squire root


2-37

of compressive strength of concrete. Later in 1965, they observed that increased

embedment length resulted in decreased the bond stresses. Also, crack growth in

concrete tended to be more severe for larger bars than for smaller bars. Finally,

increased cover caused increase in bond resistance, however was not helpful in

reducing crack width.

McDermott (1969) investigated the influence of steel strength and reinforcement ratio

on the mode of failure and strain energy capacity of reinforced concrete beams. It was

observed that within the ductile range, the yield strength of the steel bars had no effect

on the strain energy of beams of equal static bending strength that were subjected to

moderate strain rates. Therefore, it was concluded that ductility of reinforced concrete

beams is independent from yield strength of the reinforcement.

Malvar (1992) conducted a series of tests to investigate the effects of confinements on

the bond of reinforcement. In this experiment, 12 specimens were subjected to

confining stresses in the range of 3.45-31MPa. Instead of selecting a conventional two

dimensional surface to simulate the bond between steel and the concrete, a 73.5mm

process zone was defined surrounding the reinforcement. The confinement effects and

the bond slip condition were described with respect to this arbitrarily selected process

zone. When the confinement stress increase from 3.45 to 31.0MPa, almost 200 percent

increment of bond strength was observed. However, for higher confining stresses, the

effects of confinement on bond behaviour appeared less pronounced. The same

conclusion was made by Robins and Standish (1982), who conducted pulled out tests

on 8 and 12mm bars in a 100mm concrete cube. The cubes were laterally loaded on two

opposite sides and 100 percent increment was observed under the confining pressure

of about 10 N/mm2. As in the earlier observation, additional application of lateral

pressure of up to 28 N/mm2 did not increase the failure load considerably.

By conducting extensive research on effects of confinement in concrete, Hungspreug

(1981) concluded that increasing cover and transverse reinforcement increase the

confinement effects on the bar, and hence can be treated as factors which increase the

bond strength. In addition, it was observed that an increase of bond strength is a direct

result of increase in concrete with concrete tensile strength or in other words, the bond


2-38

strength rises proportionately with the increase of the square root of the compressive

strength. Similar to the observations of Malvar (1992), the bond stress linearly

increased up to a confinement stress of 2.8 MPa, however severe radial cracking

prevented the linear increment for higher confinement stresses.

2.7.2 Dynamic bond slip analysis Yan (1992) studied the bond slip under impact loading conditions. Tests were

conducted in the dynamic as well as in the static range. A drop weight impact hammer

was used to apply the impact load and the effects of different surface roughness,

compressive strength, amount of fibre content and two different fibres (polypropylene

and steel) were investigated. It was found that for smooth rebars, the bond slip

relationship was linear for both, static and dynamic loading conditions. For deformed

bars, the effects of the surface roughness were found to play an important role in bond

slip resistance. The bond-slip relationship under dynamic loading changes with time

and location along the reinforcing bar and is greatly influenced by the above

mentioned parameters. Specifically, the surface roughness affects the stress

distribution in the concrete, the slip at the interface between rebar and concrete and the

crack development.

Series of experimental tests were carried by Weathersby (2003) to investigate the

(a) chemical adhesion between smooth steel bars and concrete

(b) bond resistance of smooth bars and deformed bars and

(c) influence of concrete confinements and bar diameter.

Three modes of failure were identified from the experiment. The failure of smooth

bars mainly occurred due to pullout and the failure mode was independent of the

loading rate. The resistance to pullout mainly provided by static and dynamic frictions

with chemical adhesion. Strength of the chemical adhesion and the static friction

increased with increasing loading rates. For example, the combined static friction and

the chemical adhesion were 6.62 MPa for the quasi-static loading and 22.1 MPa for

the impact loading.

In deformed bars, failure occurred due to radial cracking regardless of the loading rates.

Compared to the smooth bars (10φ and 8φ), 70%-77% increment was observed due to


2-39

the deformations and this increment was independent of the loading rate. The ultimate

load at failure increased as the loading rate increased. It was observed that the ultimate

load at failure was 70% to 100% higher than that under quasi-static loading conditions.

Confinement of concrete increases the bond stress between the reinforcement and

concrete over all loading rates (Malvar 1992). As a result, under impact loading

conditions the failure mode between concrete and steel shifted from concrete cracking

to steel yielding. However, with increasing loading rate the increment of bond

resistance dropped significantly. It was observed that, as long as the failure mode

remained constant, the bond stress of impacted specimens was nearly twice the value

of quasi-statically loaded specimens.

2.8 Factors affecting ductility of concrete columns

2.8.1 The effects of confinement on enhancement of the ductility and strength In general, both ductility and strength of columns increase with the confinement

provided by transverse steel. In the absence of sufficient transverse steel, the

behaviour of the column is governed by the strength of unconfined concrete, which is

caused by inactivation of transverse reinforcement in providing required confinement.

Nevertheless, through experimental results, it has been proved that columns with a

larger volumetric ratio of confinement steel have larger deformability and flexural

strength capacity (Saatciglu and Razvi 1992; Mandar et al. 1984). Therefore, under the

impact loads, the effects of confinement play an important role in terms of ductility

and energy absorption.

One function of confinement steel is to provide passive confining pressure to the

concrete core by lateral expansion. Passive confining pressure depends on many

factors such as diameter, dimension of the core, and spacing of the confinement

reinforcement. The diameter of confinement steel affects the total lateral force acting

on the concrete core and the spacing affects the distribution of pressure in the concrete

core. The confinement pressure acting on the concrete core is highest at the

contra-flexure point and gradually reduces in a parabolic manner (Mandar et al. 1984).

Therefore, close distribution of the transverse reinforcement will increase uniformity

of the lateral pressure. Previous experimental tests showed that adverse effects of the


2-40

large transverse spacing can be overcome by using smaller diameter steel bars placed

at closer distance. The distance should be no longer than the distance at which the

premature buckling starts in the longitudinal reinforcements (Saatcioglu and Razvi

1998). However, the relationship between spacing and the diameter is not linear

(Johnny 2003). Because of the parabolic distribution of the confinement pressure in

horizontal and vertical directions, the effectiveness of the confinement greatly reduces

as the transverse steel spacing increases in the vertical direction. Therefore, columns

with larger link spacing fail due to the lack of confinement effects, irrespective of the

transverse steel content. Hence, the British Code, BS 8110 (1985) limits the maximum

spacing of the transverse steel to 12 times the smallest longitudinal compression bar

diameter.

Confinement effect depends on the yield strength of transverse reinforcement. It is

expected that steel of higher strength provides greater confining pressure in the

concrete core. However, lateral expansion of the concrete core will not be greater and

the tensile capacity of the steel may not be fully developed under working conditions

(Johnny 2003). It is shown that for ordinary high yield deformed bars of yield strength,

fy = 460MPa, strain hardening of the transverse steel is unlikely to occur under

serviceability conditions. ACI 318(2002) and NZS 3101 (1995) also specify an upper

limit for the yield strength of transverse reinforcements. ACI limits the allowable

maximum yield stress to 420MPa and NZS to 800MPa. However, highly localised

stresses generated under impact loading conditions may yield the transverse

reinforcement (Memari et al. 2005). Hence, the above restrictions underestimate the

possible maximum strain that can be developed in transverse reinforcement under

impact loading conditions. In other words, hoops having higher yield strength should

be used for columns susceptible to impact loads. Saatcioglu and Razvi (1998)

introduced an important relationship between volumetric ratio Vr and yield strength

'syf of confined reinforcement so that a compromise between each parameter is

possible.

ie. 'sysr fV ρ= Eq. 2.27

This equation implies that a reduction in yield strength can be compensated by

increasing the amount of transverse reinforcement.


2-41

The confinement effect of concrete also depends on the configuration of transverse

steel and distribution of the longitudinal steel. Confinement provided by longitudinal

reinforcement is particularly effective if the reinforcement consist of larger diameter

bars, which are less prone to inelastic buckling. Nevertheless, it was found that

irrespective of the distribution of longitudinal reinforcement, columns with closer

links have better confinement effects and consequently an increased flexural strength

and ductility (Cusson and Paultre 1994). Therefore, in providing the required

confining pressure, the distribution of both transverse and longitudinal reinforcement

is equally important.

Figure 2.9: Distribution of confining pressure produced by various shapes of

transverse steel (Razvi and Saatcioglu 1999) The confining pressure provided by various kinds of links is shown in Figure 2.9.

When the lateral force produced by the links is well distributed across the perimeter, as

in the case of circular columns, the efficiency of the confinement effects is improved.

However, for square columns the confinement pressure reduces from the corner to the

mid point of each side of the links and efficiency of the confinement decreases. As far


2-42

as the rectangular columns are concerned, the pressure distribution is not uniform in

each direction and hence the efficiency greatly decreases.

The following equation quantifies the relationship between confinement effects due to

transverse steel spacing and distance between laterally tied longitudinal steel (Razvi

and Saatcioglu 1999);

115.02 ≤

=l

cc

s

b

s

bk , Eq. 2.28

where bc is the breadth of the core concrete for square column, s is the spacing of the

transverse steel along the height of the column and sl is the spacing between laterally

confined longitudinal bars. According to Razvi and Saacioglu (1999), parameter k2 is

proportional to the effectiveness of the confinement provided by both, transverse and

longitudinal steel. It was observed that the effectiveness of the confinement can be

improved by reducing spacing ‘s’ between the transverse steel and the distance

between the longitudinal steel ‘sl’. However, the addition of cross ties at a fixed

volumetric ratio may or may not improve the confinement effects. Because, while

improving the confinement effects in the lateral direction, the cross ties will on the

other hand increase the transverse steel spacing.

2.8.2 Effects of concrete cover The amount of transverse steel required for confinement increases when the concrete

cover to thickness ratio c/Do for a circular section or c/h for a rectilinear section,

increases (where c is the concrete cover thickness, h is the smaller dimension of the

rectangular section and Do is the diameter of circular section). This is because when

the concrete cover is high, the column loses significant flexural strength as the cover

concrete spalls off quickly under comparatively low strain conditions. The resultant

loss of the flexural strength can be recovered by providing adequate transverse

reinforcements to confine the core concrete (Johnny 2003). For convenience, New

Zealand Standard NZS 3101 (1995) expresses the concrete cove thickness ratio c/Do or

c/h by Ag/Ac, where Ag and Ac are gross concrete area of the section and area of the

concrete core respectively. According to NZS 3101 (1995) specifications the ratio

Ag/Ac lower than 1.2 should be avoided.


2-43

2.8.3 Compressive axial load level Flexural ductility reduces significantly with the increase in compressive axial load on

the column (Paultre et al. 2001). As the axial load level increases, the concrete

becomes subjected to higher compressive stress levels and the moment capacity of the

column is then dependent mainly on the compressive strength of the concrete. Since

concrete is brittle in nature, flexural strength reduces rapidly after reaching the

maximum moment capacity of the column. Moreover, depth to the neutral axis

increases as the axial load level increases and the extreme concrete fibre is subjected to

higher compressive strains. Under these conditions the concrete will reach its ultimate

strain sooner and as a result, the concrete cover will spall off rather quickly, causing a

decline in the flexural capacity of the section. If sufficient transverse reinforcement is

provided, the reduction in flexural capacity can be compensated by increasing the

capacity of the concrete core such that the concrete core could dilate properly under

large compressive axial loads (Johnny 2003).

On the contrary, in the presence of adequate amount of transverse steel in RC columns,

the concrete core dilates and may induce transverse radial confining pressure in the

concrete core. Under this circumstance, the ultimate strain of the concrete core is

considerably larger than that of the unconfined concrete cover. The actual stress-strain

behaviour of the concrete core, therefore, could vary from that of the outer cover of the

column (Anselm 2005). This may cause considerable deviation of the results if a

common stress-strain diagram is assumed for a non-linear impact analysis of the

column.

2.8.4 Combined effects of axial load and flexure Extensive experimental research has been conducted to investigate the behaviour of

columns under the combined action of axial load and flexure. The investigation mostly

covered flexural strength and flexural ductility performance under the combined

action of axial load and flexure. However, the flexural behaviour of reinforced

concrete columns also depends upon many other factors such as column slenderness

(Kim and Yong 1995), load eccentricity (LIoyd and Rangan 1996), boundary

conditions at the ends (Majewski 2007), area and the shape of the cross section of


2-44

concrete, area and spacing of the vertical and horizontal reinforcement (Razvi and

Saatcioglu 1999) and the reinforcement ratio. On the other hand, improved flexural

behaviour of concrete columns will help minimise the adverse effects of lateral impact

loads, by converting the failure mode from brittle to ductile. Axial load level could be

a critical factor that governs the above phenomenon. Research, which covers the

flexural ductile behaviour of concrete columns and the factors affecting the ductile

behaviour is summarised in the following paragraphs.

By conducting an analytical study of flexural strength and ductility of RC columns,

having various arrangements and quantities of transverse reinforcements, Watson et al.

(1994) concluded that more transverse steel is required for confining the core of square

and rectangular columns than that of circular columns. This is due to variation of the

lateral pressure distribution inside the concrete core. On the other hand, the quantity of

transverse reinforcement required for confinement to meet any particular curvature

ductility factor demand, increased with increasing axial load level, increasing concrete

strength, decreasing longitudinal reinforcement ratio and increasing concrete cover

thickness. Additionally, large flexural strength enhancement was detected for columns

subjected to medium or high axial loads.

Sheikh and Khoury (1997) proposed a performance based approach for designing

confining steel in tied columns. By comparing the provisions of the ACI code (1995)

with their findings, it was concluded that the behaviour of column design according to

the ACI code may vary from unacceptably brittle to very ductile. To overcome the

problem, a modification was introduced to the ACI (1995) equation and when

compared with the experimental results, the proposed equation showed excellent

agreement. The final equation is as follows;

cha

h AP

PA ,

5

29131 φµ

α

+= , Eq. 2.29

where Ah,c is the ACI code provision of transverse reinforcement content, α is a factor

dependent on the tie configuration, µφ is the curvature ductility factor, P is the axial

load applied and Pa is the axial capacity of the section. It is interesting to note that this

equation accounts for the axial load effects on the column. The curvature ductility

factor is included in the equation, because the amount of transverse reinforcement


2-45

according to the current provision in ACI code, is too low for high axial load levels and

too high for a low axial load level( )4.0' <cg fAP .

2.9 Shear capacity calculations and effects of axial load

Columns subjected to impact loading may fail in either shear or flexure. Shear capacity

of structural columns under dynamic impact loads are therefore important to determine

the threshold of a shear failure. Axial load in particular, is one of the deterministic

factors that can affect the shear capacity of impacted columns. On the other hand

energy dissipation characteristics (Li et al. 1991), sequence of the load application

(Saadeghveziri 1997) and confinement effects (AIJ 1994) are also play an important

role in the shear capacity determination. Following are the information that is found

from the literature which can be used to determine the behaviour of shear in columns

under impact loading conditions.

According to the Louw et al. (1992), when a column is subjected to a hard impact, the

maximum dynamic shear force generated just after the impact, and the ratio of

maximum dynamic shear to ultimate static shear strength of the column, vary from

0.87 to 1.67. The variation of the axial load in the column due to the impact load is

quite similar to that of the dynamic shear force. As a result, the shear capacity of the

column can change and hence the dynamic shear capacity can differ from the static

shear capacity of the column. In contrast with a hard impact, the peak load occurs

much later under soft impact conditions. In addition, it is more likely to affect the

flexural shear resistance of the column rather than the pure shear and inertial stiffness.

However, the actual contribution of the axial load in resisting the dynamic shear force

is uncertain.

Kreger and Linbeck (1986) reported a test result of three double curvature specimens

under various lateral and axial load variations. Two specimens were subjected to axial

loads proportional to the lateral loads. The other specimen was tested using uncoupled

axial and lateral loads. The test results reveal that strength of specimens increased with

increasing axial load. In addition, the energy dissipation characteristics of the columns

depended largely on the axial load history.


2-46

Based on an analytical investigation of columns with proportionally and

non-proportionally varying axial loads, Saadeghveziri (1997) suggested that the

energy dissipation capacity of the columns could be reduced significantly under

uncoupled variation in axial and lateral loads. Li et al. (1991) came to similar

conclusions by conducting extensive experiments on seventeen cantilever columns,

under constant axial loads and proportionally and non-proportionally varying axial

loads. Moreover, in non-ductile columns, the proportionally varying axial load pattern

resulted in significant shear strength degradation. The result also showed that the

variation in magnitude of the axial load had significant effects on the stiffness, strength

and deformation capacity of the column.

Different formulations and parameters have been proposed over the past few decades

to calculate the shear strength of the concrete columns. The shear strength is calculated

as a summation of the strength contributions from the concrete and transverse steel.

However, representation of the effects of various parameters such as axial load,

displacement ductility and aspect ratio are vary depending on the formulation or some

times are not included.

The equation given in ACI 318-02 considered the effects of axial load on the shear

strength enhancement. The shear strength Vn, is calculated as the summation of

contribution of the concrete Vc and the transverse steel Vs. The contribution of the

concrete to the shear strength of the members subjected to shear and axial compression

is given as,

bdfA

PV c

gc '

200012

+= , Eq. 2.30

where P is the axial load, Ag is the gross cross sectional area, f ’ c is the compressive

strength of concrete and b and d are the web width and effective depth of the section,

respectively. Here all the units are in lb, in and psi.

The contribution of the transverse reinforcement was calculated as,

s

dfAV ysw

s = , Eq. 2.31

where Asw is the transverse reinforcement area within a spacing, s and fy is the yield

strength of the transverse reinforcement. Special provisions of the ACI 318-02 further


2-47

state that at column ends or in the possible plastic-hinge regions, the concrete

contribution, Vc should be taken as equal to zero, if the factored axial compressive

force including earthquake effects is less than 20'gc Af and if the earth quake induced

shear force is large.

The ASCE-ACI committee report (1977), provides a broad insight into the shear

transfer mechanism. The report describes the important mechanisms as: (a) shear

transferred by the uncracked concrete; (b) interface shear transfer in the cracked

concrete, ie. aggregate interlock; (c) dowel shear carried by the longitudinal

reinforcement; (d) arch action in deep members and (e) shear transferred by the

transverse reinforcements. The most critical mechanisms identified were the shear

transfer by the transverse reinforcement and concrete. For members with short

span-to-depth ratio, the committee recommends the use of reduced shear stress vc.

Aschheim and Moehle (1992), used laboratory test data from cantilever column tests

to identify the shear strength characteristics. The data indicated that the column shear

strength is a function of displacement ductility demand, the quantity of transverse

reinforcement (confinement), and axial load. The proposed approach is similar to the

ACI 318-02 code equation, except when the equation is used to calculate the concrete

contribution Vc. The concrete contribution is defined as,

bdfbdfA

PV cc

gc

'' 5.3)2000

1(' ≤+= α . Eq. 2.32

For design and evaluation of rectangular hoop reinforced concrete columns,

δµρ

α yww f006.0'= , Eq. 2.33

where wρ is the transverse reinforcement ratio, bsAsww =ρ .

2.10 Energy absorption characteristics under impact loads

Structural columns subjected to dynamic loading conditions must have higher energy

absorption capacity, and hence, should be allowed certain plastic deformation to avoid

premature shear failures. In addition, rate of energy absorption is among the factors

that determine the mode of failure. In brief, the energy absorption capacity of a column,


2-48

under impact conditions may depends on the factors such as material characteristics,

relative masses, stiffness, area of contact, and frequency of loading.

Bischoff and Perry (1995) investigated the energy absorption capacity of concrete by

conducting a series of tests on air-dried plain concrete cylinders. The cylinders were

subjected to slow static loading rates of 10 micro strains per second, as well as much

higher rates of about 5 to 10 strains per second. Two grades of concrete with design

strength of 30 and 50 MPa were tested. This investigation showed that absorbed

energy is greater for specimens subjected to strain rates higher than 250 micro strains,

compared to specimens subjected to quasi-static rates. Therefore, the energy

absorption capacity of plain concrete seems to depend on the strain capacity of

concrete, which in turn is governed by the failure mode. In addition, according to the

experimental results, the weaker mix absorbed more energy than its counter part. It

was, therefore, concluded that both strength and strain rate capacity play an important

role in determining energy absorption capacity.

According to Mindess et al. (1986), energy absorption capacity can be improved by

adding extra fibre volume to concrete. Also, maximum flexural load and fracture

energy increased significantly with the increase in fibre volume. The enhancement of

the moment rotation characteristics of the flexural compression zone and the

increment of local deformation characteristics due to the fibre reinforcement could be

the possible reasons for those observations.

In general, fibre reinforcement will increase impact resistance or dynamic toughness

of plain concrete, owing to the significantly increased maximum beam deflection. This

effect was predominant only after the fibre content was increased up to 0.75% (Hughes

and Al-Dafiry 1995). The most significant effect of the additional fibre content is the

delay in failure which means, for higher fibre content, the duration of the impact of

5ms compared to the duration of the impact event with lower fibre content of only 1ms

(Wang 1996). The average deflection also increases under these circumstances and

results in a lower rate of energy absorption, which reduces the possibility of a shear

failure of an element. However, improvements in the peak load and fracture energy

under impact loading may be considerably lower compared to the improvements


2-49

obtained under static loading conditions (Banthia et al. 1989). Also, the addition of

nominal shear reinforcement with fibre did not show any significant impact on energy

absorption (Wang 1996).

Fracturing effects can also contribute to the energy absorption characteristics under the

impact loads. Concrete can absorb more energy under impact loading conditions

(Remennikov and Kaewunruen 2006). The energy consumption for plasticity and

other non-liner deformations under the unstable conditions can be one possible reason.

In addition, boundary conditions may also affect the energy absorption process (Xu,

2001). Due to the influence of the boundary conditions, the resultant energy

consumption closer to the boundary will vary from that in the region far away from the

boundary. On the other hand, tensile strength of concrete gained the highest

consideration among the parameters that influence energy dissipation under the impact

loading conditions (Weerheijm and Doormaal, 2007).

Conditions at the supports and contact interface may also have considerable effect on

energy absorption under laboratory conditions. Since the displacement of the

supporting devices in experimental set-ups was found to be very small, energy

absorption at the supports can be neglected compared to that of the contact interface,

especially under hard impact conditions, where significant deformations take place

(Hughes and Al-Dafiry 1995).

2.11 Design practices and provisions of RF in critical sections

In general, existing reinforced concrete columns contain additional amount of

longitudinal reinforcement, which provides excessive strength against simultaneous

compressive axial load and flexure. However, provisions of shear reinforcement in

structural columns were just sufficient to withstand the ultimate shear force in the

columns (Ho and Pam 2003). Additionally, codes such as BS 8110 contain few

provisions for the detailing of transverse reinforcement. As a result, using current

design methods, post elastic behaviour and failure mechanisms cannot be anticipated.

Consequently, the resulting failure modes under the impact could be brittle and

sudden.


2-50

Most of the current design provisions include the inelastic behaviour of the structure

under seismic attacks and overloading. Existing design philosophy is based mainly on

the energy dissipation through extensive inelastic deformation within the potential

plastic region, especially under earthquake conditions. However, the nature of the

damage caused by impact loads is quite different from the damage caused by

earthquake loadings. In either case, proper detailing of longitudinal and transverse

reinforcement within the potential plastic hinge region is needed to avert the following

undesirable modes of failures in structural columns (Ho and Pam 2003): (a) shear

failure (b) bond and anchorage failure at the member joints (c) buckling of

longitudinal steel and (d) premature failure of concrete core due to inadequate

transverse steel. All of the above are classified as brittle and sudden failures, which

must be avoided.

In the absence of more specific information on plastic hinge formation in columns

under lateral impact loads, especially for low velocity hard impact conditions, the

experimental results in the following studies can be used to identify the factors

affecting the formation of plastic hinges. Mendis (2001) conducted a series of tests on

simply supported beams with a point load at mid span, which he subjected to various

levels of axial load to determine the plastic hinge length. It was concluded that plastic

hinge length remained constant at 0.40d or 0.33h, where d and h are the effective and

overall depth of the beam respectively. According to the observations, the plastic hinge

length is independent from the compressive axial load level on columns.

Based on the tests of four full size confined RC columns with a 550mm square cross

section, Park et al. (1982) also showed that the equivalent plastic hinge length was

insensitive to the compressive axial load level and had an average value of 0.5d, where

d is the effective depth of the column. The tested columns contained transverse steel

ratio of up to 4%. However, the effect of the transverse steel ratio on the length of

plastic hinge was not demonstrated.

Contrary to the above observations, Clause 8.5.4.1 of NZS 3101 (1995), specifies that

the potential plastic hinge length is influenced considerably by the applied

compressive axial load level. For an RC member subjected to an axial load level

(Refer Eq. 2.34) of 0.25 or smaller, the potential plastic hinge length is equal to 1.0h or


2-51

a region where the moment exceed 0.8 of the maximum moment, whichever is larger:

'' cg fAP φ Eq. 2.34

where P is the compressive axial load, φ’ is the strength reduction factor, Ag is the

gross cross sectional area and 'cf is the concrete compressive cylinder strength. The

plastic hinge length is increased to 2.0h or over a region, where the moment exceed 0.7

of the maximum moment, whichever is larger, for an axial load level larger than 0.25

but smaller than or equal to 0.5. It is further increased to 3.0h or over a region, where

the moment exceed 0.6 of the maximum moment, whichever is larger, for axial load

level larger than 0.5 but smaller than or equal to 0.7.

Based on the test results of eight large scale high strength reinforced concrete (HSRC)

columns with a square cross section of 305x305mm, Paultre et al. (2001) showed that

the potential plastic hinge zone could range from 1.0h to 2.0h, depending on the

volumetric ratio of transverse steel and the compressive axial load level. The plastic

hinge length was found to increase with an increase in compressive axial load level

and concrete compressive strength, but with an increase in the transverse steel

volumetric ratio, it was reduced. Ho and Pam (2003) also recommend that the potential

plastic hinge region of HSRC columns can be taken as h, which is the overall depth of

the column cross section. However, Ho and Pam (2003) observed that the plastic hinge

length of columns was insensitive to the spacing of the transverse steel, provided it

was less than 0.75d.

Bayrak and Sheikh (2001) presented an analytical procedure to predict the behaviour

of plastic hinges in RC columns. The analysis incorporated complex behaviours, such

as softening of longitudinal bars due to inelastic buckling and reinforcement

cage–concrete core interaction. According to their observations, the concrete

core-reinforcing cage interaction, which caused outward deflection in longitudinal

bars, did not only reduce the ductility of longitudinal bars under compression, but it

also reduced the maximum stress the bars were able to achieve. They also proposed

that for high curvature ductility demand, ratio of tie spacing to longitudinal bar

diameter should be kept under 6, which is the same as that proposed by Mander et al.

(1988). However, for moderate curvature ductility demand, the ratio should be kept

below 8.


2-52

While there is no general agreement about the effects of the volumetric ratio, axial load

and spacing of transverse steel, it is quite clear that the height, where the plastic hinge

is formed is increased with the compressive strength of concrete. That is, under impact

loads, where strain rate effects are predominant, and hence the compressive strength is

increased, significant changes of the height, where the plastic hinge is formed, can be

expected.

In addition, based on the fact that inelastic buckling of the longitudinal reinforcement

and yielding of confinement steel is excessive within the plastic hinge region, it is

proposed that the end hooks of transverse reinforcement within that region are 45o,

while those outside the plastic hinge region be 90o (Ho and Pam 2003). Moreover, the

average length of the plastic hinge region affected by the ‘Stub Effect’ was observed to

be around 50mm (Paultre et al. 2001; Ho and Pam 2003).

2.11.1 Influence of the various parameters on confinement 2.11.1.1 Effect of the cover concrete and volumetric ratio of steel on confinement

It was found that when concrete strength '

cf of the column increased, the column

strength achieved a proportional increase in capacity. The ultimate strength was gained

at larger axial deformations than the yield point. Concrete cover spall off just before

the peak strength was achieved without achieving its full compressive strength and in

general, it was approximately at 0.75'cf (Cusson et al. 1994). Some of the testing

procedures have taken in to account the effect of the longitudinal steel while many of

them did not consider that effect. The limited experiments with longitudinal steel

revealed that when concrete strength and the yield strength of longitudinal steel were

constant, the strength of the columns increased with an increase of spiral

reinforcement. The effects of spacing were more pronounced for steel ratios from

1.1% to 1.7% and s/Dc appeared to be around 1.2% and 0.24% respectively for stable

unloading behaviour (Toklucu et al. 1992). In addition, columns that contained

volumetric ratio of steel spiral around 3.2% were capacitated to replace the loss of load

carrying capacity due to cover spalling. Moreover, typical columns with more than

3.2% volumetric ratio of spiral developed a second maximum load after the first peak

but never exceeded the first maximum load at spalling (Richart 1946). Thus, it can be


2-53

concluded that confinement steel increases both the strength and deformation capacity

of concrete in compression while increasing the stresses in spirals at peak confined

concrete strength (Tomosawa et al. 1990). Even though lateral reinforcement has very

significant effects on the confined core, increasing lateral steel results in less

proportional increase in strength and ductility (Sheikh and Uzumeri, 1982).

It was also revealed that the volumetric ratio of lateral steel ranging from 0.55% to

1.64% is not sufficient to substitute for the loss of load-carrying capacity resulting

from the failure of the protective cover. On the other hand experimental tests on

150x150mm specimens made out of Grade 59 to 68MPa concrete revealed that the

stress-strain curves were not significantly influenced by the area of the longitudinal

steel or the spacing of the lateral ties including the one with 50mm spacing (Hwee et al.

1990). However, strength due to the confinement effects was enhanced with further

reduction of the lateral spacing from 50 to 15mm in 150x300mm specimens made out

of Grade 50 to 120 concrete (Tomosawa et al. 1990). Therefore the above strength

category can be treated under HSC category where the behaviour is significantly

different from the LSC. Thus, the limit of 76mm for the spiral spacing in large

diameter column was considered to be too restrictive (Toklucu et al. 1992).

2.11.1.2 Changes in the stress-strain curve due to confinement

According to the Yong et al. (1988), the stress-strain curves exhibited a relatively

linear ascending branch below the maximum axial load for the tested specimens of

plane concrete, and confined concrete with or without cover. Also the relative

enhancement of the confined strength and the corresponding strain were decreased as

the grade of concrete is increased. Furthermore, the slope of the descending branch of

the stress-strain curve was deteriorated more rapidly with higher concrete cylinder

strengths. Therefore the effectiveness of the confinement will decrease as the concrete

grade increase (Ahmad and shah, 1982). Apart from that, decreasing the spiral spacing

by keeping the volumetric ratio as a constant will not improve the slope of the falling

branch of the stress-strain curve. It is also concluded that the number of longitudinal

bars had little influence on stress-strain behaviour (Mander et al. 1988). In general, as

the volumetric ratio of confining reinforcement increased, the peak stress increased,


2-54

the slope of the falling branch decreased, and the longitudinal strain at which spiral

fracture occurred increased (Mander et al. 1988). But not proportionally (Yong et al.

1988).

2.11.1.3 Effects of the axial load and grade of concrete on confinement

In particular, lower grade of concrete exhibited large plastic deformation capacity

without losing its load carrying capacity significantly while the slope of the

descending branch is reduced with increased confinement. In contrast, the steepness of

the ascending branch of the stress-strain curve was increased with higher grades of

concrete. It is also observed that the confining effects of spirals were small for loads up

to about 30% of the unconfined column strength. Above this load, rapid increment of

the spiral stress and confinement pressure can be seen. Once the load level had reached

unconfined compressive strength there was rapid disintegration of the concrete

microstructure and hence greatly increases the confinement stress. Consequently,

spirals yield before the maximum capacity of the column was reached. With the

enhancement of the concrete strength the disintegration of the concrete at the

maximum compressive strength was diminished and hence the rapid increment of

spiral stresses was diminished. As a result, none of the spirals in HSC (Grade 55 to 83

MPa) column reached their yield strength at the maximum load (Martinez et al. 1984).

Moreover, the failure of the protective cover of HSC columns was sudden and brittle

compared with LSC columns. However, for HSC the second maximum load found to

be greater than first contrary to what would be expected (Martinez et al. 1984).

2.11.1.4 Effects of yield strength of hoops on confinement As far as the yield strength of the spiral is concerned, it did not influence the

compressive strength enhancement of concrete. Similar steel stresses were observed at

peak load regardless the yield strength of the spirals for the specimens made out of

same grade of concrete. Thus the ductility will increase with the enhancement of the

yield strength. In fact, confining pressure exerted by spiral depends on the potential

lateral expansion of the corresponding plane concrete. Thus, reluctant to the lateral

expansion of the higher grade of concrete will excrete less lateral strain on spirals for a

given axial strain. However despite the grade of the concrete the stress on spirals


2-55

rapidly increases once the peak stress is achieved (Ahmad and shah, 1982). It was also

found that lateral steel reaches its yield strength at the maximum load even when very

high strength steel (fy = 915MPa) was used (Polat 1992).

2.11.2 Theoretical stress strain curves for confined concrete by transverse RF 2.11.2.1 Stress-strain model for confined concrete

Numerous empirical models have been proposed by various researchers to predict the

non-linear behaviour of confined concrete columns under concentric loading.

However numerical modelling of the nonlinear response of the confined concrete has

not been addressed substantially in the literature. In the present study a material model

compatible with nonlinear finite element analysis of three dimensional concrete

models was used to investigate the confinement effects.

The stress-strain model developed by Mander et al. (1988) simulates the confinement

effects in this process. It was assumed that the passive lateral confining pressure

exerted by the transverse reinforcement leads to a tri-axial state of stress in the core

concrete and thus enhances the compressive strength compared to the unconfined

concrete. Eventually the equal and opposite forces acting on the lateral reinforcement

may rupture the hoops at one stage by bringing the useful ultimate longitudinal

compression strain to a residual level (see Fig. 2.10). Longitudinal compression strain

of confined concrete within the range of 0.02 to 0.08 should be maintained to satisfy

this requirement (Watson et al. 1994).

Figure 2.10: Stress-strain model for confined concrete proposed by Mander et al. (1984)

The proposed model is applicable for both circular and rectangular shaped transverse

reinforcement. The stress strain model illustrated in Figure 2.10 is based on an


2-56

equation suggested by Popovics (1973). The 'ccf is defined as compressive strength of

confined concrete and 'ccf will depend on links spacing, area of effectively confined

concrete core, diameter of the hoops and yield strength. The lateral confining pressure

can be found by considering the stability of the half body confined by the transverse

hoops as shown in Figure 2.11. In fact, the model assumes an arching action to occur in

the form of a second degree parabola with an initial tangent slope of 45o. Hence, the

effectively confined area is calculated based on the confined concrete core midway

between the levels of transverse reinforcement. In the model, the resultant

confinement stress due to the various components such as yield strength, diameter and

spacing of the hoops are expressed in terms of equivalent uniform confinement over

the core concrete. Additionally, two sets of equations were developed for rectangular

and circular columns separately by taking into account the variation of lateral

confining stress across the sections.

However the cross ties can excrete either equal or unequal confining stress along each

of the transverse axis depending on the topology of the section. As far as the circular

columns are concerned the stress distribution is uniform across the section and hence a

single equation can be used for the stress-strain equation despite the axis of bending.

c

yhb

l sb

fAf

4= Eq. 2.35

c

yhb

l sb

fAf

2= Eq. 2.36

Figure 2.11: Confining stress provided by the transverse reinforcements The effective lateral confining stress in each direction exerted on the core at yield

strength is given by:

lel fkf ='

Eq. 2.37

Where fl is the confining stress calculated as in the Eq. 2.36 and ke is the confinement


2-57

effectiveness coefficient that takes into account the arching action. s is the links

spacing and fyh is yield strength of transverse reinforcement. The coefficient ke for

circular hoops is defined as the ratio of the area of effectively confined core concrete to

the area of concrete within the centrelines of the peripheral hoop or spiral and

mathematically expressed as,

cc

se

d

s

kρ−

−

=1

21

2

Eq. 2.38

where cc

ρ ratio of area of longitudinal reinforcement to area of core of section and ds

is the distance of spiral between bar centres. s is the clear vertical spacing between

spiral or hoops. In addition, for a section with equal effective lateral confining stress in

each direction, the ratio of the compressive strength of the confined concrete 'ccf to the

comprehensive strength of unconfined concrete 'cof is given by,

254.1''

0.2''

94.71254.2'' −−+=

co

l

co

l

co

cc

f

f

f

f

f

f

Eq. 2.39

Over the last few decades many experiment have been carried out on confined

concrete columns under both concentric and eccentric loading conditions. In addition,

numerous models exist to predict stress strain behaviour of normal and high strength

confined concrete. However various limitations still persist when it comes to the

column geometry and axial loading conditions.

2.12 Effects of impact induced torsion in eccentrically loaded columns

The torsion is more likely in skewed bridges and bridges with outrigger bents though

the structural columns with irregular configurations may also susceptible to in-built

torsional moments. When the impact occurs in a direction perpendicular to the plane of

bending, impact induced torsional moments are generated. Most of the investigations

were focused only on the effects of bending moments and shear forces by neglecting

the effects of torsion. In the absence of any detail on the effects of impact induced

torsion, some insight may be provided by the tests conducted under seismic loading

conditions. For instance, Tirasit and Kawashima (2008) reported that the torsion


2-58

induced damage tends to occur above the flexural plastic hinge region and

consequently change the formation of plastic hinge zones. According to Otsuka et al.

(2005) pitch of the lateral ties can be used to control the torsion induced damage. Thus,

spiral reinforcement which might adequate for flexural design may not be adequate in

the presence of torsional moments (Prakash et al, 2009). Further, they observed that

the deformation characteristics and failure modes may also affected by the presence of

torsion. The reduction of confinement effects of core concrete followed by the torsion

induced spalling further support these observations. Consequently, columns failed

without achieving the ultimate shear capacity (Belarbi et al. 2008).

In general, three modes of failures can be identified under the combined effects of

shear flexure and torsion depending on the initiation of failure. In under reinforced

members the reinforcement will fail before concrete crushes while in partially over

reinforced columns the failure will occur either by yielding transverse or longitudinal

reinforcement. If concrete crushed before any steel yield they can be categorised as

completely over reinforced members (Prakash et al. 2009). In addition, open hoops

may unlock and become non-functional under torsional loads by causing significant

spalling of cover concrete while reducing the confinement effects particularly under

cyclic loading (Belarbi et al. 2008). Therefore close hoops or spirals may be more

suitable for columns subjected to torsional moments. In addition, the torsional strength

mainly depends on the amount of longitudinal and transverse reinforcement, the

sectional dimensions and the concrete strength. For instance, in circular columns the

torsional strength was reached first and then the flexural strength while in square

columns the two strengths were achieved simultaneously.

2.13 Impact reconstruction

Impact of a rigid object with a column is obviously a hypothetical case, which

represents more extreme situations of a vehicle impact. For example, the rigid body

induced deflection is highly concentrated to the point of impact and the bending

moment varies even after the formation of the plastic hinges (Tsang et al. 2005).

Therefore, associated shear force can be much higher than the actual impact and hence

intends to model more extreme condition, usually the upper bound of possible impact


2-59

damage. On the other hand, simulation of an actual impact using a finite element

model of a vehicle would be a tedious task and is beyond the scope of this project.

Alternatively, impact induced vehicle deformation can be used to quantify the energy

absorbed during a collision. Many impact phase reconstructions assume a linear

relationship between an absorbed energy and the residual deformation (Campbell

1974; Varat et al. 1994; Neptune 1999). Based on this assumption, the stiffness of the

vehicle frontal impact can be represented by constant spring stiffness. This Campbell

Model (1974) was further improved by Parsad (1990) with the aid of repeated barrier

impact testings. One assumption continuing through that reformulation is that a

constant liner spring rate over the entire depth is applicable. However, available crash

data indicates that vehicle frontal stiffness cannot be precisely modelled through the

use of single linear springs for all vehicles (Varat et al. 1994). Hence it is prudent to be

aware of the potential problems with this method.

Varat et al. (1994) proposed the following method to categorise the existing multiple

barrier crash test data, covering vehicle velocity of up to 50mph. When analysing the

test data, energy of the impact should be used to account for the weight differences

between different test vehicles. That is, speed of the vehicles alone may not be

sufficient to differentiate the impact of vehicles belonging to different weight

categories. On the other hand, rebound velocities were not available for all the test data

and consequently, the mathematical formulation was developed based on the energy

absorbed by the crushed vehicle. The following equation 2.40 was derived by

assuming linear dissipative spring. Factor called Energy of Approach Factor ( EAF)

was implemented to maintain a linear relationship.

xBw

Eo =2, Eq. 2.40

where x is the crushed length and w

kB = . k is the spring stiffness and w is the

crushed width. Factor w

Eo2 is called EAF and Eo represents the absorbed energy by

the crushed vehicle. The above equation describes a relationship with a zero y-

intercept. However, vehicles do require some initial onset energy before permanent


2-60

crush takes place. Adding a onset energy factor to Equation 2.40 results in the

following,

xBEAFEAF o += . Eq. 2.41

Existing records of the crashed test were then compared with the above model, using

least squares method to determine how well the set of data fits the particular model.

The liner and quadratic curve fits were considered at the initial stage and R2 values

were compared and analysed along with the percentage errors between the fits and the

actual test data points. The vehicles were classified as non-linear, if noticeable

improvement was achieved through the use of the second order fit.

Some of the vehicles exhibited consistently linear relationship between EAF and

residual crush of up to 35 mph tested speed. It was observed that the overall average

percent error for all vehicles was approximately 5.0% for speeds from 15 to 35 mph

(Varat et al. 1994). On the other hand, some of the vehicles exhibited non-linear

relationship within this velocity region. This is an indication of the structural softening

even for the low velocity levels. The second order fits yielded good correlation for this

category of vehicles and overall average percent error was limited to 1.0%. Similarly,

some of the vehicles exhibited linear correlation of up to 50 mph while others

exhibited non-liner response within the same velocity region.

2.13.1 Application to accident reconstructions When performing an automotive accident reconstruction, the re-constructionist may

not have access to impact test data to the required extent. In addition, he does not know

whether a vehicle under analysis has a linear or non liner trend considering the EAF

versus crush relationship. Therefore, the study of Varat et al. (1994) needs further

generalisation by assuming a bi-linear relationship between all the crash test data.

Varat et al. (1994) suggest that the bi-linear approximation takes the form of a straight

line for the test data below 30 mph and another strait line to approximate the data from

30 to 50 mph. These two lines share the common 30 mph data point. Then the

percentage error between the assigned linear relationships and the actual data points

was calculated. The resultant overall average error was approximately 5.5% at the

speeds between 15 and 35 mph. The average error was found to be less than 5% for the


2-61

velocity range from 30 to 50mph. However, the vehicles exhibiting liner relationship

up to 50mph cannot be satisfactorily modelled with the presented bi-linear

approximation used between 30 to 50 mph.

A similar procedure was adopted by Wagstrom et al. (2004) to simulate a frontal

collision of vehicles of different velocity categories. The vehicles were divided in to

three mass class and stiffness coefficients suggested by Summmers et al. (2001) were

assigned to find the deformation characteristics (Table 2.2). Wagstrom et al. (2004)

was further suggested that the cars can be further simplified to two main categories

based on the front stiffness.

Table 2.2: Mass and stiffness coefficient for impact reconstructions

Category Mass (kg) Front Stiffness (kN/m) Light 1200 1000

Medium 1600 1000 Heavy 2000 2000

Front stiffness values only represent the initial slope of the force deformation curve

and by assuming liner springs, the impact forces are highly overestimated for larger

measurement of deflections. In addition, it is important to note in the model, that there

is no differentiation between the resultant elastic or plastic deformations, even though

the term stiffness is usually associated with linear elastic deformation. In fact, after

reaching their maximum deformations, linear spring elements would eventually act as

a tension elements. This has to be prevented by setting the stiffness constant to zero, as

the velocity of the mass changes the sign. This means that structural restitution effects

are considered as negligible (Wagstrom et al. 2004). On the other hand, vehicles with

bi-linear characteristics can be modelled as a two linear springs in series, where the

second spring does not compress until the first spring bottoms out. In this case, the first

spring represents the engine compartment and the second spring represents the

occupant compartment (Neptune 1999).

2.14 Design guidelines

Irish Standards, I.S. EN 1991-1-7-2006 suggests some useful guidelines for the

assessment of impact and accidental loads on buildings. The guidelines considered

accidental impact forces applied from rail or road traffic, ship impact and impact due


2-62

to helicopter emergency landing on a roof of a building. The code is recommending to

use equivalent static force when the consequences are low to medium. On the other

hand, more advanced analysis is required for more serious consequences.

Special attention has been given to buildings used for parking, where vehicles are

permitted to enter the inside of the buildings and the buildings that are located adjacent

to the road or railway traffic. These buildings are highly susceptible to impact loads

and the code recommends the use of dynamic analysis or equivalent static force

method to calculate the effects on a structure. The equivalent static force is used for the

verification of static equilibrium of the structure and for the determination of the

deformations of the impacted structure. The code also considers the effects of

variables, such as impact velocity, angle of impact, mass distribution, deformation

behaviour and the damping characteristics of both, the impacting object and the

structure. In addition, the code recommends the use of upper and lower characteristic

values for the material properties of the impacting body and for the structure,

respectively.

Horizontal equivalent forces applied on column or wall in a structure are tabulated in

the code for the type of road and vehicle. The maximum force 1000 kN is used to

account for a truck impact and 500 kN for a car impact in a parking garage. No

horizontal force needs to be considered on overhead elements, unless the clearance is

less than 6m. If the clearance is less than 6m, prescribed horizontal force can be used in

the design. Such forces can then be applied on the underside of the bridge over a traffic

lane.

According to the specifications given in AS 1170.1 (1989), columns in car parks

should be designed to withstand the additional horizontal impact load arising from the

movement of vehicles. The additional live load F can be calculated as follows,

l

mvF

∆=

2

2

Eq. 2.42

where, F = impact or breaking force in newtons,

m = gross mass of the vehicle in kilograms,

v = velocity of the vehicles, in meters per seconds,


2-63

∆l = deceleration length in meters.

∆l can be taken as the sum of the deflection of the vehicle and the barrier. In the

absence of reliable data, ∆ can be considered to be 0.1m. The recommended value of m

is,

a) for domestic car parking - 1500 kg ,

b) for general car parking - 2000 kg ,

c) maximum expected gross vehicle mass exceeding 2500 kg is rare.

Barriers facing ramps longer than 20.0m must be capable of withstanding an impact

force applied by a moving vehicle having a velocity of 10.0ms-1. In the absence of

reliable data, the deceleration length ∆l, can be taken as 0.15m. The code only

considered the collision with one vehicle at a time. The impact load was considered at

0.5m above the floor level for cars and 1.0m for trucks. The impact force was

considered to be distributed over a distance of 1.5m, along the barrier or full width of a

column.

2.14.1 Dynamic design for impact Complex interaction between two objects takes place under impact process. Due to

this complexity, simplified approximations are used to quantify the impact response.

Dynamic effects as well as the non-linear material behaviour must be taken in to

account in the impact analysis. EN 1991-1-7(2006) covers the dynamic aspects of the

design briefly based on the simplified or imperial models.

Based on the initial kinetic energy dissipation process, the impact is characterised as

either soft impact or hard impact. For soft impact, approximate dynamic analysis may

be performed by using the Equations from 2.43 to 2.45. By assuming that the

impacting body deforms linearly during the impact phase, the maximum dynamic

interaction can be calculated as,

kmvF r= , Eq. 2.43

where vr is the object velocity at impact, k is the equivalent elastic stiffness of the

object and m is the mass of the colliding object. If required, force due to impact may be

considered as a rectangular pulse with a non-zero rise time. The duration of the plus is

be given by,


2-64

kmtormvtF /=∆=∆ , Eq. 2.44

k = EAi/Li and m = ρALi, Eq. 2.45

where Li is the length, Ai is the cross sectional area, E is the Modulus of Elasticity and

ρ is the mass density of the equivalent impacting object.

Hard impact Under hard impact conditions, the structure is assumed to be elastic and the colliding

object as rigid. The condition is similar to the assumption made in the validation

process presented in this thesis. The maximum dynamic force applied in this case can

also be calculated using the same equation, by substituting structural stiffness value for

the equivalent elastic stiffness k. The rest of the calculations can be carried out based

on the assumption that structure has sufficient ductility to absorb the total kinetic

energy noted as 221 rmv by plastic deformation. This requirement can be expressed

by using the following expression,

oor YFmv ≤221 Eq. 2.46

where Fo is the plastic strength of the structure and Yo is its deformation capacity.

Indicative probabilistic information for the basic variables used in the Equation 2.46 is

given in the Table C.1 in EN 1991-1-7(2006).

2.15 Knowledge gaps and literature review findings

In this chapter, the literature on impact loads applied on columns in the lateral and

longitudinal direction was reviewed.

• The investigations under dynamic loading conditions highly emphasised the

effects of strain rate, and 30% strength enhancement was expected compared to the

quasi-static conditions. However, this enhancement would be undesirable for

columns susceptible to shear failure conditions.

• A significant reduction of flexural ductility was noticed on structural columns

caused by the increase in compressive axial load level. However, no particular

effort was made in the past to identify the variation of the shear capacity of

columns under instantaneous increment of axial load followed by lateral impact.


2-65

• There is no evidence of research performed on laterally impacted columns under

varying axial loading conditions. Strain rate effects and confinement effects are

highly sensitive to the axial load level, and the resultant impact behaviour of

columns has not been investigated to a satisfactory level. For instance, concrete

elements under flexure exhibit higher strain rate sensitivity than the elements

under compression. This means that the strain rate sensitivity of a column can be

changed with the axial load.

• Most of the tested beams were simply supported and hence, exhibited flexural type

failures. As a result, the beams survived under higher bending moments and

absorbed more energy compared to the static loading conditions. However, higher

modes of vibration due to dynamic impact loads in impacted columns were

inevitable. Consequently, the shear failures were unavoidable under impact

loading and needs further investigation.

• Confinement effects are a governing factor of the ductility of the columns, and

bond stress can increase up to 200% due to confinement effects. Higher strain rate

sensitivity of tensile strength of concrete reduced the amount of cracks under

impact loads, while increasing the bond between reinforcement and concrete.

Therefore prefect bond may be assumed between concrete and steel under impact

conditions.

• Material erosion criteria associated with the Lagrangian mesh discritization may

not represent the realistic behaviour of concrete during an impact.

• Methods proposed by Campbell (1974) and Summers et al. (2001) facilitate

comparison of the hard impact data with more realistic vehicular impact events.

These methods are highly depending on the stiffness of the impacting vehicle.

Based on the literature review findings, a comprehensive research is proposed to

bridge the gaps in the knowledge. In the process, axial load is chosen as one of the

major parameters. Strain rate effects, confinement effects, elevation of the impact,

slenderness ratio, steel ratio, confinement, the amplitude and duration of the impact

are among the other parameters that are needed to be examined.


3-67

3. FE MODELLING OF SHORT RC COLUMNS UNDER LATERAL IMPACT

3.1 Introduction

In general, finite element analysis can be performed based on implicit or explicit

algorithms. The predictor-corrector method is employed by the implicit algorithm to

solve non-linear problems, in which the stiffness matrix is updated throughout the

entire analysis. On the other hand, in explicit codes, the corrector method is omitted

and no equilibrium check is preformed. The advantage of using the explicit method is

that there is no need to calculate stiffness and mass matrices for the entire system, thus

the solution can be carried out at the element level and relatively little storage is

required. In addition to its computational efficiency, it can handle contact problems

effectively. The explicit Finite Element program LS-DYNA was therefore used in the

numerical simulation process (Hallquist 2007).

The procedure used to solve the discrete equation of motion is called explicit, if the

solution at some time tt ∆+ in the computational cycle is based on the knowledge of

the equilibrium condition at time t (Homuda and Hashmi 1996). According to

Newton’s axioms, the forces that act on a structure which is not in equilibrium, cause

acceleration, that can be expressed in terms of velocity and displacement. The

summation of all the forces drives the system to acquire a position of equilibrium.

However, forces acting on the impacted structure change regularly during each time

step and the system does not reach an equilibrium position. Therefore, the stability of

the solution method strongly depends on the time step size. For a typical finite element

mesh it may be necessary to use time steps in the order of microseconds (Mathisen et

al. 1999) which indeed vary with the natural frequency of the structure and the

characteristic length of the elements in discretisation. The explicit time integration

method is conditionally stable, when the time step is sufficiently smaller than the

travel time of the material stress wave across the smallest described element or zone

width within a given FE spatial discretisation (Hallquist 2007). To avoid unpredictable

errors, critical time step size based on the so-called Courant’s criterion must not be

exceeded (Hallquist 2007). This time step, ∆te is determined automatically in a


3-68

conservative way within the relationship of the sound speed, cs and the element lengths,

Le. The sound speed is a function of Young’s modules, E mesh density, ρ and Poisson’s

ratio, ν. Q is a function of the bulk viscosity coefficients Co and C1. The Equations 3.1

& 3.2 can be used for calculation of the time step for solid elements.

( )[ ]2122s

ee

cQQ

Lt

++=∆ Eq. 3.1

( )( )( )ρνν

ν211

1−+−= E

cs Eq. 3.2

Therefore, the time step sizes usually become rather small (10-8s) for reasonable FE

discretisations (15mm). The solution for system of non-linear equations using this

time integration scheme requires the inversion of the mass matrix within each time

step (LS-DYNA 2006).

The LS-DYNA explicit code has the ability to solve advanced, robust and

computationally efficient contact algorithms. That includes crashes, drop tests and

further contact related applications. Although contact zones can be defined to any level

of precision, general contact algorithms are also available where the specified contact

algorithm for the entire model or part of the model can be selected (Rust and

Schweizerhof 2003). In LS-DYNA code, generally displacement control is preferable,

as then the structure can be well controlled globally. In large deformation analysis,

reduced integration should be used to reduce computational expenses. In particular,

reasonable mesh size, boundary and contact conditions need to be chosen by

considering the cs, Le, bulk viscosity and duration of the analysis in order to give

comprehensive results.

3.2 Finite element modelling for impact problems

When it comes to the impact problems, the Lagrange discritization is the state of the art

method. The Lagrangian method, uses a mesh which is not fixed in space but fixed in

body. Thus the cells can be distorted and change their volume (Gebbeken et al. 2007).

To overcome this shortcoming, many numerical codes use erosion criteria and

elements get deleted after achieving the predetermined distortion criteria (Hallquist


3-69

1997). However this phenomenon will not represent the actual behaviour of a material

during an impact. For example, concrete may have 11% compressive strength at the

residual state compared to its undamaged state. The advantages of the Lagrange

method are that the mass of the each Lagrange cell is constant and hence, fulfils the

principle of conservation of mass. In addition, modelling as well as the contact

definitions are easy to implement in this method. The main disadvantage of the

Lagrange method, however are very small time steps resulting from the element

distortion within a short period of time.

3.3 Finite element modelling and selection of material models

One of the biggest challenges associated with modelling the behaviour of reinforced

concrete is the difficulty of incorporating realistic material models that can represent

the observable behaviour of the physical system. This section discusses the material

model formulation used for different materials such as concrete, reinforced steel,

stirrups and rigid body used in the finite element simulation.

3.3.1 Evaluation of Constitutive material models in LS-DYNA

Concrete is known to be ductile in nature under hydrostatic pressure conditions and

may be subjected to brittle failure in tension under impact loading conditions.

LS-DYNA contain several material models that can be used for concrete, however the

actual behaviour of concrete under a three dimensional state of stress is extremely

complicated. Concrete is highly strain rate dependent and any material model should

incorporate these non-linear characteristics. However, due to changes of state within

microseconds, dynamic experiments that determine these parameters are much more

complicated and expensive than the static load test, and therefore data relevant to the

dynamic experiments are limited in the literature. ‘Concrete Brittle Damage Model 96’

presented in the LS-DYNA library is extremely useful in this circumstance as it

employs a fully anisotropic damage rule which is free from adjustable parameters for a

given failure surface (Govindjee et al. 1995). At the same time the constitutive

material model Mat_Concrete_Damage_REL3 offers an added advantage of only one

user input parameter; ie. Unconfined compressive strength is sufficient in the


3-70

calibration process (Malvar et al. 1997; Schwer and Malvar 2005). Since the

unconfined compressive strength of concrete can be easily determined from

experimental testing, these two models are very useful in impact simulation process.

Consequently, these two material models were implemented in the validation and the

superior material model was implemented in the parametric study based on the

accuracy of the outcomes.

In fact, for simplicity some other constitutive models in LS-DYNA adopt highly

restrictive assumptions and hence their applicability is limited to a certain class of

problems. For example, material model Mat_Soil_and_Form uses a perfectly plastic

flow to approximate the post-yield behaviour and is unable to capture the various

softening behaviours of concrete under different loading conditions. Similarly

Mat_Geologic_Cap_Model generates a circular deviatoric cross section where

experimental data indicate that for brittle materials, the shape of the deviatoric section

is circular only at the high pressure regime. In addition, this model cannot predict the

softening behaviour of concrete and confinement effects due to the use of an

associated flow rule (Yonten et al. 2005). The Mat_Soil_Concrete has similar

weaknesses as this model uses only two stress invariants, I1 and J2 while neglecting

third stress invariant J3. It has been also shown that the model does not give a smooth

transition at the softening region due to tri-linear variation of fc which is used define

material softening behaviour from plastic to residual (Yonten et al. 2005). Therefore,

their capability of describing the actual nonlinear behaviour of concrete under

dynamic loading can be different and hence this chapter intends to evaluate the

similarities and distinctive features of the selected models by using the results

generated from a numerical simulation of columns impacted by a rigid mass.

3.3.2 Theory on Mat_Concrete_Damage model

This model uses three shear failure surfaces namely an initial yield surface, a

maximum failure surface and a residual surface with consideration of all three stress

invariants (I1, J2 and J3). The attractiveness of this model is it allows the general

material properties to be generated based solely on the unconfined compressive

strength of concrete and the density. The compressive meridian of the initial yield


3-71

surface cyσ∆ , the maximum failure surface c

mσ∆ and residual surface crσ∆ are defined

independently as;

Figure 3.1: Failure surfaces in Mat_Comcrete_Damage_REL3 (Malvar et al. 1997)

paa

pa

yyoy

cy

21 ++=∆σ Eq. 3.3

paa

pao

cm

21 ++=∆σ Eq. 3.4

paa

p

ff

cr

21 +=∆σ Eq. 3.5

where σ∆ is the stress difference and p is the mean stress in the triaxial compression

failure test. The parameters fyyoy aaaaaaa 121021 ,,,,,, and fa2 can be determined by a

regression fit of the above equations to the available laboratory test data. Having

provided the three separate strength surfaces, the corresponding loading surfaces

representing strain hardening after yield are defined as follows;

ymL σησησ ∆−+∆=∆ )1(

Eq. 3.6

The post failure surface pfσ∆ is defined by interpolating between the maximum

failure surface and the residual surface;

rmpf σησησ ∆−+∆=∆ )1( Eq. 3.7

The variable η is called the yield scale factor which is determined by a damage

function λ which has two distinctive definitions for compression ( )0≥p and

tension( )0<p to account for the different damage characteristics of concrete in


3-72

compression and tension: η varies from “0” to “1” when the stress state advances

from the initial yield surface to the maximum failure surface and visa versa when

softening begins.

[ ]

[ ]

<+

≥+=

∫

∫p

p

pfp

d

pfp

d

bt

p

bt

p

ε

ε

ε

ε

λ

0

0

01

01

2

1

Eq. 3.8

where ft is the static tensile strength of concrete, pdε is the effective plastic strain

increment, and ( ) pij

pijp ddd εεε 32= , where p

ijdε is the plastic strain increment tensor.

The yield stress factor ( )λη governs the nonlinear behaviour of the material and

follows a piece-wise linear relationship whose control points must be pre-defined in

the data input file.

3.3.3 Definition of compressive and tensile meridians at p < fc/3

As the compressive meridian of the failure surface is determined based on the

experimental data with pressures equal or above fc/3, the pressure values below fc/3

predicted by extrapolating the existing values generally over estimate the compressive

strength in that region. The tensile meridian (image) obtained by using the

compressive meridian will also become inappropriate. To eliminate this problem, the

tensile meridian below the pressure range fc/3 is defined as a linear curve derived using

experiments such as uniaxial and triaxial hydrostatic extension tests. The compressive

meridian (image) will then be derived by dividing the tensile meridian by a factor ψ.

The tensile meridian( )tmσ∆ can be obtained from the following equation for the range

of pressures less than fc/3;

( )ttm fp +=∆

32σ Eq. 3.9

The corresponding effective stress ( )σ∆ and the tensile strength (ft) pairs will be

determined from the following testing procedures;

(i) Biaxial tension (on the compressive meridian)

(ii) Plane stress pure shear (θ = 30o)


3-73

(iii) Uniaxial compression,

The tensile to compressive meridian ψ has the following values.

=+≤−

=32321

05.0

cct

t

fpff

fψ Eq. 3.10

For higher compression pressures, the model chooses three additional points as

follows:

( )( ) ( )( )[ ]

≥==++

=

c

c

cccoc

fp

fp

fpfaafaf

p

453.80.1

3753.0

3/23/23/2/ 21 ααααψ Eq. 3.11

For computational purposes the above function is made continuous by a piece-wise

linear interpolation of the discrete points above.

3.3.4 Pressure cut-off and softening

In the model, the maximum principle stress of concrete under tension is limited by the

uniaxial tensile strength ft considering the quasi-static loading rate. In the stress

softening procedure in compression based onpfσ∆ , the yield scale factor η is

controlled by the accumulation of a scaled effective plastic strain pdε . This may cause

a problem when the stress path is closed to the triaxial extension path because there are

no stress deviators and hence no plastic strain accumulation in the model. Therefore

both the damage function λ and scale factor η remain zero. The pressure return from

EoS will decrease from 0 to –ft and will remain at that level thereafter, contrary to what

would happen in concrete after the failure surface is reached. To overcome this

problem the model implements a volumetric damage term by considering volumetric

plastic strain λ∆ in conjunction with the damage function λ;

( )yieldvvddkfb ,3 εελ −=∆ Eq. 3.12

where b3 is a user defined scalar multiplier, kd is an internal scalar multiplier, vε

represents the volumetric strain and yieldv,ε is the volumetric strain at yield. The scalar

fd is used to restrict the effect of this additional term λ∆ , only to the triaxial tensile


3-74

stress path region by limiting the fd so that 0=df when 1.03 2 ≥pJ . With the

implementation of λ∆ , the damage is incorporated for the tri axial tension stress state.

However, post-peak softening behaviour will still not be achieved since there is no

interconnection between the post failure surfaces and the triaxial tension stress path for

pressure above –ft. To solve this problem, the model shifts the pressure cutoff when

softening occurs in the negative pressure range by scaling the original pressure cutoff

by the factor η.

3.3.5 Strain rate effect

In the concrete damage model, the failure surface and the damage function λ are

modified to account for the strain rate effects. Based on the fact that rate enhancement

is experimentally obtained from the unconfined compression and tension tests, and

that they follow the radial paths from the origin in the principal stress difference verses

pressure plane, the model implements a radial rate enhancement on the concrete

failure surface as given in the following equation;

( )fcmf

cme rpr ×∆×=∆ σσ Eq. 3.13

Referring to the maximum failure surface meridian, the modified equation becomes

para

prra

f

ffo

cm

21 ++=∆σ

Eq. 3.14

After the modification to account for the strain rate effects the variation of λ becomes:

[ ]

[ ]

<+

≥+

=

∫

∫p

p

pfrpr

d

pfrpr

d

btff

p

btff

p

ε

ε

ε

ε

λ

0

0

01

01

2

1

Eq. 3.15

where r f is the dynamic increasing factor for concrete. Thus, the current model (REL3)

allows input of the different rate enhancement for tension and compression solely

based on one strain rate curve. Equations given in the CPF-FIP (1990) model code

were implemented to account for the strain rate effects in concrete. For a given strain

rate and concrete grade, the compressive and tensile strength enhancement can be

estimated from these equations. The Dynamic Increasing Factor (DIF) given in the


3-75

code is a design value, which means that the given strength increments are lower than

the values obtained from experiments.

3.3.6 Equation of state

An Equation of State (EoS) is a formula describing the interconnection between

various measurable properties of a system. For the physical state of matter, this

equation usually relates the thermodynamic variables of pressure, temperature and the

volume of a system to one another. An EoS is particularly important to define isotropic

concrete material model Mat_Concrete_Damage (Hallquist 2006).

A failure criterion for isotropic material should be an invariant function of its state of

stress. This means that it must be independent of the choice of the defined coordinate

system. Therefore, a failure criterion is defined using stress invariant. The state of

stress is subdivided into two components: hydrostatic and deviatoric. An EoS is used

to describe hydrostatic material behaviour and relate quantities such as pressure,

density and energy. In contrast, the deviatoric stress state corresponds to the shear

stress and represents the change of shape. Generally, the constitutive model relates the

stress vs. strain behaviour, while the EoS relates the hydrostatic pressure with density

and energy. In addition, the shear stress and stiffness is increased with pressure. The

deviatoric behaviour also depends on the temperature, strain rate and degradation of

material (Gebbeken et al. 2007). This means that these parameters cannot be

considered as constant for a wide range of impact scenarios and hence are not unique

to the particular system.

3.3.7 Evaluation of LS-DYNA material model Mat_Brittle_Damage

3.3.7.1 Material characteristics

This model (second selection) is primarily formulated for evaluating brittle damage in

concrete and hence is useful in the impact simulations. It is well known that numerical

results are very sensitive to the nonlinear material properties. In the absence of more

specific data for the dynamic properties of concrete, the following equation presents

an approximate value for its tensile limit (CEB-FIP, 1990);


3-76

( )

=

o

cit a

fT

2'

lim 58.1 Eq. 3.16

where the compressive strength 'cf should be in MPa and ao is 145 for SI units. Once

the principal tensile stress has been reached at a point, a smeared crack is initiated at

that point with a normal that is co-linear with the first principal direction. Once

initiated, the crack is fixed at that location though it will connect with the motion of the

body. Allowed tensile traction normal to the crack plane is progressively degraded as

the loading progresses. This behaviour is implemented by reducing the material

modules normal to the smeared crack plane by using the program based internal

parameter. In addition, the shear strength basically governs the initial shear traction

that may be transmitted across a smeared crack plane. According to Halliquist (1998),

in the absence of more specific data, the shear traction can be calculated by using the

following equation.

[ ] )exp1)(1( αβ ss Hf −−− Eq. 3.17 According to Equation 3.17 it is interesting to note that the shear degradation is

coupled to the tensile degradation through the internal variable alpha (α) which

measures the intensity of the crack field. In general, the shear degradation factor

accounts for the reduction in the shear stiffness of the material parallel to the smeared

crack plane. The evolution of alpha is governed by a maximum dissipation argument.

Here β is the shear retention factor and βfs represents the shear traction that is allowed

across the smeared crack plane as the damage progresses. The parameter Hs represents

the softening modulus. Govindjee et al. (1995) present a full description of the tensile

and shear damage part of this material model.

As far as the viscosity of the material is concerned, the Perzyna regularization method

was used to implement the viscose behaviour of the material. In order to avoid error

termination, values of viscosity (η ) between 0.71 and 0.73MPa are recommended

(Hallquist 2007). Apart from serving as a regularizing parameter which stabilises the

calculations, the viscosity of the material also allows the inclusion of first order rate

effects during the simulation. Other than the viscosity of the material, the stability of

the calculation process will also depend on the fracture toughness of the material.


3-77

Once the facture toughness is entered, the softening modulus (Hs) is automatically

calculated based on element and crack geometries. Instead of using stress-strain

relationship, the impact behaviour of concrete was defined by using various

parameters which simulate its non-linear behaviour. Table 3.1 shows the reasonable

values selected for Grade 47 concrete. In the absence of definite data, the modulus

elasticity of concrete (E) was calculated by using the following equation (CEB-FIP

1990);

( )[ ] 3/1' / cmoccoci fffEE ∆+= . Eq. 3.18

where 'cf is the characteristic cylindrical strength of concrete, ∆f is 8 MPa, fcmo is 10

MPa and Eco is 2.15x104 MPa. The shear retention factor was selected by using a trial

and error method. The cylinder compressive yield strength (σy) should be used here;

and can be directly calculated from the uni-axial cubic strength by using Table 3.1 of

the CEB-FIP code (CEB-FIP, 1990).

Table 3.1: Material properties used for the concrete

To achieve damage degradation the model employs three damage surfaces which

evolve with material damage. In order to cater for compressive failure, the model

adopts a very simple J2 flow correction method which is not capable of representing

the enhancement of material shear strength due to the presence of high pressure. The

use of a simplified resistance function for concrete elements may ignore the nonlinear

behaviour of concrete specifically near the ultimate conditions.

3.4 Development and validation of a numerical model of a RC column

3.4.1 Introduction

With a view to assessing the vulnerability of columns to low elevation vehicular

impacts, a non-linear explicit numerical model has been developed and validated using

existing experimental results. The numerical model accounts for the effects of strain

rate and confinement of the reinforced concrete, which are fundamental to the

successful prediction of the impact response. The sensitivity of the material model

parameters used for the validation is also scrutinised and numerical tests are performed


3-78

to examine their suitability to simulate the shear failure conditions. Conflicting views

on the strain gradient effects are discussed and the validation process is extended to

investigate the ability of the equations developed under concentric loading conditions

to simulate eccentrically loaded columns. Two of the most sophisticated material

models presented in the LS-DYNA library were compared and one of them was

selected for parametric studies based on its capacity to generate most accurate results.

Experimental data on impact force time histories, mid span and residual deflections

and support reactions have been verified against corresponding numerical results.

3.4.2 Experimental test set up

Feyarabend (1988) tested a square RC column in a horizontal position as shown in

Figure 3.2. Fixed support conditions were achieved at one end by stationary steel

sections fixed at the ground and the other end was attached to a 20t mass that can slides

over horizontal low friction rails simulating roller support. The axial load was applied

through a system of prestressing wires located on either sides of the column and the

impact was generated by dropping a 1.14t mass on to the column at mid span. In order

to account for the shear critical conditions, a separate numerical test was conducted on

the material model and parameters such as fracture toughness and shear retention was

proved to be adequate in representing the shear critical condition by using CPF (1990)

model code and the existing literature on material properties. In the absence of

experimental results from testing of a concrete member with axial load impacted

nearer the support, the above validation was carried out (Thilakarathna et al. 2010).

Figure 3.2: The test set-up by Feyerabend (1988)


3-79

The testing procedure involved impact test on 0.3×0.3m square column specimens

made of 47MPa concrete. The properties of the test specimen are given in Table 3.2

and detailed description of the experimental program can be found in Feyerabend

(1988).

Table 3.2: Characteristics of the Feyerabend’s test specimens (Feyerabend 1988)

Details of test No.SB2 Cross-section (m×m) Span (m) Concrete cube strength, fcu (MPa) Steel Yield stress, fy (MPa) Main bars, ds (mm) Shear stirrups, Avs (mm/mm) Restraining mass (kN) Initial axial load (kN) Striker mass (kN) Impact velocity (m/s) Velocity at which fy was reached (m/s)

0.3×0.3 4.0 47 548 4φ25

12φ@150 196.2 -197 11.18 3.0 ±1.8

3.4.3 Numerical simulation of the physical testing

According to El-Tawil et al. (2005) the peak force generated at the impact is not

representative of the design structural demand, as the structures do not have enough

time to respond to a rapid change of loading. Consequently, it is assumed that the

lateral movements of the restraining mass and the elongation of the prestressing cable

system during the impact do not affect the impact behaviour of the column. Hence the

prestressed cable was excluded and the axial load applied as a ramped up surface

pressure over the cross section. The small fluctuation of the axial load due to the

impact is also neglected as the fluctuation is unlikely to alter the flexural or shear

capacity of the column.

Figure 3.3: The simplified test set-up and the cross section of specimen

}


3-80

The experimental set-up of the impacted column is simplified as shown in Figure 3.3.

Half of the column was modelled and appropriate boundary conditions were

introduced to maintain the symmetric conditions. One support was restrained against

rotation while allowing translation along the longitudinal direction at the other end to

simulate partial restrained conditions at that end as shown in Figure 3.3. These

boundary conditions allow representation of the much complex fixed and free sliding

supports with constraints as close as possible.

3.4.4 Convergence study and mesh discritization

Discretization is the process that transfers continuous models into discrete

counterparts. The discrete elements generated in the simulation process is solved by

assuming linear equations where displacement at each node is calculated for the given

load and its constraints, and is then used to approximate the stress contours of each

element. As the stress approximation is based on the relative displacement of the

individual nodal locations, the stress contour will not necessarily be continuous from

one element to the next thus causing an error. This discritization error can be

minimised by decreasing the size of elements. However, small mesh discritization will

significantly increase the duration of the analysis and demand a compromise between

the duration and accuracy.

Figure 3.4: Convergence of the numerical model A convergence study is used to solve this problem and a typical rule in convergence

study is to double the element density for each iteration in the area of interest.

According to Burnett (1987), drastic local refinement should be avoided in mesh


3-81

convergence studies. Therefore, in this investigation, the element size homogeneously

varies from 50mm to 5mm at each stage. The total energy accumulation in the

impacted column is shown in Figure 3.4 under the 3.0ms-1 impact. Figure 3.4

illustrates that the total energy of the model converges towards 2565J as the size of the

elements decrease. With the convergence of the total energy, peak displacement is also

converging by limiting the hourglass energy accumulation to an acceptable level.

According to the analysis the 25mm mesh discritization provides the optimum

solution. It is also evident that a further decrease of the mesh size had very little effect

on the accuracy of the results, while at the same time, increased the duration of the

analysis dramatically. The error due to the selected mesh size is only 2% compared to

the 5mm mesh, and hence negligible.

Figure 3.5: Mesh generation for the impacted column & rigid body

Consequently, the concrete column was modelled using 25mm eight node hexagonal

‘constant stress’ solid elements with one point integration (see Fig. 3.5). Similarly

25mm long beam elements were used for both vertical and the lateral reinforcements

with 2×2 Gauss integration. The vertical reinforcements are defined as truss elements

and the links are defined as a Hughes-Liu beam element with cross sectional

integration (Flanagan and Belytschko 1981). 35mm cover is assumed for the main

reinforcement. Eight noded hexagonal solid elements with one point integration were

employed for the drop mass. No attempt was made to simulate the guide rail or

restraint when the drop mass bounces back after the first impact. A nominal radius was

maintained at the bottom of the mass to avoid stress concentration along the perimeter

and initial analysis was carried out by assigning rigid body conditions (Hallquist 2007).

Rigid body

Impacted column with 25x25mm hexagonal elements


3-82

Movement caused by the gravitational acceleration were not considered in the analysis

as the main impact pulse lasted for a maximum of 10ms.

As far as the contact between concrete and steel is concerned, chemical adhesion,

frictional resistance and rib-bearing are the main components of the bond between

concrete and the steel. Under the influence of the small stress difference, the chemical

adhesion predominates over the other two components. Wedging action becomes

predominant when chemical adhesion alone is not sufficient to resist the stress

difference. Longitudinal and radial cracks are generated under the influence of

wedging action. Restraining effects due to wedge action are replaced by the frictional

forces only after propagation of the longitudinal and radial cracks. In addition, the

entire process may alter due to the confinement effects, strain rate, phase

transformations and the pressure variations. However, due to their complexity, all

these factors cannot be taken into account in a simulation.

As far as deformed reinforcement bars are concerned, the ultimate dynamic bond at

failure was 70-100% higher than that under quasi-static loading conditions

(Weathersby 2003). Also the steel deformation under the impact load was limited to a

region only few centimetres long beneath the point of the impact (Bentur et al. 1986).

This means that there was not enough time to develop extensive bond slip along the

length of the bar under impact condition. Therefore perfect bond was assumed

between the reinforcement and concrete. Under these circumstances, strain hardening

characteristics of the longitudinal steel may not alter the numerical results

significantly.

3.4.5 Contact algorithm and prevention of initial penetration

Segments with high velocity conditions often come into contact under impact

conditions. This results in large deformation of the segments. In the finite element

modelling, instead of being deformed after the collision, one segment penetrates the

other segment, especially if the stiffness of the segments is different. This is known as

initial penetration. The initial penetration occurred at the contact interface between the

column and the rigid body, leading localised stresses and strains that were quite

unusual.


3-83

Suitable contact interfaces themselves can be effectively used to eliminate the overlap

or penetration between the interacting surfaces. The detection of a penetration is also

an important part in this elimination process. In brief, to detect penetration due to

contact, the LS-DYNA code firstly performs a global search followed by a local search

using incremental search techniques. After detecting any penetration condition, the

amount of penetration is calculated and a force is applied to remove the penetration.

This force can in some instances be very large and may create negative energy

conditions. To avoid negative energy conditions, a suitable contact definition can be

used as a default treatment. Detailed description of the contact definition including

penetration detection and elimination process can be found in LS-DYNA theoretical

manual (LS-DYNA 2007).

In order to prevent contact break downs similar mesh discretisation was used in this

simulation for the drop weight and the column including the contact treatment

AUTOMATIC_SURFACE_TO_SURFACE (Hallquist 2007). This standard penalty based

formulation consists of interface springs between all the contact surfaces which apply

interface forces proportionate to the amount of penetration. In a typical discrete

environment the nodes of the softer body (slave) will be penetrated by the harder body

(master). The selection of the slave and master surface can however be arbitrary with

this formulation, since there are two checks for possible penetration between slave

and master surfaces and vice versa.

Contact stiffness can also be effectively used to handle the initial penetration

conditions. Since the default stiffness option depends on the material properties and

the size of the segment, it is inadequate to handle the initial penetration between

dissimilar materials. In addition, the default treatment may distort the original

geometry at location, where the penetrations are detected. The soft constrained-based

approach (SOFT=1) was effective in this circumstance where contact stiffness

calculations are based on stability considerations by taking into account the time step.

This option was particularly useful to address the dissimilarities of the mesh

orientation and stiffness of the two bodies. The contact nonlinearity was further

stabilised by assigning a value of 30 for the Viscous Damping Coefficient. In addition,

friction coefficients were introduced to simulate the frictional forces that were


3-84

transmitted across the contact interface due to the nominal radius that was maintained

at the bottom of the impacting mass. The friction coefficients were assumed to be

dependent on the relative velocity of the column (concrete) and rigid body (steel) that

were in contact and consequently the selected values for the static and dynamic

friction coefficients were 0.6 and 0.5 respectively (McCormick 2009).

3.4.6 Validation of the finite element model using Mat_Concrete_Damage

3.4.6.1 Material characteristics

Impact loads generate tri-axial state of stress in concrete columns. For instance, from a

material point of view, spalling occurs at the contact interface as a result of tri-axial

extension stress conditions. Subsequently impacts on the column generate tri-axial

compression stress conditions in the core concrete. In the meantime, uni-axial tensile

stresses will be generated at the opposite face by scabbing the concrete as a result of

wave reflection at the boundaries. Therefore, a material model that can replicate the

results of tri-axial tension tests, tri-axial compression tests, and uni-axial tensile tests

may be much suited in the impact simulation process. Mat_Concrete_Damage fulfils

this requirement. Each of these tests will represent different damage modes. However,

the possibility of combined mode of failures may not be negligible.

3.4.6.2 Elimination of mesh dependency of the fracture toughness

Figure 3.6: Single element under uni-axial tensile test

As the softening part of the unconfined uni-axial tension stress-strain curve is

governed by the values of two parameters b2 and b3 (Refer to Equations 3.12 & 3.15),


3-85

these parameters have to be adjusted to minimise the differences between the

numerical and the experiment results (Schwer and Malvar 2005). For example, the

softening behaviour becomes mesh-dependent if it is not governed by a localisation

limiter or characteristic length. Thus, the mesh dependency of the fracture energy has

to be corrected by changing the parameters assigned for the tensile softening of the

material. The mesh dependency can be eliminated by selecting ‘h’, ie. the size of the

finite element so that the ratio hGf is equal to the area under the stress-strain curve

for the uniaxial unconfined tensile test, where fG is the fracture energy of the concrete.

This procedure eliminates the mesh size dependency on the fracture toughness.

Figure 3.6 shows the stress-strain response of a single element (1×1×1mm) subjected

to uni-axial tension test. According to the CPF-FIP model code, the fracture energy of

Grade 47 concrete should be in the range of 100 Nm/m2 to 125 Nm/m2 (CEB-FIP

1990). Hence, the default value of b2 (b2=1.35) will overestimate the fracture

toughness of 47MPa concrete. Therefore, the value of the b2 is adjusted until the area

under the stress-strain curve becomes 120 Nm/m2. However, no specific guidelines are

found in the CEB-FIP code related to the strain rate effects on fracture toughness.

Therefore strain rate effects are exempted from the fracture toughness and the selected

value is based on the static material characteristics similar to the one used by Unosson

(2001). Similarly, parameter b3 is determined from the hydrostatic tri-axial tensile test.

The default value of b3 (b3=1.15) leads to desired results and hence accepted without

alteration. The effects of these parameters are crucial where tensile and shear failure

characteristics are more predominant.

3.4.7 Material properties of steel

Longitudinal reinforcement and hoops were modelled as elastic perfectly-plastic

materials by using Mat_Plastic_Kinematic model which has the reputation of

minimising the duration of the analysis and is available for Hughes-Liu beam elements

and truss elements (Hallquist 2007). Kinematic hardening is implemented for

reinforcement with strain rate effects. Table 3.3 represents the material properties

adopted for main reinforcement and Strain rate was incorporated using the

Cowper-Symonds model given in Equation 3.19.


3-86

P

s

d

C

1

1

+=′ ε

σσ &

, Eq. 3.19

where, d

σ ′ is the dynamic flow stress at a uni-axial plastic strain rate ε& , and σs is the

associated static flow stress. C and P represent the material constants. The relevant

values for steel can be found in Table 8.1 of Stouffer and Dame (1996). Failure strain is

not defined here as there is no evidence of a steel rupture. The hardening parameter is

βh=0 to represents kinematic hardening characteristics.

Table 3.3: Material properties used for the main reinforcement Density (kg/m3) Es (GPa)

Poisson’s ratio ρ

σ (MPa)

Et (GPa)

Hardening Parameter (βh)

C p

7800 210 0.30 548 2.0 0 40 5

As the falling weight of 1.14t did not experience excessive deformations, rigid

material model was implemented for the impacted mass. Rigid elements are bypassed

in the element processing and no storage is allocated for storing history variables.

However the inertial properties of the materials are calculated from the geometry of

the elements and hence realistic values for the density and the modules of elasticity

must be provided. In addition when the rigid body interacts with the column, the

contact interface parameters are determined by using the given Young’s modules ‘E’

and Poisson’s ratio ‘ρ’ values of the material (Hallquist 2007). The material

characteristics used for the drop mass are given in Table 3.4. The displacement and

rotational constraints for the rigid body were restrained about global X and Y

directions.

Table 3.4: Material properties used for the impacting body Density (kg/m3)

Young’s Modulus (GPa)

Poisson’s ratio

7800 210 0.3

3.4.8 Load simulation for a dynamic system

When imposing static loads in explicit environment the ramp up loading would be the

better solution to avoid stress fluctuation. The ramp up load must be increased from

zero to its final value and the load curve should extend beyond the termination time for


3-87

stability. In general, the ramping duration should be greater than the fundamental

vibration period (0.004s) of the column. Furthermore, it is observed that the

stabilisation depends on the wave speed and the superimposing effects of the reflected

waves at the boundaries. The minimum fluctuation was observed when the ramping

duration was around 0.02s, above which there was instability again. The lateral impact

force was applied after achieving the vertical load stabilisation. The rigid body is

accelerated from zero velocity to 3ms-1 at impact over a 10mm gap. Realistic value for

the density is essential for the rigid body since conventional equations of motion are

involved in the velocity generation.

3.4.9 Hourglass energy and damping effects

To determine the stiffness, damping and mass matrices and the external force vector, it

is necessary to integrate the different terms that are present in the equation of motion,

with respect to the volume of the object. The method typically used for the integration

is the Gauss quadrature rule. In particular, in LS-DYNA code the integration is

conducted with a low order Gauss method. For example, reduced integrated solid

elements use only one integration point, located at the centre of the element. This low

order integration is beneficial in terms of CPU time but may come across some

deformation modes, which create zero strain at the Gauss point, determining a zero

internal energy. This is usually known as hourglass problem. Since these

zero-energy-modes have no stiffness, they may cause numerical errors, decrease the

time step size and interrupt the calculation. If this formulation is implemented in a

simulation, an hourglass control algorithm is mandatory (Schwer et al. 2005). The

other types of formulations, such as, fully integrated selective reduced (S/R) solid and

quadratic eight node element with nodal rotation, generally do not exhibit any

hourglass problems. However, these types of element formulation often increase

computational cost. Additionally, they are unstable in large deformation where some

tendency to 'shear-lock' and thus behave too stiffly in applications particularly when

the element shape is poor.

Elements with reduced integration are more robust in impact simulations because the

strain terms evaluated at the integration point remain well conditioned at larger


3-88

deformations. The hourglass resisting force vector kif α is given in Equation 3.20 and

descriptive details of each parameter can be found in (LS-DYNA 2007). In brief, the

parameter ακΓ depends on nodal velocity and hiα depends on nodal coordinates

respectively.

kihk

i haf ααα Γ= Eq. 3.20

Where 4

32 s

ehgh

cvQa ρ= . Eq. 3.21

In which eν is the element volume, ρ is the density, cs is material sound speed, and Qhg

is a user defined constant. When Qhg is equal to 0.05, improved results have been

observed. As this equation contains a component of the volume ve, theoretically the

hourglass error should reach zero with the mesh refinements. Moreover, the default

setting of LS-DYNA which is given by the Equation 3.20 is not orthogonal. Therefore

the orthogonal approach, as described by Flanagan and Belytschko (1981) is used in

this simulation. As far as the damping effects are concerned, damping forces do not

impose significant effects during impacts where duration of the event is shorter than

the fundamental natural period of the specimen (Zeinoddini 2008; Jones 1997; Strong

and Yu 1993). However, when the damping effects are introduced to the system based

on ‘mass weighted damping’ method which is ideal for damped low frequencies and

rigid body motion (LS-DYNA 2007), the deflection of the column and impact force

reduced slightly with negligible increment of the contact duration. Since the procedure

does not significantly improve the post peak behaviour of the column the damping

effects were excluded. Moreover, the application of a specific damping coefficient

cannot be justified in a realistic situation where the supporting structure has many

elements that are interconnected.

3.4.10 Procedure for axial load application

Axial load on a column can be applied as a uniform surface pressure over the gross

cross-sectional area of the column or as combination of loads on the steel and concrete

areas separately by assuming uniform strain distribution. In the absence exact

knowledge on load sharing between the concrete and steel in a column, researchers


3-89

used the first option based on the assumption that perfect bond between the concrete

and steel can generate a uniform strain condition (Shi et al. 2008). However, there is an

uncertainty in the axial force transferred to the longitudinal steel when the total axial

load is applied as a surface pressure over the gross surface cross-sectional area of the

column. Therefore it is prudent to investigate the above options in detail and select the

one that minimises the error due to the contact enforcement.

In order to achieve the uniform strain condition a column subjected to pure axial

compression must fulfil the following requirement.

sc εε = Eq. 3.22 where εc and the εs represent the longitudinal strain in concrete and steel respectively.

For linear elastic behaviour, stresses in concrete and steel are in proportion to the

modular ratio, cs EEn /= . Then by considering axial load compatibility, it is possible

to derive the following equation.

totalcc nA

P

ρσ

+=

1

1

Eq. 3.23

where P is the axial load on the column and Ac is the cross sectional area of the

concrete column. ρtotal steel ratio of the section. This equation is based on the

assumption that strain of concrete and steel vary in an identical manner under the axial

loads, which is not particularly true at the ultimate stage. However the longitudinal

steel yields before concrete. In addition, with the implementation of the axial load

reduction factor 'φ =0.6 according to AS 3600 (2004) the error induced due to the

assumption of strain compatibility can be neglected.

Deflection of the column under the separate application of axial load on steel and

concrete areas is different from the one with uniformly distributed load. This may

cause a substantial difference of the failure load at near ultimate stage under lateral

impacts and hence the full bond between concrete and steel will not ensure the strain

compatibility particularly close to the load application region. In fact, this would affect

the shear capacity of the column under the lateral impact conditions. On the other hand

wave propagation effects through the material may alter due to the artificial stress

concentration. Because of the same reason separate applications of axial load perform

well as far as hourglass is concerned. Consequently, the axial loads must be applied on


3-90

steel and concrete areas separately and the numerical simulation process must contain

a contact enforcement verification phase other than the mesh discretisation. By this

way, the numerical simulation can represent the actual behaviour of structural columns

particularly when it forms a part of a structure.

3.4.11 Confinement effects under strain gradient

The confinement effects can improve the strength and the deformability of the

concrete columns under impact loading conditions while mitigating the damage. The

confinement effects were therefore taken into account in the numerical simulation by

assigning enhanced concrete characteristics to the core concrete and unconfined

characteristics to the cover concrete based on the equations proposed by Mander et al.

(1988).

a) Section under concentric loading b) Section under eccentric loading

Figure 3.7: Lateral pressure distribution and the resultant strain gradient

It is worth investigating the capacity of this method to simulate concentrically as well

as eccentrically loaded columns under impact. In fact, numerous models have been

proposed to simulate the confined characteristics of concrete both under concentric

(Mander et al. 1988; Sheikh and Uzumeri 1982) and eccentric loading conditions

(Sargin 1971; Lokuge et al. 2003). Enhanced stress-strain characteristics that develop

under concentric loading conditions may not be applicable to eccentrically loaded


3-91

conditions where the stress-strain distribution across the section substantially differs

from that under concentric loading conditions (see Fig. 3.7). Therefore the resultant

strain gradient in eccentrically loaded columns was numerically simulated by dividing

the cross section into a number of strips and then assigning various stress-strain

characteristics based on the depth to the neutral axis (Lokuge et al. 2003). The validity

of such a method may be uncertain as impacted columns may generate higher

vibration modes where the compression and tension sides interchange over the height.

However, in similar investigations, the confinement characteristics developed under

concentric loading conditions were applied to eccentrically loaded columns (Lokuge

et al. 2003; Saatcioglu et al. 1995). The results confirmed that the capacity predicted

from the confinement characteristics developed under concentric loading conditions

was adequate for evaluating the behaviour under eccentric loading conditions,

particularly for low strength (56MPa or less) concrete columns. One of the

assumptions of such investigation was that strength decay is essentially a function of

confinement stress and does not vary with the strain gradient (Sargin et al. 1971). In

addition, the investigation considered the fact that the stress-strain curves in the strips

closer to the neutral axis may not be substantially different to the one that under fully

confined conditions particularly in the preloading strain range (Saatcioglu et al. 1995).

Moreover, it is important to note that the flexural cracks appearing on the column at

the ultimate stage under the impact will minimise the stress differences in various

layers across the section. All these observations lead to the conclusion that

vulnerability assessments of the eccentrically loaded columns are independent of

confinement characteristics resulting from strain gradient. Therefore the present

vulnerability assessment techniques could be extended to the eccentrically loaded

columns by assigning uniform confined compressive characteristics to the core

section.

3.4.12 Numerical and experimental results for Mat_Brittle_Damage

The aim of the validation process is to investigate the ability of the material model

Mat_Brittle_Damage to simulate the dynamic response of the impacted columns. As

an axial load is present on the column, the effects of axial stress propagation, axial

shortening and shear deformation characteristics must also be reflected in the


3-92

simulation process. The results of the numerical simulation follows the displacement

characteristics of the experimental results accurately up to t=60ms as shown in Figure

3.8. However, the residual deflection characteristics of the impacted column were not

simulated by the numerical model where non-linear characteristics of the concrete play

a vital role. This deviation reflects the error introduced by the use of a simplified

resistance function for concrete elements which ignores the nonlinear behaviour of

concrete especially near the ultimate stage. In fact, this material model implements

only first order strain rate effects by using a viscosity parameter for concrete. However,

the major deflection characteristics are well reflected by the model. This indicates that

the second order strain rate effects can be exempted from the numerical simulations of

the impacted columns. This is a reasonable assumption as it was observed that the

average strain rate across the impacted column is less than 0.1s-1. The reaction force

generated at the interface also agrees reasonably well with the experimental results

(see Fig. 3.9). As a whole, these results are an indication of the accuracy of the

stiffness and axial inertia characteristics of the model as there is strong interaction

between the axial and lateral inertia effects for the dynamic elastic-plastic deformation

of a column with axially moving and stationary ends (Karagiozova and Jones, 1996).

Figure 3.8: Comparison of the resultant deflections Figure 3.9: Interface forces during the

impact 3.4.13 Comparison of numerical and experimental results for Mat_Concrete

_Damage

Time histories of the mid span deflection of the impacted column as predicted by the

FE model is compared with the published experimental data as shown in Figure 3.10.

The resultant maximum deflection and duration of the impact are well reflected in the


3-93

results. Hence the inertia and stiffness of the impacting bodies, boundary conditions

and effects of confinement are accurately simulated in the numerical model. Also, it is

important to note that the residual displacement of the column was reasonably

approximated by the numerical results even after cracking and slight crushing of the

concrete occurred simultaneously during the impact as shown in Figure 3.11. Local

crushing is important as this can reduce available energy and consequently modify the

post impact behaviour of the impacted column. In addition the result confirmed the

fact that the stress-strain characteristic developed under concentric loading conditions

can be assumed over the entire core section to simulate the flexural failure conditions

of impacted columns made of low grade concrete.

Figure 3.10: Comparison of the resultant deflections

Figure 3.11: Crack Propagation of the impacted column and numerical simulation

In Figure 3.12, the continuous line represents the contact force history obtained

through the numerical simulation and the dotted line represents the one that was

obtained from the experimental test. It can be seen that the differences between those

two graphs are insignificant and the impact-time histories are almost identical. On the

other hand, exact simulation of the peak contact force is evidence of the accurate


3-94

representation of the stiffness and boundary conditions as it would mainly depend on

the inertia characteristics of the column for given boundary conditions. The

comparison is quite good for both of these global parameters (Impact force and

deflection). Therefore the model as well as the input data used for the simulation can

be regarded as satisfactory.

Figure 3.12: Comparison of the resultant Figure 3.13: Comparison of the resultant reactions impact force

Some differences are observed in reaction forces generated by the numerical

simulation compared to the experimental results (see Fig. 3.13). The sources of the

error may be explained as below: In the experiment, the load cells were placed in

between the bearing plates and the supports to measure the reaction. In the absence of

descriptive details of the bearing system, the load cells were not modelled in the

numerical simulation. However, the reaction forces are found to be very sensitive to

the recording position (Zeinoddini 2008). Other possible reasons would be the

inevitable error induced by the partial fixity of the supports due to the stationary steel

sections and the filtering procedure used to extract the data from the data logger

system (Feyerabend 1988). In spite of the differences in local peaks, the FE model

prediction appears to be tracing the trend quite well.

The FE model predicted crack pattern in the impacted columns and those observed in

the experiment are presented in Figure 3.11. Tension cracks initiated at the bottom and

top of the beam followed by the crushing of the material beneath the impacted zone.

The simulation reproduced the tension cracks at the bottom while displaying dense

tensile crack propagation at the top surface of the beam. The dense crack concentration

occurred directly beneath the impacted zone with the partial separation of the material

represented as a region with higher stress accumulation. Based on these factors it can


3-95

be concluded that the FE numerical simulation reasonably predicts the impact

behaviour of the axially loaded column.

On the whole, LS_DYNA material model Mat_Concrete_Damage_REL3 generated

better results. It has been also proven that the default parameters generated by the

software based on the unconfined compressive strength of concrete satisfy the

majority of the well characterised tests results for 45.6MPa concrete (Schwer and

Malvar 2005). In addition, many researchers have proven that this material model

generates better results in dynamic simulations (Zhenguo and Yong 2009; Yonten et al.

2005; Bao and Li 2009). However, the model is still subject to the possible deviations

that can occur with other concrete grades. To minimise the associated errors the

parametric study will be limited to 30MPa to 50MPa concrete by avoiding High

Strength Concrete (HSC).

3.5 Conclusions

In fact, validated computational models can dramatically simplify the analysis process

and greatly reduce the cost and time involving in physical testing. The underlying

challenge in this method is the capability of the constitutive models of concrete to

represent the realistic response of the columns under the impact loading conditions.

Based on its distinctive features, reliability and capacity to simulate contact impact

problems, the explicit finite element software LS-DYNA was selected for the

numerical simulation process. Two of the most sophisticated material models present

in the LS-DYNA library were compared and one of the material models was selected

for parametric studies based on its capacity to generate the most accurate results.

The main conclusions of this chapter are summarised below:

1. It is found that the numerical simulations of the column tests can be simplified

greater by isolating the impacted column from the connecting structure and by

assuming perfect bond between steel and concrete. However axial load must be

applied separately on steel and concrete in order to maintain uniform strain

distribution.


3-96

2. A material model that can replicate tri-axial state of stress must be used in the

impact simulation process. LS_DYNA material model

Mat_Concrete_Damage_REL3 fulfils the requirement. However the mesh

dependency of the fracture toughness has to be corrected by a single element

analysis to enable simulation of tensile and shear failure conditions. In addition,

contact algorithms, hourglass problems and initial penetration condition must be

treated carefully in the simulation process. Conversely, strain hardening and the

strain rate effects of the longitudinal steel may not alter the result significantly.

Damping effects may also be exempted from the numerical simulation.

3. For impact simulation, it is suggested that the application of the stress-strain models

developed under concentric loading conditions is valid under eccentric loading

conditions particularly for low grade of concrete.

4. Comparison of the results shows that the both concrete material models are

satisfactorily describes the behaviour of the impacted columns with some

limitations under specific conditions. As the Mat_Concrete_Damage generated the

most accurate results it will be used for the further analyses.

5. A better understanding of the impact behaviour of the column is reached with the

supplementary information from the numerical simulation. An extended model

generated from the validation process can be used for a parametric study of typical

columns in multi-storey buildings which are highly susceptible to vehicle impacts.


4-97

4. IMPACT RECONSTRUCTION AND PARAMETRIC STUDIES

4.1 Introduction

Analyses of concrete columns under vehicle impact loads are limited in the literature.

Therefore noticeable differences can be seen in the magnitude of the specified loads

prescribed by different authorities for columns subjected to lateral impact loads

(Vrouwenvelder 2000). Another reason for this difference is the complex behaviour of

the vehicle column interaction (Louw et al. 1992). For instance, Reinschmidt et al.

(1964) and Feyerabend (1988) conducted a series of tests by applying hard impact

loads in their experiments whereas experimental studies on the structural behaviour of

the rail and the road vehicles demonstrated that the colliding vehicle would exert soft

impacts (Dodd and Scott 1984; Sutton and Lewis 1984; Varat et al. 2000). As far as

soft impacts on concrete columns are concerned, there are several questions that

remain unanswered, namely;

(a). What is the overall dynamic strength enhancement factor of the impacted column

compared to its static capacity?

(b). What are the effects of the boundary conditions on the strength enhancement?

(c). What are the influences of the parameters such as steel ratio, concrete grade,

effective height on ultimate capacity (Louw et al. 1992)?

According to Popp (1965) the ratio of dynamic to static loads at failure could be as

high as 2.7 for a 3.8m high hinged column that is struck at 1.2m elevation by an 18t

truck. However, this enhancement may not be representative of the entire impact

loading particularly when the flexural shear is critical. Nevertheless, the DIF of the

shear capacity always exceeds that of the moment. Also, it must be kept in mind that

during a hard impact the high strain rates occur when the flexural failures are small,

whereas during a soft impact the greater flexural strains are accompanied by high

strain rates (Louw et al. 1992). Therefore selection of a realistic impact duration is

mandatory in vehicle impact simulations, in addition to the elevation of the impact.

Even though non-linear finite element analysis is a speedy process compared to


4-98

experimental testing, impact reconstruction is still a challenging task. Therefore, finite

element methods have been directed towards the development of a proper simulation

technique, which, in turn, enables researchers and designers to have better insight of

vehicle impacts. To this end, an extensive parametric study based upon the finite

element code LS-DYNA proves to be a useful tool and the findings can be easily

summarised to develop a rational basis for the design of impacted columns.

Consequently, a comprehensive study has been conducted to determine the appropriate

shape of the impact pulses, the effect of the pulse characteristics, mass of the vehicle,

strain rate, contact area and duration of the impact (or the stiffness). As a result, a

universal technique which can be applied to determine the vulnerability of the

impacted columns against collisions with new generation vehicles under the most

common impact modes is proposed. Additionally, the observed failure characteristics

of the impacted columns are explained using extended outcomes. Columns having

circular sections have been comprehensively studied under low to medium velocity

impacts.

4.2 Impact reconstruction by using crash test data

Vehicle-column interaction plays a vital role in the impact simulation process.

However, generation of realistic numerical models of vehicles for impact simulation is

quite complex and difficult to obtain. On the other hand, specified methods (or models)

used in the past are limited in their application to assess the vulnerability of columns to

impacts from a general vehicle population (El Tawil et al. 2005; Tsang et al. 2005).

The crash response depends on the mode of impact, rate sensitivity of the vehicle,

dynamic crush characteristics, restitution, collision partner and vehicle specific

parameters (Varat and Husher 2000). Therefore the force history of an impact at one

particular velocity is different from vehicle to vehicle even though the mass remains

the same. Consequently impact simulation with a rigid body or even with the

simplified deformable body assumptions may not be adequate. For instance, a typical

contact force history of a rigid body impacting a concrete column has two force pulses

(see Fig. 3.12). The main pulse lasts only for about 6ms and this will highly exaggerate

the strain rate effects.


4-99

In fact, a linear relationship between the absorbed energy and residual deformation of

the vehicle can be assumed under the deformable body assumption (Campbell 1974;

Varat et al. 1994; Neptune 1999). Based on this assumption, the stiffness of the vehicle

in frontal impact can be represented by a constant spring stiffness. This Campbell

Model (Campbell 1974) was further improved by Prasad (1990) with the aid of

repeated barrier impact testing. One assumption maintained during the reformulation

is that a constant liner spring rate over the entire depth is applicable. However,

available crash data indicates that vehicle frontal stiffness cannot be precisely

modelled through the use of single linear springs for all vehicles, in particular when

the crash propagates to the passenger compartment (Varat et al. 1994). On the other

hand, the maximum resulting dynamic interaction force calculation based on the ‘Hard

Impact’ assumption (EN 1991-1-7: 2006) is exceptionally sensitive to the equivalent

elastic stiffness of the impacting object (i.e. the ratio between force F and total

deformation) (EN 1991-1-7: 2006). Hence it is inappropriate to represent the impact

response of a vehicle population having a unique mass but different stiffness, due to

the inherent relationship between the force (amplitude) and the total deformation.

Having observed the limitations of the existing methods, this research implements an

impact pulse generated from a typical car to rigid barrier impact to reconstruct the

vehicle collisions so that the generated results can be applicable to a general vehicle

population over common modes of collisions.

4.2.1 Vehicle-Column Interaction

Figure 4.1: Front and side views of an impacted column (NHTSA)

Pressure transducers

Location of the impact


4-100

The area of contact has some correlation with the mode of failure and the dissipation of

kinetic energy through structural deformation. Hence, the vulnerability will depend on

the distribution of the pressure over the contact interface. Euro code EN 1991-1-7

(2006) suggested that the contact area should be 0.25m high for car impacts. However,

there is no guidance on the lateral distribution of the pressure across the section,

particularly for circular columns. Figure 4.1 shows a rigid pole used to conduct several

full scale impact tests on cars (NHTSA). It is evident that the effective contact area is

around 25% of the perimeter if the frictional effects are neglected. Therefore, uniform

normal pressure distribution is assumed across the 25% of the perimeter and the

resultant lateral pressure distribution used in this study is shown in Figure 4.2.

Figure 4.2: Lateral pressure distribution across the diameter of the 300mm column

When the column diameter increases, the contact area also increases and contact pressure will

reduce any particular impact. The resultant pressure distribution affects only the local damage. On

the other hand, to account for the possible changes to the resultant impact duration, a method is

proposed to calculate the peak forces under 50ms to 150ms impact durations.

4.2.2 Effects of Impact Pulse Parameters

As impact pulse parameters play a vital role in the generation of a universal method to

determine the vulnerability of columns under all modes of collision, it is intended to

investigate the effects of pulses generated from full scale vehicles to rigid barrier

impact in the impact reconstruction process. To investigate the shape of the various

curves to simulate the collision pulses, force histories generated from several full scale

frontal collision scenarios are studied. The ‘MATHLAB’ program can be used to


4-101

generate the ideal curve by using accelerometer data alone and hence derivation of

mathematical model for different curves by using parameters such as velocity profile,

frontal stiffness, residual deformation and restitution is not considered at this stage.

Based on this method several force time histories derived by using accelerometer data

presented in National Highway Traffic Safety Administration (NHTSA) are compared

in the following paragraph.

Figure 4.3: Force Time histories of full scale crash tests (NHTSA 1997)

Comparisons of the Force -Time history of the Honda Accord, Ford Taurus & Renault

Fuego are shown in Figure 4.3. Durations of the impacts were 100ms and triangular

pulse shape is best fitted with the force history diagrams. This pulse shape has already

been identified as a useful collision pulse model to simulate the frontal impact

conditions (Breed et al. 1991). Sine, Haversine (Tsang et al. 2005; Campbell 1974),

and square formats (Brach 1991; EN 1991-1-7: 2006; Vrouwenvelder 2000) are the

other standard formats used widely to represent the frontal vehicle impacts.

Comparative analysis revealed that the peak forces generated by the side impacts are

less significant as the peak force is small and the impact duration is high and

consequently the strain rate effects are also minimized. Numerical approaches to

determine impact pulse parameters are rare in the literature as collision pulse

characteristics depend on the mode of impact, collision partner and vehicle specific

parameters. In fact, equations derived based on velocity profile, frontal stiffness,

amount of deformation, and restitution may not be applicable to a general vehicle

population (Varat and Husher 2000).


4-102

Table 4.1: Comparison of the force time history data with properties of the Impulse

Vehicle Velocity v(ms-1)

Mass m (kg)

Impact Duration (s) (mv)

Area of the Equivalent Triangle

Percentage variation

Ford Explorer 16.9 2242 0.150 37890 39573 1.04 Ford Taurus 15.6 1619 0.100 25274 27794 1.09 Renault Fuego 13.3 1329 0.100 17793 16687 0.94 Honda Accord 13.4 1329 0.100 17830 19500 1.09

A comparison of the area of the triangular impulse generated from the full scale impact

tests (NHTSA 1997) with the product of mass and the velocity of the respective

vehicles is given in Table 4.1. The force time history was obtained using the data

generated by the accelerometer placed at the centre of gravity of the impacted vehicle.

It is evident that the force pulse generated from the realistic impact events agree well

with the product of mass and the velocity for the range of impact scenarios. This will

ensure the accuracy of the test data and eliminate the uncertainty of the amplitude from

the Force-Time history diagram for that particular velocity range which is specified

differently by various authorities (Vrouwenvelder 2000). This will also enable

accurate prediction of the vulnerability of the column for the respective vehicle impact.

In fact, force time history data generated from vehicle impact with rigid barriers are

always conservative for the vulnerability assessment of deformable bodies such as

concrete columns. Therefore in the absence of realistic numerical models of vehicles,

the vehicle column interaction can be conservatively regenerated by using the force

time history diagram resulting from the rigid barrier-vehicle impact events. This

observation is also effective in determining the basic pulse characteristic in the

absence of precise numerical methods to quantify the collision pulses. Based on these

results, duration of the typical impact pulses can be assumed to be 100ms on average

as they belong to a group of vehicles with various masses, stiffness and velocities. On

the other hand it is worth to note that frontal stiffness of a vehicle must be satisfied

both passenger safety and better driving performance. Therefore the practical range of

the frontal stiffness is cannot be changed dramatically. Therefore average impact

duration of around 100ms may also be applicable to a new generation vehicles. These

force pulses will also be used for comparison purposes once the collapse load of the

column is determined.


4-103

4.2.3 Impact pulse modelling and effects of the impact pulse parameters

Effects of the pulse shapes are investigated in this subsection. Figure 4.4 shows the

various pulses that lead to equivalent damage (or Iso-damage) conditions. Triangular

(1.0875MN) and Haversine (1.080MN) failure amplitudes are almost equal and the

square pulse with equivalent ramp up duration is the one that causes the same damage

with minimum peak. The observed impact behaviour is more sensitive to the peak

impact force than the associated impulse similar to quasi-static loading. It seems that

the impact force is located close to the quasi static region. The strain rate sensitivity of

the impact can be examined by reducing the slope of the iso-damage functions.

Figure 4.4: Iso-damage pulses

Figure 4.5: Effects of the strain rate or frontal stiffness

When the slope of the ramp function is increased, the column fails at a lower amplitude

(see Fig. 4.5) even though the triangular force pulses implied the opposite. In fact,


4-104

inertia effects and the strain rate effects are the main parameters that govern the failure

as the damping and deformation characteristics have negligible effects on energy

dissipation during the impact. The inertia effects are more predominant than the strain

rate effects for rectangular pulses in the quasi-static region (see Fig. 4.5) and the

opposite is true for triangular pulses. As the resultant variations are negligible, any of

these characteristic curves for the vehicle impact generated force history space can be

used to define a force-impulse diagram for the impacted columns when the structural

damage is controlled by the shear capacity of the section. Therefore, in this research

only the triangular pulse is used to reconstruct the vehicle impact due to its simplicity.

The peak force has been varied until the columns reach near ultimate stage while

keeping the duration constant in order to account for the various masses and velocities.

A comprehensive discussion of the duration will be provided at a later stage.

4.2.4 Simulation of impact of axially loaded columns in medium rise buildings

Impact capacity of typical columns of five to twenty storied buildings made of 50MPa

concrete was investigated by using comprehensive numerical simulations. The

columns support 6m spanning slabs in each direction, subject to 3kN/m2 imposed (live)

load identical to the design load capacity of an office building, classrooms or lecture

theatres at each floor level (AS3600 2004). Vulnerability of a ground floor column was

assessed for a typical frontal collision of light weight vehicles such as cars or vans. The

structural design is based on the Australian Standard AS3600 (2004). The optimum

column diameters are rounded to the nearest 50mm and consequently, to facilitate ease

of comparison of various columns made of one particular concrete grade, the axial

stresses on the columns are maintained constant, rather than the axial load. This has

lead to smoother graphs at the later stage. In addition, two alternative design options

with two different steel ratios were considered and Figure 4.6 represents the cross

sections of the selected ground level columns supporting different number of stories.

The design axial load capacity Pd was calculated as,

( ) 6.085.0 ' ×+= ysccd fAAfP Eq. 4.1

Where 'cf is the compressive strength of concrete, Ac is the net area of concrete, As is

the area of steel and fy is the yield strength of steel. Even though this analysis was


4-105

conducted by assuming 500MPa for longitudinal steel, the results can be extended to

other steel grades by using the equivalent steel area method proposed by Shi et al.

(2009). The spacing between the longitudinal steel bars was kept close to 100mm as

much as possible even though it inevitably varied from 70mm to 110mm depending on

the diameter and steel ratio of the columns. No further effort has been taken to

minimise the distance between the bars as it violates the general procedures used in

practice.

Figure 4.6: Cross sectional areas of the circular Figure 4.7: Support conditions and external

concrete columns loads applications

As the entire structure did not have enough time to react under impulsive loading (El

Tawil et al. 2005), fixed support conditions were assumed in this analysis as shown in

Figure 4.7. The translations as well as rotations of all the nodes in all directions were

restrained at the bottom. However, only the outer vertical faces of the column head

were constrained against horizontal movements (ie. X and Y directions) in order to

permit axial shortening during the impact. The restrained height was maintained equal

to the diameter of the column. If this condition is satisfied, the support condition can

be treated as fixed (BS8110 1985). Based on a convergence study, 25mm×25mm

quadratic solid elements with one point integration were used for concrete while

25mm long beam elements with 2×2 Gauss integration were used for reinforcement.

Hourglass control is mandatory with reduced integrated elements, and the method

described by Flanagan and Belytschko (1981) was implemented in the hourglass


4-106

control. A detailed description about the selection can be found in (Schwer et al. 2005).

Moreover, a 35mm cover was provided for the reinforcements and at least two cubic

elements were used across the thickness.

Figure 4.8: Plan view of the half models

Quadratic solid elements were used at the core concrete zone (see Fig. 4.8) as the

wedge element has caused stability problems under severe deformations including

element interlocking. Obviously, the number of elements generated inside the

rectangular segment is quite large and hence, stability of this model is high when it

undergoes large deformations. These quadratic solid elements are the smallest

elements in the entire model and the time step size has therefore been governed by the

size of these elements.

4.3 Impact behaviour of the columns and possible damage modes

Once a triangular pulse was applied, the displacement of the column increased

simultaneously with the impulsive load until its peak is reached. Then the

displacement was decreased with several minor peaks in the post peak region until the

residual displacement was achieved. This behaviour accounted for the axial load

acting on the column, which has developed second order bending effects. In an actual

impact event there could be some contact losses due to the relative movements of the

bodies in this region as the speed of the deformation of the column exceeds the

velocity of the vehicle even though they are moving in the same direction. According

to further investigations a rapid change of residual deflection occurred with a small

increment of the impact force at a later stage. Consequently the permanent damage to

the column can be identified by the continuous increment of the residual deflection

without recovering.

However, failure due to vehicle impact varies from the usual flexural type of failures

during a mid span impact and hence, a conventional hypothesis based on the energy

absorption capacity of the column may not be applicable as the energy absorption


4-107

characteristics mainly depend on the flexural deformation of the column. Since the

column has not been subjected to flexural deformations, a small portion close to the

impacted region has undergone highly localised stress and has absorbed an excessive

amount of energy. This localised stress has exceeded the crushing stress of the

concrete and hence may fail abruptly during the impact. This will considerably reduce

the effective area of the column and the resultant eccentric axial load has finally

diminished the axial load carrying capacity of the column. Under these circumstances,

the column has failed due to shear failure initially and subsequently by flexural failure

leading to collapse. The observed failure modes can be categorised as shear or shear

flexural type of failures depending on the test variables as observed during the many

simulations.

Accordingly, the strain rate effects are more predominant in the vicinity of the impact

and hence localised effects on the dynamic material characteristics could be significant.

However the resultant overall capacity enhancement seems to be insignificant as the

average strain rate was only marginal. For instance, the standard definition of the strain

rate, tδεδε =& can also be expressed as

Lt

l

L

νδδε

&& == 1 , where L is the original length of the

element (column) and v& is the rate of deformation. This definition generates a much

broader view of the strain rate effects while allowing the average strain rate to be

calculated. According to the Feyerabend (1988) column test, the maximum axial

deformation of the column is around 1.3mm and the duration of the impact is

approximately 6ms. This leads to an average strain rate of approximately 5.4×10-5s-1.

As far as medium velocity vehicle impacts are concerned, the duration of the impact is

around 100 ms and hence the average strain rate is well below 0.1s-1.

4.4 Vulnerability prediction

The analyses include 300mm to 500mm diameter axially loaded columns which are

adequate in capacity for five to ten story buildings with two different steel ratios as

shown in Figure 4.6. The critical impact pulses for each column are given in Figure

4.9 along with the collision force-time histories of some cars - Honda Accord, Ford

Taurus & Renault Fuego which are in frontal impacts similar to the one that shown in


4-108

Figure 4.10. Calculations based on the areas under the curves show that the axially

loaded five-story building column design for gravity loads cannot withstand impact

velocities more than 15ms-1 or 40 km/h (see Fig. 4.9). This velocity range is common

in urban areas and hence axially loaded column having diameters equal to or less than

340mm are likely to collapse under medium velocity car impacts. Though statement

cannot be generalised, this method will allow prediction of the vulnerability of the

impacted columns against new generation of vehicles for most common modes of

impacts and hence provide a common base for comparison purposes.

Figure 4.9: Comparison of impact capacities of columns with full scale crash tests (NHTSA)

Figure 4.10: Honda Accord in a frontal collision at a speed of 48.3km/h (NHTSA)

If the impact duration remains constant, the peak force of the tri-angular pulse will

determine the severity of the impact due to the negligible effects of the slope of its legs

(see Fig. 4.5). This observation implies that the peak forces can be interpolated to

quantify severity of an impact. Based on this argument, it can be concluded that


4-109

collision severity can be predicted by interpolating a known collision pulse as the

influence of the slope of triangular pulse (or the strain rate effect) is negligible.

Comparison of the maximum impact capacities of impacted columns up to

twenty-story buildings in terms of Force-Time history is given in Figure 4.11.

According to the Figure 4.9, the maximum impact force applied by medium velocity

car impact is around 675kN. Therefore it is evident that axially loaded columns made

of 50MPa concrete in ten-story and above will not susceptible to collapse under

medium velocity car impact.

Figure 4.11: Ultimate capacities of columns Figure 4.12: Support reaction and Impact pulse

Figure 4.12 represents the support reaction and the applied lateral impact pulse on the

column during an impact simulation. The reaction forces are nearly equal to the

applied impact pulse as inertia effects and damping effects are less pronounced under

shear failure conditions. By definition, for ‘hard impact’ it is assumed that the structure

is rigid and immovable and that the colliding object deformed linearly during the

impact phase (EN 1991-1-7: 2006). Based on that assumption an equation was derived

in EN 1991-1-7 (2006) to calculate the maximum resulting dynamic interaction force

applied on the impacted column. However, the aforementioned difference actually

represents the probable error that might occur in interaction force determination based

on the undeformable impacted body assumption. Therefore this assumption may only

be true for the shear failure predominant columns. Consequently the analysis based on

this assumption may need amplification factor which can account for the inertia and

damping effects under flexural failure predominant column. Therefore undeformable

body assumption used in EN 1991-1-7 (2006) is conservative for flexural failures

predominant columns.


4-110

4.5 Bending moment and resultant shear due to impact

Variation of the resultant bending moment due to impact was measured at three

different locations along the height of the column in order to identify the impact

behaviour of the column. Variation of the resultant bending moment due to impact can

be measured using ‘SECFORC’ in LS-DYNA code (LS-DYNA 2007). The ‘SET’

option under the ‘SECFORC’ uses a set of nodes to define a cross section. Forces from

the elements belonging to the node set are summed up to form the moment in this

method and hence interconnected elements belonging to only one side of the nodes

should be defined at one particular section.

The measured moments at each cross section (CS) are given in Figure 4.13 with the

relevant distances measured from the bottom. The highest moments are generated in

the vicinity of the impact (0.975m) and close to the support (0.100m), but with

opposite signs. The moment is gradually reduced away from the bottom support

beyond the point of the impact. However another directional change can be seen close

to the top support (3.9m) which signifies the generation of third order vibration mode

in the impacted column. Consequently, excessive shear forces are generated at the

contraflexure points located close to the supports. This observation may be cited as a

potential reason for the failure of the column shown in Figure 4.14 which indicates

typical shear critical situation. It is also important to note that laps forming in this

region worsen the consequences. Thus, conventional design and detailing practices,

which lead to impact damage, need modification. Laps should be avoided and

transverse reinforcement should be provided close to the support where shear strength

is vital for survival.

Figure 4.13: Resultant bending moments Figure 4.14: Damaged column under vehicle impact


4-111

4.6 Effects of the diameter of the column, concrete grade and steel ratio

The impact capacities of the columns supporting different number of storeys and the

capacity variation with the grade of concrete are shown in Figure 4.15 along with two

steel ratios. In fact, this plot is based on the assumption that one particular structure

with certain number of stories constructed by using different concrete grades. In other

words, variation of concrete grade and steel ratio will automatically change the

diameter of the column to resist the constant axial loading. The concrete grade and

steel ratio have profound effects on the impact capacity of the columns. For instance,

the impact capacity of a column made of lower concrete grade is higher compacted to

an equivalent (capacity) column made of higher concrete grade. Conversely,

equivalent column with a high longitudinal steel ratio becomes more vulnerable to the

impact loads. The diameter of the column has to be increased as the steel ratio

decreases to maintain the same amount of axial load and the capacity enhancement is

therefore partially due to the increase in the diameter of the column.

Figure 4.15: Effects of the diameter of the column Figure 4.16: Effects of the concrete grade

Comparative average increments of the impact capacities, with respect to the Grade

50MPa concrete columns with 4% steel, are given in Figure 4.16. The area in between

the two curves shows the range of capacity increment that can be achieved by

changing the longitudinal steel ratio and the concrete grade. The diagram clearly

exhibits the influence of the steel ratio and concrete grade while emphasising the

possibility of improving the impact capacity up to 2.4 times by selecting various

combinations of concrete grades and longitudinal steel. The rate of capacity

enhancement of columns with 1% steel towards the 30MPa concrete reflects the

greater influence of the diameter of columns made of lower concrete grades and the


4-112

resultant shear capacity enhancement. Diameter of the column reduced with the

increment of the steel ratio and hence the columns having higher steel ratios become

increasingly vulnerable. In other words the dynamic impact capacity of columns

mainly depends on the area of the concrete, and conversely the influence of

longitudinal steel is not so significant compared to the behaviour under static loading

conditions. This observation confirms that there is not enough time to develop

extensive bond failure along the length to activate the steel under the impact

conditions, particularly for circular columns.

4.7 Effects on the slenderness ratio

Figure 4.17: Effect of the slenderness ratio Figure 4.18: 500mm column with 4% steel

The effectiveness of slenderness ratio is investigated in Figure 4.17. As the

slenderness ratio decreases, the failure plane will change its inclination. Consequently,

the diagonal cracks are formed between the support and the point where maximum

moment is reached followed by initial tension cracks (see Fig. 4.18). This change will

increase the fracture energy dissipation through the cracked surface while increasing

the number of effective links in preventing the crack propagation. Thus, noticeable

50% increase in the critical (collapsed) peak force can be achieved on average. Apart

from that, improvements are increased with the columns having larger diameters (see

Fig. 4.17). As far as the effect of boundary conditions are concerned, there is no

significant contribution from the supporting conditions to resist the impact force.

However, slight increments can be seen when the slenderness ratio is higher and the

difference will be decreased with the enhancement of the diameter.

2 m

Crack formation


4-113

4.8 Energy absorption due to the impact

Energy absorption will primarily depend on the failure mode which in turn governed

by the specific parameters such as slenderness ratio, steel ratio, concrete grade, contact

interface parameters etc. However, energy consumption for plasticity and nonlinear

deformation are comparatively low under impact conditions. Due to the localised

damages and the micro-cracks, most of the energy dissipation will takes place closer to

the contact interface. Therefore, fracture toughness of the concrete gained the highest

consideration among the parameters that influence the energy dissipation of concrete

rather that the tensile strength. Consequently, the energy absorbed by the concrete is

comparatively low and the most of the internal energy is stored as strain energy in the

longitudinal steel.

4.9 Effects of impact duration

Figure 4.19: Equivalent impulse diagrams Figure 4.20: Iso-damage pulses for 600mm column

Figure 4.19 exhibits the impact characteristics of a 600mm diameter column with 4%

steel. The column was subjected to equivalent impulses (F∆t = mv = constant) with

different durations (∆t) and the 100ms impact represents the critical pulse in which the

column reached near failure conditions. In fact, these impulses replicate the cars with

constant mass (m) and velocity (v) but with different frontal stiffness. Numerical

results revealed that the column failed under 50ms impact yet remained unaffected

under 150ms impact. This observation confirmed the importance of the frontal

stiffness characteristics of the vehicle that govern the impact behaviour of columns. At

the next stage, the peak forces of the triangular pulses are adjusted until the

iso-damage (or equivalent damage) at near failure conditions are achieved as shown in


114

Figure 4.20. This proves that the column is highly sensitive to the amplitude of the

impact and not to the associated impulse (Shi et al. 2008). Thus, the vehicle impact

generated forces are theoretically closer to the quasi-static loading category where the

response becomes somewhat insensitive to impulse (Shi et al. 2008). This comment is

applicable to all the other column diameters varying from 300mm to 750mm.

Another observation is that the impact pulses and resultant support reactions are nearly

equal in terms of shape, duration and amplitude as shown in Figure 4.21. According to

Baker et al. (1983), if the load and response functions terminate simultaneously, such a

scenario can be included in the dynamic region which is between the impulsive and

quasi-static regions. However, as the average strain rate was shown to be well below

0.1s-1, it can be concluded that the strain rate effects are less pronounced in the vehicle

impact analysis.

Figure 4.21: Force vs reaction histories for pulses with different durations

Figure 4.22: A typical Pressure impulse curve Figure 4.23: Iso-damage contours for impact

By considering all these factors, it is suggested that the vehicle impacts belong to a

specific domain in the pressure impulse diagram (see Fig. 4.22). To investigate this


4-115

hypothesis, force and the relative impulses that lead to the iso-damage conditions are

plotted in a log scale for different column sizes (see Fig. 4.23). In fact, the curve

reasonably agrees with the typical shape of the Pressure-Impulse diagram which is

commonly adopted to predict the structural damage under blast loading conditions.

Hence, the impact capacity (in terms of amplitude P, impulse I) can be expressed by

the following mathematical expression which is valid between the dynamic and the

quasi static regions.

( )( )B

cccc

IPAIIPP

+=−− 22

Eq. 4.2

In the above equation, Pc and Ic represent a known force and impulse pair located on

the iso-damage curve, A and B are constants to be determined based on the shape of the

iso-damage contours. Pc and Ic can be expressed in terms of the column diameter,

concrete grade, steel ratio, slenderness ratio and hoop spacing. A parametric study and

the derivation of equations for the impacted column based on this observation will be

presented in the following chapters. Polynomial equations will be provided for more

accurate estimations along with linear equations for approximate predictions.

4.10 Conclusions

Numerical model of an axially loaded column subjected to transverse impact loads has

been validated and extended to simulate the behaviour of vehicle impacted columns.

This analysis has confirmed the feasibility of using numerical simulation techniques in

vulnerability assessment of impacted columns while enhancing our understanding of

the impact behaviour of columns under vehicular impacts. The main outcomes and

findings of the investigations are summarised below.

1. Impact pulse generated from a typical car to rigid barrier impact is successfully

used to reconstruct the vehicle collision. This method can be used as the foundation

to generate a data base which can be used to determine the vulnerability of column

against most common modes of collision. Additionally, the rigid barrier-vehicle

impact data are conservative in the impact reconstruction due to the low

deformation of the column under shear predominant conditions.

2. If the duration of impact and peak force remain identical, the effects of the shape of


4-116

the pulses are insignificant in vulnerability assessment. Strain rate also has a minor

effect if the pulse characteristics belonging to the vehicle impact generated force

history space. Inertia effects are more predominant than the strain rate effects for

the rectangular pulses and the opposite is true for triangular pulses even though the

comparative advantages are insignificant. Hence triangular pulses can be

effectively used for impact simulations and the average duration can be taken as

100ms. In addition, collision severity can be predicted by interpolating a known

collision pulse as the strain rate effect is negligible.

3. The generated results will be more reliable if the peak force occurs at a time of

25ms or more, which is derived by assuming 50ms triangular impact pulses with

equal legs. If this condition can be satisfied for particular impact duration, the

arrangement of the mechanical components of the impacting vehicle is immaterial

as the effects of the shape of the impact pulses are insignificant.

4. The analysis also revealed typical columns adequate in capacity for five-storey

buildings made of 30MPa to 50MPa concrete with 1% to 4% steel are vulnerable to

medium velocity car impacts. However at the design stage, impact capacity of the

columns can be increased by 20% by selecting the alternative design method with

the low amount of steel.

5. Resulting bending moment revealed the generation of third mode of vibration in the

impacted column and generation of maximum shear forces at contra flexure points

close to the supports. Consequently, the impacted column may tend to fail close to

the supports under shear critical conditions. Thus, laps should be avoided close to

the supports and maximum transverse reinforcement should be provided in the

vulnerable regions to avoid shear failures.

6. The impacts of columns treated herein are theoretically close to the quasi-static

loading region where the response becomes less sensitive to impulse but more

sensitive to the peak force. Iso-damage curves of the impacted columns will follow

the shape of the conventional pressure impulse diagrams under blast loading

conditions. Hence analytical equations can be derived to quantify the impact

effects.


5-117

5. CAPACITY OF THE AXIALLY LOADED COLUMNS UNDER LATERAL IMPACTS

5.1 Introduction

Structural columns are seldom designed for vehicle impacts due to inadequacies of

design guidelines. This chapter presents an extensive numerical study on the response

of exposed concrete columns in multi-storey building to vehicle impacts with a focus

on evaluating their impact capacity. Due to the low elevation impact conditions, most

of the impacted columns failed in flexural shear mode. Localised stress concentration

exceeded the compressive and tensile capacities of the concrete at the contact interface

and diminished the strength of cover concrete by reducing the effective area of the

column. Under this circumstance, the columns failed due to shear initially and flexure

subsequently, causing collapse. Based on the above observations, it is expected that the

parameters that govern flexural shear type failures under quasi-static loading

conditions can be effectively used to mitigate the damage under impact conditions. A

comparative analysis revealed that the flexural-shear failure conditions of the columns

are well predicted by the AASHTO equations for bridge piers, ATC/MCTTER

equations, equations by Ascheim and Moehle, and equations by Priestly et al. (Lee et

al. 2003). In these equations, the total shear capacity of a column was calculated by

adding together the contribution of the steel and concrete. It can be seen that concrete

grade, steel area, diameter of the column, hoop spacing, area and yield strength of

hoop are the key design parameters and therefore the effect of these parameters are

further investigated in detail. As analyses have been conducted in the previous

chapters by assuming nominal confined conditions, the effects of the confinement are

yet to be determined.

Another aspect of was to determine the vulnerability of the structural column during

the construction process when the applied load is only a portion of the design load and

hence the shear capacity and the stiffness have not reached their full potential (Abrams

1987; Zeinoddini et al. 2008). Proper impact damage assessment is vital to determine

whether the column has to be replaced or can be repaired for further use. Design

guidelines developed on partially loaded columns subjected to earthquake (Esmaeily


5-118

and Xiao 2004) and blast (Shi et al. 2008) loading may not be adequate in this

circumstance where mode of failure, strain rate effects and inertia effects are

substantially different. Moreover, a decision on the portion of total load that can be

allowed during the rehabilitation process has to be made. Proper damage assessment

will also minimise the risk to rescue workers and those who enter into the building

following an impact, or when the affected bridge structure has to be used as a vital

supply line.

5.2 Design against accidental loads

As the low elevation impacts initiate the shear failure conditions, the limit states

formulated based on the kinetic energy or deformation capacity (EN 1991-1-7: 2006)

may not be adequate. In the absence of specific guidelines for accidental design,

Accidental Limit State (ALS) design was considered as an alternate solution. In

particular, the ALS mainly used for floating and fixed offshore structures made of steel

which are subjected to Ship Collisions and Explosions (DNV-RP-C204). Possible

extension of the same philosophy for the concrete columns is discussed here in brief.

In the process, the inherent uncertainties associated with the frequency and magnitude

of the accidental loads should be determined in advance. To demonstrate the concept

of Accidental Limit State (ALS) a column with fixed geometric properties can be

taken in to consideration. Then the Design resistance (Rd) is a constant and allowable

design load (Sd) can be varied depending on the particular loading. The requirement to

be fulfilled in the Accidental Limit States may be written as;

dd RS ≤ , Eq : 5.1

where, fkd SS γ= = Design load effect, m

kd

RR

γ= = Design resistance, Sk is the

characteristic load effect, γf is the partial factor for loads, Rk is the Characteristic

resistance and γm is the Material factor. For ALS design, the load and material factor

should be taken as 1.0 (DNV-RP-C204). Thus it seems to be closely related to the

serviceability limit state (SLS). However, it should be in conjunction with the factors

involved in the safety evaluations (Moan 2009). For instance, substantial residual

capacity should remain in the impacted columns to avoid damage to the supported

structure or vital component to be used in rescue missions. Consequently, the structure


5-119

should be checked for all relevant limit states and in particular accidental limit state

can be declared in-between the serviceability and ultimate limit states depending on

the requirement.

5.3 Parametric studies of impacted columns

5.3.1 Finite element analysis of confined circular columns

The effects of the confinement as a damage mitigation technique were investigated at

the initial stage of the analysis. The simulation was conducted by assigning

unconfined material characteristics to the cover concrete and confined material

characteristics to the core concrete to account for the various confinement stresses due

to the hoop spacing, hoop diameter and steel grades. The stress-strain model

developed by Mander et al. (1988) was used in the simulation to account for the

confinement effects. In particular, this equation expresses the ratio of the compressive

strength of the confined concrete 'ccf to the compressive strength of unconfined

concrete 'cof , for a section with equal effective lateral confining stress in each

direction.

5.3.2 Effects of the confinement

(a) Hoop distribution (b) Resultant strain variation in the column

Figure 5.1: Effects of confinement under lateral impacts


5-120

According to the Liu and Foster (1998) early cover spalling in confined high strength

concrete is caused by the tri-axial state of stress at the cover-core interface and high

tensile strength is observed between cover-core interfaces under low axial loading

conditioned. To provide an explanation for similar observations under impact loading

conditions where cover concrete spall off quickly, the 12mm spirals are placed at

100mm spacing in 300mm column as shown in Fig. 5.1(a). This arrangement

generates higher confinement stresses in the core concrete and hence the resultant

stresses in core and cover concrete vary in all directions. The extensive lateral stresses

generated at the core-cover interface has measured and it has exceeded the tensile

strength of concrete during impact (see Fig. 5.1(b)). This is one of the reasons why

concrete cover fails abruptly during impact. However there is no excessive stress

difference under the serviceability conditions which represented by the stresses up to

20ms where the impact begins (see Fig. 5.2). The cover concrete in cracked sections

did not contribute to the axial capacity of the concrete column at the residual stage. In

fact, these sections are the weakest sections which govern the residual capacity of the

concrete columns. However the stress difference between cover and core concrete

gradually reduced towards the top support as the stress in the vertical direction did not

excessively increase in those sections due to impact as shown in (see Fig. 5.1(b)). This

observation also supports the argument that sufficient stress must apply on the core

concrete in order to activate the confinement effects.

Stress diffrence in lateral direction

-6

-5

-4

-3

-2

-1

0

1

0 0.05 0.1 0.15 0.2

Time (s)

Stre

ss(N

/mm

2 )

Stress diffrence

Figure 5.2: Stress difference at the cover-core interface

5.4 Effectiveness of confinement under impact loads

In general, transverse reinforcement confines the compressed concrete and prevents

premature buckling of compressed longitudinal bars while acting as shear


5-121

reinforcement. One of the objectives of this investigation is to quantify the

enhancement of the impact capacity due to the transverse reinforcement present in the

vulnerable region of the impacted column. In particular, the effects of the hoop spacing

s, diameter d and yield strength 'syf are investigated in this section for columns made

of 30MPa and 50MPa concrete. 6mm hoops with 350MPa yield strength at 250mm

spacing are used as the datum of the analysis as this arrangement will not be effective

for improving the confinement effects. The diameter of the hoops is varied from 6mm

to 12mm while spacing is varied from 50mm to 250mm. In fact, Hwee and Rangan

(1990) as well as many others are varied the lateral steel distribution within this range

in their experiments. The maximum and minimum yield strength of the hoops are

limited to 500MPa and 250MPa respectively.

Confinement effects are particularly effective when the hoops are below 100mm

spacing. It was observed that for the 350mm diameter columns made of 30MPa

concrete, nearly 32% improvement can be achieved by reducing the hoop spacing to

100mm and it can be further improved up to 44% by reducing the spacing to 50mm

with 6mm diameter bars. However this enhancement is limited to 10% for 900mm

diameter columns due to the lack of confinement stress. This indicates that

confinement has to be increased with the diameter of the column by a suitable means

to achieve the same level of capacity enhancement.

As far as the effectiveness of the individual parameters is concerned, the vulnerability

of the column reduces as the hoop spacing decreases and as the hoop diameter and

yield strength increase. In other words, individual or collective use of these parameters

is possible to enhance the impact capacity of the columns. In particular, provision of

hoops with a larger diameter is more effective for damage mitigation compared to

reducing the hoop spacing for the range of values under consideration. This

observation is effective when the hoop spacing cannot be reduced due to practical

issues. However, the capacity enhancement due to the increment of yield strength of

the hoops is negligible. In fact, yielding of the lateral reinforcement depends on the

axial load, and hence provision of hoops having higher yield strength may not be

effective under serviceability loading conditions (Janke et al. 2009). As a whole, the

highest confinement stress is gained by the small diameter columns and hence their


5-122

effectiveness is more pronounced.

ACI-ASCE Committee 441 (1997) recommended the ratio csys ff ''ρ , referred to as the

confinement index, to evaluate the effectiveness of the confinement. However, it was

revealed that the confinement index cannot capture the effects of yield strength of the

confinement steel, particularly under P-∆ effect (Paultre et al. 2001). Consequently

another confinement index, which can also account for the distribution of confinement

steel and yield strength is introduced as,

'

'

co

le f

fI = Eq: 5.2

where 'lf

is the effective lateral confining stress and 'cof

is the unconfined strength of

the concrete. A detailed description can be found in Mander et al. (1988). Paultre et al.

(2001) also demonstrated that this method was very efficient in predicting the

effectiveness of yield strength of lateral steel.

Impact capacity increment vs f' l /f' co

1.0

1.1

1.2

1.3

1.4

1.5

1.6

0.00 0.10 0.20 0.30f' l /f' co

Cap

acity

incr

emen

t

350mm; G30 500mm; G30 600mm; G30700mm; G30 900mm; G30

Figure 5.3: Capacity enhancement for 30MPa concrete

The capacity enhancements with 30MPa and 50MPa concrete are shown in Figure 5.3

and 5.4 respectively for the range of parameters under consideration. It is evident that

columns made of a lower grade of concretes gain the highest capacity enhancement

due to confinement compared to those with a higher concrete grade. Flexural shear

failure characteristics of columns made with 30MPa concrete leads to an additional

capacity enhancement as the column diameter reduces. This is because flexural

characteristics become more predominant as the column diameter decreases. However,

almost all the columns made of 50MPa concrete show shear failure characteristics

where the impact capacity is based entirely on shear strength of the columns. Hence,


5-123

the capacity increases proportionately with the diameter of the column and hence there

is no proportional change of the capacity increment with the diameter of the column as

shown in Figure 5.4. Consequently the shear capacity increment is responsible for

observed behaviour of the columns with larger diameters while tensile strength

increment is responsible for the capacity improvement in the 300mm column.

Impact capacity increment vs f' l /f' co

1.00

1.05

1.10

1.15

1.20

1.25

1.30

0.00 0.05 0.10 0.15 0.20f' l /f' co

Cap

acity

incr

emen

t

300mm; G50 600mm; G50750mm; G50 800mm; G50

Figure 5.4: Capacity enhancement for 50MPa Concrete

As the effectiveness of the confining stress is further reduced with the increase in

concrete grade, the columns made of higher grade concrete need an even higher

confining stress compared to those with lower grade concrete. For instance,

improvements due to the reduction of hoops spacing down to 100mm is not effective

for 50MPa concrete and the increment is nearly 4%. Under these circumstances the

hoop spacing needed to achieve the required level of confinement under impact

loading conditions may be highly underestimated by the general provisions of the

codes (BS8110 1985; AS3600 2004), which are based on the maximum diameter of

the longitudinal steel.

Moreover, diameter of the hoops also has a profound effect on the impact capacity of

the columns. Impact capacity can be improved by 12% and 22% on average by

increasing the diameter of hoops from 6mm to 8mm and 12mm respectively.

Consequently overall 35% improvement can be made if the 12mm hoops with 50mm

spacing are selected for 300mm columns. On the other hand yield strength of the

lateral reinforcement has a similar effect. The investigation is conducted by assuming

hoops having 350N/mm2 yield strength and it is observes that almost 10% increment

of the impact capacity can be achieved with 500MPa steel. Hence the final capacity


5-124

can be increased up to 49% if 12mm bars with 500MPa placed at 50mm spacing. Thus

the diameters of the hoops and the hoop spacing have the most profound effects on

enhancement of the impact capacity of the columns made of higher grade concrete.

These conclusions were made based on the assumption that hoop spacing is constant

throughout the column.

5.5 Effects of the unconfined cover and use of external wrapping

(a) 30MPa concrete (b) 50MPa concrete

Figure 5.5: Confined strength for different concrete grades

In general, compressive strength increment due to the confinement effects can be

expressed in term of concrete grade as the hoop spacing, diameter of the hoops and

steel grade will finally contribute to increase the compressive strength of the core

concrete. A point on graphs in Figures 5.5(a) & 5.5(b) represents the improved grade

of core concrete that can be achieved by changing the confinement effect alone. It can

be observed that the rate of the change of enhancement is more pronounced for

columns with larger diameter even though the range of variation of the grade of

concrete not as much as widespread as for smaller diameter columns. These graphs can

also be used to estimate the impact capacities of the columns at an intermediate

combination which is not covered by the selected spacing and bar diameters in the

present analyses. The 600mm diameter column is common for both the concrete

grades and comparison of the impact capacity at a particular concrete grade would

indicates the strength reduction caused by the unconfined concrete properties assigned

to the cover concrete. For instance, comparison of the impact capacities for confined

compressive strength of 50MPa concrete indicates that the cover concrete can enhance

the impact capacity around 13%. The contribution of the cover concrete may increase


5-125

around 20% for confined compressive strength of 60MPa concrete. This means that

there is a potential for external wrapping can be used to enhance the impact capacity of

the confined concrete columns significantly where full cross section experiences a

tri-axial state of stresses.

5.6 Effects of the slenderness ratio on capacity enhancement

Figure 4.17 of the Chapter 4 shows the impact capacity (in terms of peak force) for the

columns made of 50MPa concrete with 4% steel and nominal hoop spacing. As the

slenderness ratio decreases, the shear failure plane increases its inclination.

Consequently, this change will increase the fracture energy dissipation through the

cracked surface while increasing the number of effective hoops in preventing crack

propagation. Hence, the investigation continued to examine the impact behaviour of

columns with 50mm hoops spacing and 6mm bars.

Peak force vs Slenderness ratio

0

1

2

3

4

5

6

7

3 6 9 12 15Slenderness ratio (L/D)

Fo

rce

(MN

)

300mm Col 450mm Col500mm Col. 600mm Col.300mm Col. Uncon. 450mm Col. Uncon.500mm Col. Uncon. 600mm Col. Uncon.

Figure 5.6: Columns confined with 12mm links at 100mm spacing

Capacity enhancement due to the confinement effect is shown in Figure 5.6 along with

the capacities of the columns with nominal confinement. The enhancement is more

predominant in larger diameter columns and the average increment of 38% is observed

except for the 300mm column which shows a lesser increment of about 8 to 20%

depending on the effective height. Therefore, the confinement effects may not provide

substantial extra capacity particularly for the 300mm diameter short columns. On the

other hand, 450mm and 500mm diameter columns changed the mode of failure from

flexural-shear to flexure, even for low slenderness ratios, by enhancing the capacity

due to the confinement. However, the failure mode of 600mm column remains


5-126

unchanged and hence 500mm is the limiting diameter which changes the response due

to the confinement effects.

5.7 Comparison of the dynamic and static shear capacities

Dynamic shear capacity enhancement

1.5

2.0

2.5

3.0

3.5

4.0

4.5

0.30 0.40 0.50 0.60 0.70 0.80

Diameter (mm)

Vd/

φVs

G30 100c/c G30 50c/c G50 100c/c G50 50c/c

Figure 5.7: Comparison of the dynamic and static shear capacities

The shear capacity enhancement under dynamic loading condition is compared in

Figure 5.7. The design static shear capacity (φ'Vs) calculated by using AASHTO

specification for circular bridge piers where the strength reduction factor φ' can be

taken as 0.85 (AASHTO-LRFD 2002). Fully loaded columns of 30MPa and 50MPa

concrete with 50mm and 100mm hoop spacing were selected in the analyses. The

dynamic shear capacity (Vd) was equal to the peak force (collapse) of a 100ms

triangular impulse. It can be observed that the dynamic to static shear capacity ratio

varies in between 1.6 and 4.2 depending on the diameter and concrete grade of the

columns. This clearly indicates that the static shear capacity is not an indication of the

maximum allowable (peak) force during an impact, even though the peak force has

some correlation with the static shear capacity of the column. If the correlation factor

is known, the static shear capacity can be used for approximate vulnerability

assessment. According to these results the dynamic amplification factor ϕdyn (varies

from 1.0 to 2.0) given in EN 1991-1-7:2006 highly under estimate the dynamic impact

capacity of the larger diameter columns. In general, the columns made of lower grade

concrete have the highest dynamic capacity enhancement due to their flexural-shear

failure characteristics.


5-127

5.8 Impact capacity of partially loaded circular columns

5.8.1 Introduction

Axial load is one of the key design parameter in the analysis and substantial discussion

is therefore provided on impact behaviour of partially loaded columns.

Comprehensive understanding of the status of the damage to an impacted column is

vital for prevention of progressive collapse of the supporting structure as well as to

determine whether the column has to be replaced or can be repaired for further use.

Moreover, a decision on the portion of total load that can be allowed during the

rehabilitation process has to be made. Proper damage assessment will also minimise

the risk to rescue workers and those who enter into the building following an impact,

or when the affected bridge structure has to be used as a vital supply line. Each of these

decisions has to be made based on the residual capacity of the column after an impact.

Current knowledge on damage assessment is incomplete and major decisions are made

based on personal experience.

Figure 5.8: Rehabilitation of a bridge after catastrophic failure of a column

It was evident that the behaviour of reinforced columns under simultaneous axial

loading and lateral impact has not been given significant consideration in the literature.

In fact, there is very limited data on the dynamic failure of pre-loaded concrete

columns subjected to lateral impacts. Bao and Li (2009) and Shi et al. (2008) have

studied the residual strength of concrete columns under blast loading. The studies

emphasised the importance of pre-loading when the effects of the damage is

determined.


5-128

To bridge the gap, a numerical analysis was conducted on partially loaded concrete

columns. This analysis included 300mm to 800mm columns which are adequate in

capacity for five to twenty storied buildings. Two steel ratios were considered in the

analysis and the effects of the hoop spacing, slenderness ratio, support fixity and load

eccentricity were extensively investigated in the parametric studies. The finite element

analysis continued to find out the residual capacity of the columns under arbitrary

impact velocities and was extended to investigate the residual capacity of the columns

under specific velocity conditions. The damage criteria used by Shi et al. (2008) was

implemented for damage assessment.

5.8.2 Damage criterion

According to the requirements it is clear that the selected damage criteria for the

impacted columns should express the residual capacity of the impacted columns in

terms of its design load capacity. Therefore the damage index D is defined as;

d

ri P

PD −=1 , Eq: 5.3

where Pr is the residual axial load carrying capacity of the damaged RC column and

Pd is the design capacity of the circular concrete columns according to the Australian

standards. As will be seen later, Pr is the ultimate on-factored residual axial load

carrying capacity. The degree of capacity degradation was defined as follows (Shi et al.

2008):

( )

.)0.18.0(

,)8.05.0(

,)5.02.0(

,2.00

collapseD

damagehighD

damagemediumD

damagelowD

i

i

i

i

−=−=−=

−=

There is no comprehensive definition given for each term. However the physical

meaning is that the supporting building is increasingly at high risk at each stage and

the column has to be replaced when the damage index is in the 0.8 - 1.0 range. The

advantage of this index is none of the commonly used damage criteria, such as residual

deflection, maximum stress and strain conditions, satisfy the above requirements. On

the other hand, the axial load capacity degradation will reflect the shear damage,

flexural damage or local damage conditions due to impact while expressing the global


5-129

behaviour of the impact, and these factors are easily obtain from numerical simulation

techniques.

5.8.3 Effect of axial load on the duration of the impact It is worth identifying the effects of the varying axial loading on the duration of the

impact, in a realistic vehicle collision, even though the impact simulation was

conducted in this thesis by applying an equivalent impact pulse of constant duration at

a specific height. The duration of the impact may depend on the percentage of the axial

load on the column as axial compression can influence the strength, stiffness, and

deformation capacity of reinforced concrete columns. In addition, it has long been

known that axial force has significant effects on stiffness (Zeinoddini, et al. 2002),

flexural strength and behaviour of reinforced concrete columns (Abrams 1987). For

instance, the rebound velocity of the striker decreases as the axial load on the column

increases, particularly for mid span impacts (Zeinoddini et al. 2002). It was also

observed that the first natural frequency decreases with an increase in the level of axial

pre-loading due to a decrease of the flexural stiffness. On the other hand, the

occurrence of plastic deformation under mid span impacts causes the impact duration

to be further extended.

Moreover, reduction of the time lag between successive impacts caused by the

impacting object as it rebounds can be observed with the increase of the axial load

level. However, the effect of the percentage of axial loading on the duration could be

negligible under low elevation vehicle impacts where the damage mode is

predominantly shear. On the other hand, the consequences due to the varying impact

duration with the percentage of axial loading are not very significant as the peak force

is the governing factor of the vulnerability under the vehicle impact. Therefore the

duration of the impact is kept constant in further analyses.

5.8.4 Simplified method to investigate the residual capacity of columns The investigation conducted by reducing the axial load and then restoring the load at

the post impact stage in consecutive steps. In fact, this will simulate the axial loads

applied during construction and at the serviceability stage on respective columns. The

initial dead weight on the column is considered as 20% of the design load in this study.


5-130

However it has been shown that columns behave in a ductile manner if the axial load is

less than 20% of the axial load capacity, particularly for HSC (Li et al. 1994; Patrick et

al. 2009). Here axial load capacity was measured as'cg fA , where Ag is the gross

concrete area and 'cf is the compressive strength of concrete. Two other initial loading

cases, namely 0.6Pd and 0.4Pd are considered in this analysis. The 60% axial load will

ensure that failure will occur by crushing the concrete above the balanced point while

40% load will tend to initiate flexural-shear failure conditions.

Staged ramp up loading

0

5

10

15

20

25

30

35

0 0.1 0.2 0.3 0.4 0.5

Time (s)

Axi

al p

ress

ure

(M

Pa)

Figure 5.9: Axial pressure application on 300mm diameter column

The initial axial load is applied gradually as a ramped up pressure on cross section and

then the impact load is applied over the front surface of the column, which takes

approximately 120ms as shown in Figure 5.9. Due to the close proximity of the impact

pulse to a quasi-static load, the column gained its residual state just after the impact.

From this time onwards a slight fluctuation of the residual deformation of the column

was seen even though its consequences on the post impact response were negligible.

Thus the axial load is applied gradually on the column in steps as separate ramped up

functions following the impact. In particular, this method avoided the explicit-implicit

transformation required in similar analyses (Shi et al. 2008). At each stage, the

capacity of the column to resist the respective axial load increments is investigated.

If the column can withstand several consecutive axial loading steps at the post impact

stage, it is an indication of considerable axial load capacity remaining in the column.

In fact, this indicates a low damage state. Thus, the analysis was conducted by

increasing the impact load gradually in steps while increasing the axial load at the post


5-131

impact stage. For example, the 20% loaded 450mm column completely collapsed

under a 1.75 MN impact (Di=0.8), and was partly damaged under a 1.7 MN impact

(Di=0.2). This is illustrated in Figure 5.10. As the difference has not much practical

significance the damage index Di, would not be sensitive as far as impacted columns

are concerned. Di=0.2 represents low damage conditions and Di=0.8 represents a near

collapse stage (Shi et al. 2008). The column failed under 1.7MN in flexure and 1.75

MN in shear critical conditions. Therefore the local failure due to impact caused the

collapse and the difference between the two critical impulses represents the energy

absorbed that causes local damage to the concrete. This damage reduces the effective

area of the column at the contact interface.

Deformation due to the post axial load

-0.45

-0.30

-0.15

0.00

0.15

0.30

0.05 0.15 0.25 0.35 0.45

Time (s)

Def

lect

ion

(m

)

D=0.2 ; 1687kN D=0.8 ; 1750kND=0.4 ; 1725kN D=1.0 ; 1812kN

Figure 5.10: Deflection characteristics of 450mm column

Even though the 400mm column with 20% axial load clearly demonstrated the various

stages of failure under the post impact loading, the 600mm and 750mm columns

immediately collapsed and hence the critical impulses for each damage index were

hardly noticeable. Therefore, if the columns are damaged due to impacts with small

deflections reflecting shear failure, the only option is to replace the damaged column

as it may catastrophically fail under further loading. Consequently only the mean

failure impact force is used for comparison purposes under 20% to 60% axial loading.

In general, the impact capacity of the columns reduced by 10%, 20% and 30% under

the 60%, 40% and 20% axial loads respectively. This is evidence for the improved

capacity of the impacted column with the axial load, due to its improved stiffness as

shown in Figure 5.11.


5-132

Axial load sensitivity of impacted columns

0

1

2

3

4

5

6

0.3 0.45 0.6 0.75Diameter (m)

Pea

k fo

rce

(MN

)

P=100% P=60% P=40% P=20%

Figure 5.11: Axial load sensitivity of impacted columns

However, it was evident that flexural ductility reduces significantly in flexure

dominated columns with the increase in compressive axial load (Paultre et al. 2001;

Zeinoddini et al. 2002; Gopalaratnam et al. 1984). As the axial load increases, the

concrete is subject to higher compressive stress levels, such that the moment capacity

of the column depends mainly on the compressive strength of the concrete. Thus the

impact capacity has decreased with the axial load enhancement in flexure dominated

columns. Conversely the shear capacity of the columns is increased with the enhanced

axial load (see Fig. 5.11). In other words, the compressive stress enhancement will

reduce the lateral impact capacity of flexure dominated columns while in shear

dominated columns the opposite occurs. Either of these two factors can govern the

impact capacity of the column depending on the elevation of the impact. For instance,

shear failure will be initiated by a low elevation impact and hence the impact capacity

will be improved with the axial load increment as a result of shear capacity

enhancement.

Since concrete is brittle in nature, flexural strength reduces rapidly after reaching the

maximum moment capacity of the column. Moreover, the depth to the neutral axis

increases as the axial load level increases and the extreme concrete fibre is subjected to

higher compressive strains conditions under the lateral impact. Under these

circumstances the concrete will reach its ultimate strain sooner and as a result, the

concrete cover will spall off rather quickly, causing a decline in the flexural capacity of

the section. Thus the moderately damaged column fails in flexure under the axial load

at the post impact stage.


5-133

When concrete strength increases, the amount of transverse reinforcement has to be

increased to acquire the same level of ductility if the axial load level remains the same

(Paultre et al. 2001). If sufficient transverse reinforcement is provided, the reduction in

flexural capacity can be compensated by increasing the capacity of the concrete core,

such that the concrete core could dilate properly under large compressive axial loads

(Johnny 2003). However if the axial load is further increased, the same ductility level

has to be achieved by increasing the amount of transverse reinforcement due to

premature dilation of the core concrete. This will result in congested transverse

reinforcement and using lateral steel with a high yield strength is suggested to

overcome the problem. This may not necessarily increase the ductility when the lateral

steel ratio remains constant (Azizinamini 1994) since the lateral expansion of the

concrete core is not greater and the tensile capacity of the steel may not be fully

developed under working conditions (Johnny 2003). Therefore the parametric study

was extended to investigate the effect of the yield strength of transverse reinforcement,

hoop diameter and spacing under varying axial loading conditions.

5.8.5 Effects of transverse reinforcement on capacity enhancement

To investigate the influence of the transverse steel characteristics the hoops spacing,

diameter and yield strength were varied while the configuration was kept constant.

The hoop spacing varied from 50 to 250mm and an initial investigation was performed

using 6mm diameter hoops with 350MPa yield strength. The 250mm spacing

generated the nominal confinement conditions while 50mm allowed the behaviour of

the impacted column to be investigated under enhanced ductile conditions.

5.8.5.1 Effects of hoops spacing on partially loaded columns

According to the observations there is a substantial improvement of the impact

capacity of the 450mm column under 20% loading. The formation of plastic hinges

near the top and bottom supports of the column is the reason behind this observation

and the numerical simulation was extended to check whether the desired ductile

behaviour could be generated in the rest of the columns by increasing the transverse

reinforcement. In general, the ductile capacity depends on the amount and distribution

of transverse reinforcement within the plastic hinge region and this concept is

particularly effective under earthquake loading conditions.


5-134

Displacement characteristics of the column

15

20

25

30

35

0.2 0.4 0.6 0.8 1

Axial load ratio

Dis

pla

cem

ent (

mm

)

300mm Col.; L250 300mm Col.; L50

Figure 5.12: Deformation characteristics of 300mm column

As observed in the present study, the ductile behaviour can not be improved to a

greater extent in the impacted columns by providing hoops at closer intervals.

Behaviour of the 450mm column can be considered as an exceptional case where the

ductile behaviour of the column was further improved by the closer hoop spacing. As

far as the overall behaviour of the impacted columns is concerned, there is no plastic

hinge formation of the columns under low elevation impacts. In this circumstance, the

effects of transverse reinforcement are limited to enhance the shear capacity of the

columns and to provide more confinement to the core concrete. Such enhancement is

still desired for the impacted columns as they need to bear a certain amount of axial

load despite the damage caused by the impact. In contrast, the ductility is slightly

improved when the axial load is enhanced. This is the desired behaviour for shear

critical columns and concurs with the hypothesis that dilation of concrete will be

further improved with the enhancement of the axial load. This argument is confirmed

by the fact that the displacement characteristics of the fully loaded column are

improved by the closer transverse steel spacing (see Fig. 5.12). For instance, 65%

improvement of the ultimate deflection is observed under confined conditions

compared to only 15% improvement under the nominal conditions with varying axial

load. A 300mm column is considered for comparison purposes as it magnified the

deflection characteristics.


5-135

Impact capacity of the 300mm column

0.35

0.40

0.45

0.50

0.55

0.60

0.65

0.2 0.4 0.6 0.8 1

Axial load ratio

Pea

k fo

rce

(kN

)

300mm Col.; 4% steel; L250 300mm Col.; 4% Steel; L50

Figure 5.13: Impact capacity of 300mm column under varying axial loads

In fact, both the ductility and shear capacity improvement contribute to the impact

capacity enhancement of the 300mm column (see Fig. 5.12) and consequently the

capacity improves by 10% with the confinement (see Fig. 5.13). The contribution of

the displacement (via tensile strength activation) may be predominant with the axial

load enhancement as the deflection of the column increases. This will reflect the

secondary moments generated by the eccentric axial loads and in practice the

deflection characteristics may be even higher as concrete can dilate properly under

high axial loading conditions and resist higher compressive stresses generated by

flexure before failure. However, the capacity improvement due to the enhanced

displacement is insignificant and consequently the two lines in Figure 5.13 are almost

parallel. Thus the confinement characteristics have improved the ductility of the

column, while the shear capacity is the main factor that contributes to the impact

capacity enhancement. These comments are applicable to all the other columns except

the 450mm diameter column.

5.8.5.2 Role of hoops spacing as a early warning system prior to collapse

Figure 5.14 compares the impact capacity increment of the columns with the hoops

spacing. Columns are axially loaded 20% to 100% of their design capacities. In

general the percentage reduction of the impact capacity is 10%, 20% and 30% when

the axial load decreases by 40%, 60% and 80% respectively. There is a series of

critical impulses where the 300mm and 450mm confined columns behave similar to

the slightly damaged and highly damaged conditions according to the intensity of the


5-136

pulses. However there is no significance improvement of the critical range so that each

column can exhibit some warning signs at the post loading stage before collapse. In

other words, collapse of the impacted columns will be brittle and sudden upon post

impact loading even though the residual deflections of the columns are insignificant.

The condition remains unchanged even after introducing hoops at closer intervals even

though it contributes to the overall capacity enhancement. As far as all the confined

columns are concerned there is a trend to enhance the capacity with the enhancement

of axial loading although the percentage increment randomly varies from 10% to 18%.

Figure 5.14: Enhanced capacities for confined columns with different axial loading In fact, typical structural columns may not be subjected to substantial shear forces

during their serviceability state and hence their full shear capacity is available to

supply the demand. It is also worth to note the probability of catastrophic shear failures

is increase with the intensity of the vertical loading. However compromise between

axial load intensity and the allowable impact load may not be worth as there is no

residual capacity remains in the impacted columns.

5.8.5.3 Effects of hoop diameter and yield strength on capacity enhancement

The effects of the transverse reinforcement on columns with reduced axial load were

comprehensively investigated. The aim is to check whether the capacity reduction can

be compensated by shear strength enhancement, by increasing the diameter or yield

stress of the transverse reinforcement as opposed to reducing the spacing. If the


5-137

transverse reinforcements are effective, the impact capacity will suddenly change

when the shear capacity exceeds the flexural capacity of the columns or vice versa.

Impact capacity vs f' l /f' co

0.0

1.5

3.0

4.5

6.0

7.5

0.00 0.05 0.10 0.15 0.20f' l /f' co

Impa

ct c

apa

city

(M

N)

P=1.0; 300mm P=1.0; 450mm P=1.0; 600mmP=1.0; 750mm P=0.6; 300mm P=0.6; 450mmP=0.6; 600mm P=0.6; 750mm P=0.4; 300mmP=0.4; 450mm P=0.4; 600mm P=0.4; 750mmP=0.2; 300mm P=0.2; 450mm P=0.2; 600mm

Figure 5.15: Capacity reduction due to axial load

A range of possible reinforcement ratios are achieved by changing the diameter from 6

to 12mm and the yield strength from 250MPa to 500MPa, with the hoop spacing

changing from 50 to 250mm similar to the earlier analyses. However, further

improvement of the ductility is not possible and all the results are summarised in

Figure 5.15. As mentioned in the previous paragraph, the small improvement of the

ductility is not effective for enhancing the capacity of the columns. Thus the shear

capacity governed the failure throughout the analyses. Therefore, it can be concluded

that the transverse reinforcement does not have a significant influence on the failure

mode of the columns under impact loading, particularly when the axial load ratio

exceeds 20%.

Impact capacity improvement vs f' l /f' co

0.65

0.75

0.85

0.95

1.05

1.15

1.25

1.35

0.00 0.05 0.10 0.15 0.20f' l /f' co

Ca

paci

ty in

cre

me

nt

P=1.0; 300mm P=1.0; 600mm P=1.0; 750mmP=0.6; 300mm P=0.6; 450mm P=0.6; 750mmP=0.4; 300mm P=0.4; 450mm P=0.4; 750mmP=0.2; 300mm P=0.2; 600mm P=0.2; 750mm

Figure 5.16: Capacity enhancement due to confinement


5-138

The impact capacity increment due to the reduction of the axial load and confinement

is given in Figure 5.16. The results are compared with fully loaded columns having

6mm hoops with 350MPa at 100mm spacing. As the axial load reduces, the impact

capacity decreases and the maximum capacity drop is around 65% under 20% axial

load, as designated by the bottom line of Figure 5.16. With confinement, the impact

capacity increases and up to a minimum of 75% of the capacity drop can be recovered

by providing 12mm bars with 500MPa at 100mm spacing. If the axial load on the

column can be maintained around 40%, then 90% of the impact capacity drop can be

recovered as shown in the same figure.

5.8.6 Effects of longitudinal reinforcement ratio

Equivalent columns in capacity for 5 to 15 storey buildings made of 50MPa concrete

with 1% and 4% steel ratios are simulated in this study to investigate the influence of

the steel ratio on the impact behaviour of columns under varying axial loading

conditions. Axial load varied from 20% (0.2Pd) to 100% (Pd) in the columns with

nominal hoops spacing as shown in Figure 5.17. In general, the vulnerability reduces

with reduction of longitudinal steel ratio and capacity enhancement is more

pronounced in fully loaded 5 storey column. However there is no considerable change

in the ductility due to steel ratio even for 5 storey building columns and hence there is

no abrupt change in the capacity enhancement except in 450mm column in which the

exceptional ductile behaviour occurs as discussed previously in detail. However

350mm column shows some brittle behaviour with the axial load increment and hence

the entire shear capacity of the column is fully utilised. Therefore highest increment of

around 84% is observed in 5 storey building columns under fully loaded conditions

while 15 storey columns shows only 43% increment (see Fig. 5.18). Due to abrupt

behaviour of the 450mm column, 10 storey building columns excluded from the

comparison. As far as 15 storey columns are concerned, ductile behaviour of the

columns do not affected by either steel ratio or axial load ratio on the columns.

Therefore capacity increment is remained constant throughout the analyses. In

addition, the range of impulses that can causes the damage to the columns designated

by the parameter Di, varying from minor (Di=0.2) to severe (Di=0.8) conditions is

further shrink with the reduction of the steel ratio. In fact this is a sign of flexural


5-139

capacity reduction with the reduction of longitudinal steel ratio even though the

overall impact capacity is enhanced. Therefore columns with low steel ratio fail

abruptly without giving any warnings prior to collapse.

Figure 5.17: Impact capacity under varying Figure 5.18: Capacity increment under

axial load varying axial load

5.8.7 Effects of the slenderness ratio

Figure 5.19: Impact capacities of short columns In this analyses columns with diameters 350, 500 and 700mm with 1% steel are

investigated. Transverse reinforcement is provided at 250mm spacing and 6mm bars

with 350MPa yield strength was used in the process. The main objective is to

investigate the impact behaviour of short columns under low axial loading conditions

and effective height is varied from 2m to 4m range. The results are shown in Figure

5.19. Because of the various diameters are in use the slenderness ratio provides a

common base for comparison purposes. Overall capacity of the columns increases

with the reduction of the effective height as expected. According to the observations


5-140

higher capacity enhancement is noticed in small columns. On average it is 10% to 15%

for 700mm column and 350mm column respectively when the effective height is

decrease to 3m from 4m height. The enhancement is further increased 35% to 50%

with reduction of the effective height to 2m. However there is no firm evidence for

interrelationship between the reduced axial load, diameter and the percentage of

capacity enhancement. Moreover, mode of failure sticks to the shear type of failures

under 2 to 4m effective height despite the diameter of the column and the percentage

of the axial loading. Thus the columns fail due to bulging at the impacted point as the

volumetric strain exceed the elastic limit under the application of post axial loads and

this behaviour is remained unchanged for a range of impulses ranging from Di=0.2 to

0.8. Again the failure will be catastrophic even though the capacity of the column

enhanced with reduction of the effective height.

5.8.8 Anomalous behaviour of columns under post impact loading

(a) Flexural failure (b) Shear failure with concrete crushing

Figure 5.20: Different failure characteristics of structural columns The following observations were made on the impact behaviour of a 20% loaded

300mm diameter column under subsequent increase of the axial load at the post impact

stage. The main observation was that the residual displacements change direction

rapidly with the intensity of the impact loading even though the deflection

characteristics during the impact are almost the same.

Bulging at the impacted point


5-141

Figure 5.21: Axial load sensitivity of the counterintuitive effect

From a concrete mechanics point of view, micro cracks are generated at the opposite

side, due to the displacement caused by impact forces. Once these cracks are formed

the concrete has reached its residual state as the elements do not carry any stress

afterwards. At the same time, the compressive stress on the impacted surface does not

exceed the compressive strength of the concrete and thus the surface remains

unharmed. When the axial load is applied at the post impact stage, the column tends to

deflect inwards (towards impacted face) as there is load eccentricity due to the element

failed in tension in the opposite face of the impact. Figure 5.20(a) shows the points

where cracks initiated in the opposite face. This behaviour becomes more predominant

when the amplitude of the impact force is increased, where the column subjected to

brittle failure under 20% axial load deflecting towards its impacted face.

When the impact force is further increased, the compressive strength of the impacted

surface is also reached in the residual stage due to crushing of the concrete. Therefore

there is no substantial load eccentricity for buckling and the concrete core will tend to

bear the axial load without any initialisation for buckling. Thus the column will fail by

crushing the core concrete, and as the volumetric strain exceeds the elastic limit,

compaction will occur and the concrete will tend to disintegrate (see Fig. 5.20(b)).

When the initial axial load on the column is further increased, it will tend to increase

the moment capacity of the columns and hence the residual deflection towards the

opposite direction of the impact is minimised. Further axial load enhancement will

minimise tensile stress development in the opposite side while crushing the concrete in


5-142

the vicinity of the impact. Resultant load eccentricity will deflect the column towards

the opposite face of the impact (see Fig. 5.21). This behaviour gradually reduces as the

column diameter increases. In addition, provision of lateral steel at closer intervals will

also minimise this behaviour by enhancing the moment capacity of the columns.

Figure 5.22: Typical failure pattern of rectangular columns under eccentric loading

Similar observations were made when rectangular columns were subjected to

eccentric loading as shown in Figure 5.22 (Nemecek et al. 2005). The collapse of the

columns was initiated by concrete softening accompanied by symmetric buckling of

reinforcement bars on the compression side around the mid height. The bars always

buckled between the hoops and failure was localised. The damaged zone was larger for

dense (50mm c/c) stirrups spacing and small for coarse (150mm c/c) stirrups spacing.

From a comparison point of view, the similarity is the eccentric loading conditions

where the load eccentricity is generated by the localized damage to the concrete due to

the impact. The location of the affected region differs as the damage is initiated within

the impacted region.

There have been cases where the observed deflection was on the same side of the

impact under impulse loading. This is also known as counterintuitive or anomalous

column response and is supported by the experimental results published by Li et al.

(1991) and Kolsky et al. (1991). It has also become evident that counterintuitive

behaviour is extremely parameter sensitive and is constrained to a particular region of

impulse loading. In other words a narrow zone of loading sensitivity can be identified

for such a system within which the dynamic rebounding can occur and the final

deflection of the column (or beam) rests in the opposite direction of the loading. It has

also been shown that this transition is abrupt with respect to the change of loading


5-143

parameters. Similar phenomena were also reported for several other dynamic systems

under radial pulse pressure loading (Forrestal et al. 1994), blast loading (Galiev 1996)

and underwater explosion (Galiev 1997).

According to Li et al. (2006), the time increment, element type or a change of

simulation software may influence the existence of the counterintuitive phenomenon.

These factors can also change the width of the sensitivity region. Apart from that, the

anomalous behaviour is predominant closer to the region of transition from elastic to

relatively plastic. The deviation observed here can be categorised as a probabilistic

response of the system due to variation of the parameters. That is, the uncertainty of

the response occurs mainly due to the randomness of the system parameters. If the

system parameters can be determined accurately to a certain extent and the system

itself is not parameter sensitive, the results can be predicted with sufficient accuracy.

Although the parametric sensitivity of this system is comparatively less than that of a

chaotic system, the uncertainties encountered in the practice may still be able to cause

some response uncertainty through the parametric sensitivity. However, research

conducted up to date has provided the solution only for very simple beams and plates.

Therefore the data accumulated up to now is more restricted to the mathematical and

theoretical aspects of the problem rather than engineering and practical applications

(Galiev 1997).

Formulations developed for a generalised event may not be valid in the region within

which the parametric sensitivity is predominant. The sensitivity normally occurs

closer to the nonlinear region and hence the equilibrium equations based on dynamic

formulation under impact may need further modifications to calculate the column’s

residual strength particularly under the effect of axial load. Therefore the structural

response cannot be predicted by solving the equation of motion. On the other hand,

parameters dealing with this problem are also complicated. Hence the finite difference

method may be more reliable in the sensitivity analysis process. However either

truncation errors or condition errors may occur under the finite difference method

depending on the selected perturbation step length. For example, the truncation error

may be excessive if the selected load increment step size is too large. Alternatively if

the step size is too small, condition errors may occur due to inaccuracies in the


5-144

calculation of the displacement and round off errors in the finite difference

calculations. Therefore selection of a proper step size to capture the dramatic variation

itself would be a difficult task (Adelman and Haftka 1986; Li et al. 2006).

5.8.9 Buckling of reinforcement under impact

(a) Concrete crushing (b) Internal forces generated in longitudinal steel

Figure 5.23: Failure of columns by concrete crushing When the column fails as a result of concrete crushing under post impact loading

conditions, the reinforcement in the vicinity of the impacted area tends to buckle

outward by opening the longitudinal steel. A typical example is shown in Figure

5.23(a). All the longitudinal reinforcement tend to buckle simultaneously and the

resultant tensile forces on the hoops tend to detach the hoops from their anchorage by

releasing all the confinement effects on the surrounding concrete. Consequently the

column catastrophically fails from the point of impact with concrete crushing. This

behaviour has been reflected in numerical simulations with magnificent accuracy. The

axial forces developed in the longitudinal and lateral steel at the point of impact are

shown in Figure 5.23(b). Once the longitudinal steel buckles, the hoops tend to fail in

tension. Compressive forces that develop in the longitudinal steel are shown with a

negative sign and tensile forces in the hoops are shown with a positive sign. It is clear

that all the reinforcements in the non-linear region activate simultaneously within a

short period of time.


5-145

5.9 Derivation of empirical relationships to predict critical impulse

Whilst the individual parametric studies reveal some interesting features of the

impacted columns, they do not reveal their collective contribution as far as routine

column designs are concerned. To comprehensively understand the cumulative effects

of key design parameters, a thorough statistical multivariate regression analysis was

carried out. This section reports part of the regression analysis.

Empirical relationships are developed in stages by varying each parameter at a time

and then combined to produce an equation based on the least square method, which

can be used to quantify the peak force and the associated impulse at the near collapse

stage for fully loaded columns. Some of the terms in the empirical relationships have

theoretical explanations and therefore the relationships can be considered as

semi-empirical. The final results are approximate values for the characteristics of the

critical impulse in terms of logarithm of Peak Force Log Pc, and logarithm of Impulse

Log Ic, particularly for a 100ms impact. The relationship is valid under specific

conditions as discussed later in this paper.

5.10 Derivation of simple linear regression equations

A simple linear correlation between the parameters namely the diameter of the column

D, steel ratio v

ρ , concrete grade 'cf , effective height H, and yield strength '

syf , area

hA and spacing s of the hoops, is determined by using a statistic program ‘StatistiXL’.

The objective of the statistic analysis is to check whether the equation can take into

account cumulatively the key variables to predict the peak force and the associated

impulse which leads to a near collapse of a column during a 100ms impact. A multiple

regression analysis is performed on the data set to obtain the correlation coefficients of

a possible linear relationship. The outcomes of the analysis are described in the

following section.

5.10.1 Descriptions of the outputs

Table 5.1 gives the descriptive statistics for the dependent variable Log P, followed by

all independent variables in the order of the column entry. A total of 141 data records

are used, and the standard deviation of each term is also given.


5-146

Table 5.1: Descriptive Statistics

Variable Mean Std. Dev. Log P 6.468 0.349 Steel ratio; ρv 0.032 0.012 Com. Strength; f ’ c 39.299 9.208 Height; H 3.674 0.660 Diameter; D 0.589 0.176 Yield strength of hoops fsy 0.240 0.129 Hoop spacing s 353.546 48.050 Area of hoops; Ah 35.846 22.152

5.10.2 Pearson Correlation

The Pearson correlation is a number between -1 and +1 which measures the degree of

association between two variables such as X and Y. A positive valve indicates a

positive association of large values of X and Y pairs or small values of X and Y pairs.

Conversely, negative values indicate that a negative or inverse association and large

values of X tend to be associated with small values of Y and vice versa. The Pearson

Correlation is computed as;

( )( )( )

yx

ii

x

i

SSn

YYXXr

11

−−−∑= = Eq : 5.4

where Xi and Yi are two arbitrary variables, and X and Y are the associated mean

values. x

S and y

S are the standard deviations respectively. The term ( )( )YYXXii−−

governs the sign of the correlation depending on the respective values of i

X and i

Y.

The correlation coefficients measure the strength of a linear relationship between two

variables. A value of ±1 indicates a perfect linear relationship and the relationship

tends to decrease when the coefficient decreases. In general, a valve between ±1.0 and

±0.7 indicates a strong association and a value between ±0.7 to ±0.3 represents a

negative association. This definition is however somewhat arbitrary and can be

deviated. In addition, the correlation also depends on the sample size and will not

reflect the practical significance. As there are a number of independent variables the

correlations between every pair can be arranged into a matrix as shown in Table 5.2. It

can be seen that there is an inverse correlation between the logarithm of peak force

Log P, and Height H. The relationship of Log P, to the parameters such as s, 'syf and Ah

also agree with the general perception.


5-147

On the other hand, considerable correlations that exist among the individual

parameters indicate that alternate options may be adopted. However, for a general

case of axial loading on a column and in the absence of design guidance for impact,

some of the other variables such as spacing of hoop steel and its diameter can be

considered as arbitrary. Therefore the correlation factors do not reflect the actual effect

on the capacity enhancement and hence the correlation does not affect the regression.

Table 5.2: Pearson correlations

5.10.3 Coefficient of Determination and Analysis of Variance

According to Table 5.3 the Coefficient of Determination (R2) indicates that 94% of the

variation in Log P, is explained by variation in the independent X variables, and the R

value 0.97, which is the square roof of R2, indicates a strong correlation between Y and

X variables. The Standard Error of Estimate, 0.084 is only 1% of the mean of Log P,

6.47 and thus indicates that the Multiple Regression model has accurately calculated a

large amount of the Log P values. The adjusted coefficient of determination (R2adj)

provides an unbiased estimate of the coefficient of determination by allowing for the

degrees of freedom of R2 particularly with the numerous independent variables.

Table 5.3: Coefficient of determination of the equation

R2 R Adj. R2 S.E. of Estimate 0.944 0.972 0.942 0.085

Analysis of Variance is used to determine whether there is a difference between three

or more categorical sets of values while t-test is used to compare two groups. First row

of the Table 5.4 indicates the significance of the multiple regression model. The much

larger mean square for the regressing 2.29 than the residual error 0.007 indicates that


5-148

the model is highly significant with zero probability of error. F statistic for linear

regression indicates the statistical probability of the partial regression coefficients for a

multiple linear regression is equal to zero.

Table 5.4: Analysis of Variance Source Sum Sq. D.F. Mean Sq. F Prob.

Regression 16.078 7 2.297 321.157 0.000 Residual 0.951 133 0.007 Total 17.029 140

5.10.4 Interpretation of partial (regression) plots

A partial plot is a graphical representation of the relationship between a given

independent variable (Xi) and the response variable (Yi), with adjustment to the

independent variable to reflect the effect of other independent variables in the model.

Thus it plots the residuals of regressing response variable (Yi) predicted from all other

independent variables except Xi verses the residuals of that independent Xi variable

regressed against all the remaining independent variables. For instance, in Figure

5.24(a) the Y axis represents the residuals from a multiple regression of Log P against

all the other independent variables such as height, concrete grade etc. except D. The X

axis represents the residuals of a multiple regression of diameter D against the other

independent variables such as concrete grade, steel ratio etc.

When performing a linear regression analysis with a single independent variable, each

scattered plot provides a good indication of the nature of the relationship between the

response variable and each independent variable. As there is more than one

independent variable in this analysis, the partial regression plots specifically indicate

the proper relationship when the plotted independent variable has a strong correlation

with other independent variables in the model. A linear trend indicates a significant

relationship between Y and the Xi. Also the regression of the sets of residuals should

pass through the origin and have the same slope as the regression coefficient for that

particular independent variable Xi. The influence of individual data values on the

estimation of regression coefficient is clearly seen on these plots and the partial plots

also enables examination of non-linearity and outlying points or points that contribute

heavily to the regression.


5-149

Partial Plot of D

-1.0

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

-0.4 -0.2 0.0 0.2 0.4

Column diameter D

Log

(P)

Partial Plot of H

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

-2 -1 0 1Height H

Log

(P)

(a): Partial regression plot of diameter (b): Partial regression plot of height

Partial Plot of ρ v

-0.30-0.25-0.20-0.15-0.10-0.050.000.050.100.150.20

-0.02 -0.01 0.00 0.01 0.02

Steel ratio ρ v

Log

(P)

Partial Plot of s

-0.3

-0.2

-0.1

0.0

0.1

0.2

-0.2 -0.1 0.0 0.1 0.2

Hoop spacing s

Log

(P)

(c): Partial regression plot of steel ratio (d): Partial regression plot of hoop spacing

Partial Plot of Ah

-0.30-0.25-0.20-0.15-0.10-0.050.000.050.100.150.20

-40 -20 0 20 40 60 80

Area of hoop Ah

Log

(P)

Partial Plot of f' sy

-0.30-0.25-0.20-0.15-0.10-0.050.000.050.100.150.20

-200 -100 0 100 200

Yield strength of hoops f' sy

Log

(P)

(e): Partial regression plot of Ah (f): Partial regression plot of '

syf

Figure 5.24(a-f): Partial regression plots of each parameter against Log P

The outlying points can be identified by examination of the X-Y plots. It is not unusual

that one or more data points in a sample do not comply with the chosen model.

However, a formal statistical test is needed to identify these outlying points to avoid

classifying too many points as outlying. As the test results are based on finite element

analyses, the deviations occurred as a result of exceptional outcomes resulting from


5-150

the higher order vibrations and other inherent properties of the impacted columns.

Therefore no outlying points have been declared in this analysis as they do not result

from numerical errors.

5.10.5 Regression coefficients and derivation of the linear equations

Table 5.5 represents the regression coefficients of the linear regression equation. Sy.x is

the Standard Error of Estimate which is defined as the square root of the error mean

square; it is the variance in Y after accounting for the dependency of Y on X. Sβ is the

Standard Error of the Slope (β) given in dotted lines in Figures 5.24(a) to (f) and is

defined as ∑ 2

.XS

xy. 95% confidence intervals (C.I.) are given in the next column

and t represents the t Statistic. Here the t is based on the assumption that each partial

regression coefficient (βi) follows a normal distribution. The t value tests whether a

partial regression coefficient differs from a particular value b and is calculated as;

( )iiXYi

cSbt.

−= β , Eq : 5.5

where cii is an element in the inverse matrix of the corrected sum of cross products. As

the probability of the regression coefficient of steel ratio ρ has a substantial deviation,

the steel ratio does not have a strong co-relationship with the impact capacity of the

concrete columns. One of the main reasons would be the non-identical configurations

of the longitudinal steel across the circumference of the various columns. This is

unavoidable in construction practice, and hence to account for this in the finite element

model, the peripheral spacing between the longitudinal steel was kept close to 100mm

as much as possible. However, the spacing is inevitably varied from 70mm to 110mm

depending on the configuration of the longitudinal steel. No further effort has been

taken to minimise the distance between the bars as it violates the general procedures

used in practice. Consequently, the depth to the neutral axis deviated from one column

to the other, so that no comparison can be made between columns having various

diameters. Even though this database does not reflect the influence of steel ratio, it is

included as a variable because the selected steel configurations are much closer to the

practical applications.


5-151

Table 5.5: Regression Coefficients for linear equations Parameter Coefficient Sy.x Sβ -95% C.I. +95% C.I. t Prob. Intercept 5.458 0.097 0.0 5.266 5.649 56.416 0.000

ρv -0.181 0.727 -0.006 -1.620 1.257 -0.249 0.803 f ’ c 0.004 0.001 0.100 0.002 0.005 4.754 0.000 H -0.071 0.012 -0.134 -0.095 -0.047 -5.848 0.000 D 1.897 0.041 0.958 1.816 1.978 46.349 0.000 s -0.209 0.075 -0.077 -0.357 -0.061 -2.790 0.006

f ’ sy 0.0004 0.000 0.014 0.000 0.000 0.674 0.502 Ah 0.001 0.000 0.042 0.000 0.001 1.911 0.058

46.5001.021.09.107.0004.018.0 ' ++−+−+−= hcv AsDHfPLog ρ Eq : 5.6

Equation 5.6 can be used to calculate the predicted Peak Force P of the critical impulse

for a typical 100ms vehicle impact. The standard errors of the regression coefficients

are also given and the significance of the regression coefficient is determined by a

t-test. For instance, for diameter, D the multiple regression coefficient is +1.897±0.041

and ranges from its -95% confidence limit of +1.816 to its +95% confidence limit of

1.978. The t-value for this coefficient of 46.35 is significant with zero probability of

error. In contrast, the coefficient for the longitudinal steel ratio is not significant since

t = -0.25 and hence the effect on the final value is not considerable even though the

probability is 0.803. The standard partial regression coefficient, Sβ signifies the

relative importance of each independent variable. A variable with a high Sβ is

relatively more important than a variable with a lower Sβ. The Sβ essentially takes into

account the possible variation in scale of the different X variables. If 'cf is measured in

Pascal, Pa rather than in MPa, then very different partial regression coefficients would

be expected due to the difference in scale, whereas the standardised partial regression

coefficients take this difference in scale into account in the comparison.

Residuals vs Predicted Log (P )

y = -0.5595x2 + 7.2421x - 23.373

R2 = 0.6323-0.30

-0.20

-0.10

0.00

0.10

0.20

5.5 6.0 6.5 7.0 7.5

Predicted Log (P)

Re

sidu

als

Figure 5.25: Accuracy of the prediction by linear equations


5-152

Figure 5.25 represents the accuracy of the predicted data points compared to the

observed values. The residuals are calculated as (Observed (Log P) - Predicted (Log

P)) and hence the positive and negative residuals indicate an under and over-prediction

of the data points respectively. By considering the distribution of the residuals the

accuracy of the Log P can be further improved. The corrected Log Pc and Log Ic are

given by;

Eq : 5.7 Eq : 5.8

The final over and under prediction of the Peak Force, P is within the range of ±15 %.

Hence polynomial equations are generated for more accurate estimation.

5.11 Derivation of Polynomial equations

Polynomial equations are derived by assuming that all the parameters are independent

and there is no correlation among them. The mathematical relationships between

individual variables and the peak of the impulse force are determined by changing one

variable while keeping the others fixed. As this procedure involves an enormous

amount of modelling, the relationship between the column diameter and the peak

impact force is determined first by keeping all the other parameters fixed. Once this

relationship is derived as a polynomial equation of (D/0.5), the constant term of the

equation should represent the effects of all other parameters if it is assumed that there

is no correlation between the diameter and the other independent variables. This

assumption will considerably reduce the total number of simulations needed to form

the database for derivation of equations. Once the correlation between the diameter

and the peak force is known, the known terms can be transferred to the left hand side of

the equation except the constant term. The correlation between peak impact force and

steel ratio is then determined by varying the steel ratio of individual columns. For the

remaining calculations the same procedure is repeated (see Fig. 5.27(a) to (g)). In these

Figures the X axes values are normalised with respect to typical values of the

respective parameters. As there is no correlation between steel ratio and effective

height, concrete grade, and hoop characteristics, the correlation between the individual

parameters and peak collapse force can be calculated as a second order polynomial

30.1

37.23242.8)(56.0 2

−=−+−=

cc

c

PLogILog

PLogPLogPLog


5-153

equation which generates much more accurate results compared to the linear

equations.

Log P vs (D /0.5)

y = -0.3715x2 + 1.8787x + 4.7598

R2 = 0.9936

5.5

5.8

6.0

6.3

6.5

6.8

7.0

0.6 0.7 0.8 0.9 1 1.1 1.2 1.3 1.4 1.5 1.6(D /0.5)

Lo

g P

Log P vs (D/0.5) Poly. (Log P vs (D/0.5))

Log P-f(D /0.5) vs (ρ v /0.04)

y = -0.0893x2 + 0.0568x + 4.7903

R2 = 1

4.75

4.76

4.77

4.78

4.79

4.80

4.81

0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

(ρ v/0.04)

Lo

g P

-f(D

/0.5

)

Log P-f(D/0.5) vs (r/0.04) Poly. (Log P-f(D/0.5) vs (r/0.04))

(a): Log P vs. diameter (b): Log P-f(D) vs. steel ratio

Log P-f(D /0.5)-f(ρ v /0.04) vs (f' c/45.6)

y = 0.2374x2 - 0.2803x + 4.8334

R2 = 1

4.74

4.75

4.76

4.77

4.78

4.79

4.80

0.50 0.60 0.70 0.80 0.90 1.00

(f' c /45.6)

Lo

g P

-f(D

/0.5

)-f( ρ

v/0

.04

)

Log p-f(D/0.5)-f(r/0.04) vs (f/45.6) Poly. (Log p-f(D/0.5)-f(r/0.04) vs (f/45.6))

Log P-f(D /0.5)-f(ρ v /0.04)-f(f' c/45.6) vs (H /4)

y = 0.3138x2 - 0.7969x + 5.3111

R2 = 1

4.80

4.84

4.88

4.92

4.96

5.00

0.50 0.60 0.70 0.80 0.90 1.00(H /4)

Lo

g P

-f(D

, ρv,

f' c)

Log(P)-f(D/0.5)-f(r/0.04)-f(f'c/45.6) vs (H/4)

Poly. (Log(P)-f(D/0.5)-f(r/0.04)-f(f'c/45.6) vs (H/4)) (c): Log P-f(D,ρ) vs. compressive strength (d): Log P-f(D,ρ,f ’c) vs. height

Log P-f(D /0.5)-f(ρ v/0.04)-f(f' c/45.6)-f(H /4) vs

(s/0.35)

y = -0.0565Ln(x) + 5.3098

R2 = 0.9999

5.305.325.345.365.385.405.425.44

0 0.2 0.4 0.6 0.8 1(s/0.35)

Lo

g P

-f(D

, ρv,

f' c,H

)

Log P-f(D/0.5,f'c/45.6,H/4,r/0.04)Log. (Log P-f(D/0.5,f'c/45.6,H/4,r/0.04))

Log P-f(D /0.4)-f(ρ v/0.04)-f(f' c/45.6)-f(H /4)-

f(s/0.35) vs (f' sy/350)

y = 0.225x2 - 0.3433x + 5.429

R2 = 1

5.28

5.32

5.36

5.40

0.7 0.9 1.1 1.3 1.5(f' sy/350)

Log

P-f

(D, ρ

v ,f'c

,H,s

)

Log p-f(D/0.5,f'c/45.6,r/0.04,H/4,s/0.35)

Poly. (Log p-f(D/0.5,f'c/45.6,r/0.04,H/4,s/0.35)) (e): Log P-f(D,ρ,f ’c,H,s) vs. hoop spacing (f): Log P-f(D,ρ,f ’c,H,s) vs. yield strength of hoops

Log P-f(D /0.5)-f(ρ v/0.04)-f(f' c/45.6)-f(H /4)-

f(s/0.35)-f(f' sy/350) vs (Ah /28.27)

y = 0.0497Ln(x) + 5.424

R2 = 0.9987

5.42

5.44

5.46

5.48

5.50

1.0 1.5 2.0 2.5 3.0 3.5 4.0(Ah /28.27)

Log

P-f

(D, ρ

v ,f'c

,H,s

,f'sy

)

Log P - f(D/0.5,r/0.04,f'c/45.6,H/4,s/0.35,f'sy/350)Log. (Log P - f(D/0.5,r/0.04,f'c/45.6,H/4,s/0.35,f'sy/350))

Residuals vs predicted Log(P)

y = -0.2232x2 + 2.831x - 8.924

R2 = 0.3406-0.20

-0.10

0.00

0.10

0.20

5.5 6.0 6.5 7.0 7.5

Predicted Log(P)

Re

sid

ua

ls

(g): Log P-f(D,ρ,f ’c,H,s,fsy) vs. area of hoops Figure 5.26: Accuracy of the polynomial equations

Figure 5.27 (a-g): Steps of the derivation of polynomial equations


5-154

Eq : 5.9

Eq : 5.10

Eq : 5.11

Where Log P is the logarithm of the Peak Force P (uncorrected), Log Pc is the

(corrected) logarithm of the Peak Force P, Log Ic is the (corrected) logarithm of

Impulse I, D is the diameter of the column in m, ρ is the longitudinal steel ratio, 'cf is

the compressive strength of concrete in N/mm2, H is the height in m, 'syf is the yield

strength of hoops in N/mm2, Ah is the area of hoop in mm2 and s is the hoop spacing in

m. With the introduction of the corrected equation, the over and under prediction of the

Peak Force, Pc is reduced up to ±12%. This is a significant improvement as far as the

uncertainties associated with the impact behaviour of the concrete columns are

concerned. Once the Corrected Impulse Ic is known, the critical velocity, v can be

calculated for a known impacted mass m (kg) of a vehicle, in meters per second (ms-1)

by using the relationship given in Equation 5.12. For instance, Eurocode EN 1991-1-7

(2006) suggested that mean mass of 1500kg for cars and 20,000 kg for trucks.

vmI c = Eq : 5.12

In general, polynomial models are among the most frequently used empirical models

for curve fitting functions. They are popular because of their simple form and

moderate flexibility of shapes. However, polynomial models do not extrapolate

reliably. The valid range provides good fits, but deteriorates rapidly outside that range.

Higher degree polynomials are notorious for oscillations between exact-fit values.

Therefore, complex relationships which lead to higher degree polynomials are avoided

to allow interpolation of values within the valid range.

301.1

92.883.3)(223.0

35.525.0

ln056.0

27.28ln05.0

350

'34.0

350

'225.0

48.0

431.0

6.45

'28.0

6.45

'24.0

04.006.0

04.009.0

5.088.1

5.037.0

2

22

222

−=

−+−=

+

−

+

−

+

−

+

−

+

+

−

+

−=

cc

c

hsysyc

cvv

LogPILog

pLogpLogPLog

s

AffHHf

fDDPLog

ρρ


5-155

The valid range of the equations is given as follows;

5.12 Conclusions

This chapter has confirmed the feasibility of using numerical simulation techniques in

vulnerability assessment of impacted columns while enhancing our understanding of

the optimum usage of critical parameters. The main findings are summarised below.

1. It has been shown that numerical simulation techniques can be used for

quantification of the critical impulse of axially loaded circular columns. Empirical

equations are developed in the process to predict the critical (collapse) load and the

associated impulse. Polynomial equations are provided for more accurate

estimation along with linear equations for approximate assessment.

2. Shear failure characteristics are initiated by the low elevation impacts and hence

the concrete grade, diameter of column, steel ratio, slenderness ratio and

confinement effects become the key design parameters that govern the

vulnerability of the columns. In fact, there is a correlation between the dynamic

and static shear capacities that can be used for approximate vulnerability

assessment. According to the limited investigation, dynamic amplification factor

suggested in EN 1991-1-7:2006 generates over conservative results.

3. The columns susceptible to impacts should be checked for all relevant limit states.

In particular the accidental limit state can be declared in-between the

serviceability and ultimate limit states depending on the expected level of safety.

Low shear demand under serviceability conditions strengthen this argument.

However, the limit states formulation based on the kinetic energy or deformation

msm

mmAmm

MPafMPa

mHmMPafMPa

mDm

h

sy

c

v

25.0050.0

1.11327.28

500'250

4250'30

04.001.0

75.03.0

22

<<<<<<

<<<<<<

<<ρ


5-156

capacity (EN 1991-1-7:2006) may not be appropriate under low elevation impact

conditions.

4. Confinement effects are particularly effective when the hoop spacing is less than

100mm. The confinement has to increase particularly with the diameter of the

columns and with concrete grade to achieve the same level of capacity

enhancement. Therefore, a method based on the maximum diameter of the

longitudinal steel is not effective for determining the lateral steel spacing of the

columns susceptible to vehicle impacts.

5. The confinement induced capacity enhancement is more predominant in short

columns as the number of effective hoops increase as the inclination of the failure

plane increases with the confinement effects. Increasing the diameter of the hoops

is recommended rather than yield strength, where restrictions may apply on the

minimum allowable spacing of the hoops. Additionally, the impact capacity

enhancement due to reduction of effective height is more pronounced in columns

exceeding 500mm in diameter.

6. The collapse of the impacted columns will be brittle and sudden upon post impact

loading even though the residual deflections of the impacted columns are

insignificant. Therefore the damage index D, is not a sensitive index for impacted

columns. In particular, there are no substantial warnings before collapse and the

condition remains unchanged even after introducing hoops at closer intervals even

though this contributes to the overall capacity enhancement. Under these

circumstances, it may be more appropriate to replace the impacted columns rather

than repair them for further use. However, the decrease in impact capacity

resulting from partial axial loading can be recovered by providing transverse steel

at closer intervals.

7. The impact capacities reduced by 10%, 20% and 30% under the 60%, 40% and

20% axial loading. This behaviour is typical for shear critical columns where shear

capacity is decreased with the axial load reduction. Even though the possibility of

catastrophic shear failures are increase with the axial load increase, compromise


5-157

between axial load intensity and the allowable impact load may not be worth as

there is no residual capacity remaining in the impacted columns.

8. It was concluded that the transverse reinforcement does not have a significant

influence on the failure mode of the columns under impact loading, particularly

when the axial load ratio exceeds 20% where capacity drop is around 65%.

However, the capacity drop can be recovered by increasing the confinement effects

and almost 90% of the impact capacity drop can be recovered, if the axial load on

the column can be maintained around 40%.

9. Reduction of the steel ratio minimised the band width the region defined from

D=0.2 to D=0.8. Therefore columns with low steel ratio fail abruptly without

giving any warnings prior to collapse. Due to low slenderness ratios most of the

columns failed due to concrete crushing at the impact location where volumetric

strain exceeds the elastic limit under the application of post axial loads.

10. It is observed that there is an extremely parametric sensitive region where the

residual deflection is in the same side of the impact which is known as

counterintuitive behaviour of columns. A narrow zone of loading sensitivity can be

identified for such a system within which the dynamic rebounding can occur and

the final deflection of the column rest in the same side of the loading. Such

response cannot be predicted by solving the equation of motion.

11. Spalling of the cover concrete under impact can be explained using the stress

variation in the core-cover interface resulting from the confinement effects. In

addition, possible alteration to the duration of the impact due to variation of

stiffness resulting from partial loading conditions can be neglected under shear

critical quasi-static loading conditions where peak force determines the

vulnerability of the column.

12. An innovative technique was developed and introduced to ensure the accuracy of

the equations developed for predicting the critical impact force and impulse where

the other techniques are failed due to the shape of the error distribution under

logarithmic scale.


6-159

6. IMPACT ON COLUMNS UNDER UNIAXIAL BENDING 6.1 Introduction

Columns on the front of buildings are a typical example of columns under uniaxial

bending, and they are always adjacent to main roads. Eccentric loads may also result

from misalignment, initial imperfections, movements of joints or support settlements,

while the possibility of eccentric loading induced by vehicular acceleration or

deceleration cannot be neglected in bridge type structures. Eccentrically loaded

columns under static conditions have gained sufficient attention from the scientific

community. However their dynamic capacity under lateral impact loading is yet to be

determined. Initial deformation present in the eccentrically loaded columns may

enhance or reduce the impact capacity of the columns depending on the direction of

the impact. Some insight may also be provided by the column tests under mid-span

impacts where flexural failures are predominant (Remennikov & Kaewunruen 2006).

However there is little or no knowledge on the exact amount of the capacity reduction

due to the presence of eccentric loading.

Axially loaded steel columns under transverse mid span impact have already been

discussed in the literature (Zeinoddini et al. 2008). As the flexure is critical in those

columns there may be some similarity to the eccentrically loaded impacted columns at

the ultimate stage. It was also observed that strain rate, geometry, axial load, the shape

of the impacting body, its velocity and impact position have considerable effects on

impact capacity. Moreover, structural stiffness drops suddenly with transverse impacts

(Sastranegara et al. 2005). It has also been observed that the duration of the transverse

impact has a significant influence on ultimate capacity while buckling of columns can

be controlled by applying transverse impact (Adachi et al. 2004). However, the

behaviour of eccentrically loaded concrete columns impacted at a shear critical height

has not been considered in the past and there is a significant room for improvement

using both experimental and numerical analyses. In addition, the dynamic buckling of

columns under axial impact has attracted much attention in recent decades (Cui et al.

2002). Lateral impacts are likely to have a secondary component along the

longitudinal direction particularly when the eccentrically loaded column deforms


6-160

against the impact. This means transverse impact can be considered as a reverse

scenario of axial impact. In fact, dynamic elastic buckling analysis of simply

supported steel columns under intermediate velocity axial impact has also confirmed

the importance of both amplitude and the duration to determine the severity of the

buckling effects (Cui et al. 2002). Obviously, information presented in the literature

has only little contribution to understanding the lateral impact on concrete columns.

To address this perceived need, an investigation has been conducted on reinforced

circular columns made of different concrete grades, steel ratio, slenderness ratio and

effective height. The duration of the impact was kept constant at 100ms by assuming

shear critical columns do not have much effect on stiffness change. The load

combinations were selected according to the Australian standard AS3600 (2004).

Comparison based on the number of stories will not be considered in the selection of

load combinations which depend on the column spacing rather than the storey height.

The main aim was to investigate the sensitivity of the moment present on the columns

to transverse impact and to derive a numerical equation that can be used to quantify the

impact capacity of the eccentrically loaded columns. In the process different load

combinations were used and initial analysis was limited to impact loads applied in the

plane of bending. Out of plane bending will be investigated in detail in the next chapter

due to its complexity. The numerical equations proposed at the end of this chapter to

quantify the uni-axial bending are the first of their kind, known to the author, on this

matter.

6.2 Behaviour of the impacted columns under single axis bending

Figure 6.1: Plan view of the column head (under uni-axial bending) The main objective of this chapter is to investigate the impact behaviour and capacity

change in the impacted columns under eccentric loading conditions. The columns


6-161

under single axis bending were considered in the first phase. Moment was applied

about Y axis so that the impact induced deflection and the deflection due to the

moment are acting in the same direction (see Fig. 6.1). This is the most vulnerable

situation for columns under single axis bending. At a later stage, the impact behaviour

of the column will be investigated by applying the moment in the opposite direction.

6.2.1 The load application procedure

The vulnerability of typical columns adequate in capacity for 5 to 15 storied buildings

with load eccentricities are investigated in the analysis. Axial load and bending

moments are applied with load reduction factors according to the AS3600 code.

Several options are considered at the initial stage of the numerical simulation of the

eccentric loading. There is no single standard way of applying a moment across a

section so that the applied moment distribution along the column follows the rules of

moment distribution. Therefore, indirect ways of applying the moment are

investigated in the numerical simulation which does not influence the initial

conditions of the impacted column. For instance, applying a displacement to a plane or

a section of the column to generate the moment introduces an artificial shear

distribution across the column. Similarly, generating a moment by assigning a moment

directly to a set of selected nodes is also not possible as the hexagonal elements

without rotational degrees of freedom used in this analysis do not allow proper

moment distribution. Consequently, the moment simulation was performed by using a

coupling action of axial loads applied on a bulk head of the column (see Fig. 6.1 & 6.2).

The bulk head of the columns was designed to have flexural and shear strengths well

exceeding the external loads applied on the columns. This was to ensure that failure

occurred in the column as intended. Hence, the top face of the column was projected

0.5m at 450 to form the lower part of the head and then projected another 0.25m in the

vertical direction. Bearing plates were placed on the top of the head so that the load

application area and eccentricity of the load were certain. The axial load required to

produce the moment was applied as an eccentric load while the remaining load was

applied directly on the plate at the centre. This bulk head reduced the stress

concentration in the concrete due to the applied eccentric load.


6-162

Figure 6.2: The Bulk head of columns used for the application of moment

6.2.2 Material models and mesh generation

Due to symmetry of the structure and the loading, only half of the column was

modelled and out of plane displacements were eliminated along the symmetric edge to

maintain the symmetric conditions. The column consists of three separate parts; the

column, column head and bearing plates, according to their intended purposes. The

circular column was simulated using quadratic solid elements with a rectangular core

area to minimise errors due to element distortion, as described in Chapter 5. Uniform

element distribution across the height was maintained as this mesh generation derived

better results. According to Xie et al. (1996), the behaviour of the concrete cover is

significantly different from the core concrete. Therefore, at least two elements were

used to simulate the thickness of the cover concrete. The column head was simply a

projection of the top surface and hence its mesh distribution was very similar to that of

the circular section. The mesh generation of the vertical portion of the head was

conducted in a similar manner. It is important to note that the sizes of the elements

increase towards the top of the head as the diameter increases. In fact, deformation of

this portion, due to the load application, is negligible as the inertia is high compared to

the circular portion. Hence ‘Rigid material’ characteristics were assigned to the bulk

head, which is known as a very cost effective material as far as the duration of the

analysis is concerned. The rigid material behaviour is mesh independent and this will

also allow global or local constraints to be applied to the mass centre. This is an ideal

way to transfer the bending action from the head to the column without applying

constraints directly on the column, as this may cause unusual stress concentrations as

well as constraint column rotations. However, the effective height of the column was

slightly changed due to the bulk head as it translates and rotates about its centre of


6-163

gravity. Thus the centre of gravity was maintained at a certain height for all column

heads for comparison purposes. It is also observed that the initial deflection at the

head-column joint is negligible at the time of impact, as will be discussed later in

detail.

Both the longitudinal and transverse reinforcement were assumed to be

elastic-perfectly plastic. The same material parameters used for steel in the Chapter 5

were used in this simulation and complete strain compatibility was assumed between

the embedded bars and the concrete.

6.2.3 Axial load and eccentric load applications in an explicit environment

Figure 6.3: Numerical simulation of eccentrically loaded columns

Monotonically increasing axial compression was applied on the columns with constant

end eccentricity during the simulation. The eccentric load simulating moment was

gradually increased followed by the direct axial load as a ramped up surface pressure

over the bearing plates to avoid stress fluctuation. This procedure also avoids the

premature flexural failures due to the eccentric load alone acting on the column, while

minimising the potential effects of cyclic loading as the eccentric load is applied. In

particular, a separate ramp function has to be defined for applying the moment. One

important observation is that the pitch of the fluctuation is minimised substantially

when the duration of the ramp function (eg. 8ms) exceeds twice the duration of the

fundamental period of vibration of the column (4ms). However, the optimum ramp

duration reduces as the column diameter increases.


6-164

Three different hoop spacings (50mm, 100mm, 250mm) and three different hoop

diameters (6mm, 8mm, 12mm) were considered in the analysis without altering the

yield strength ( 'yf =500MPa). However, lateral ties were not included as they may

complicate the problem, and the addition of cross ties at a fixed volumetric ratio may

or may not improve the confinement effects. That is, cross ties will increase the

transverse steel spacing, while improving the confinement effects in the lateral

direction.

Figure 6.4: Interaction diagram for the 300mm Figure 6.5: Extreme strain in the 300mm

column column

Figure 6.4 shows the nominal interaction diagram of a 300mm column made of 30MPa

concrete. Interaction diagrams indicate the nominal column capacities under most and

least favourable confinement characteristics according to AS3600, including two other

options in between. However, the design capacity remains well below the nominal

capacity due to the usage of capacity reduction factors at the design stage. The load

eccentricity e, may vary in the range of about 0.02 to 0.25 which corresponds to the

axial load ratio of 20% to 80% compared to the concentric loading conditions which

includes the upper level of load eccentricity used in practice.

According to Figure 6.5 the maximum strain in steel is increased with low amount of

steel content. Therefore contribution of the longitudinal steel could be limited in

damage mitigation of columns under transverse impact loads. For instance, Figure 6.6

shows a typical interaction diagram of a typical circular column. It can be subdivided

into compression control and tension control zones as shown. Characteristic points

such as pure compression, balanced failure and pure bending can be identified using

the relation to the strain in extreme tension steel. The strain in extreme tension steel


6-165

reaches a maximum value under the zero axial loading conditions where pure bending

takes place. Columns with low steel ratios show higher tensile strain development and

fail more quickly by crushing the concrete at the compressive side. Yielding of

longitudinal reinforcement can be expected at the collapse stage depending on the test

variables.

Figure 6.6: A typical Interaction diagram of a column

6.3 Deflection profiles and resultant bending moment

The axial loading system consists of two vertical loads applied on the bearing plates

that simulate axial load and the bending moment. Half the design load is applied on the

bearing pads as one half of the column is used in the numerical simulation. To

investigate the accuracy of the selected method, 64kNm moment and 407kN axial

force were applied on the 300mm column, which is equal to 20% of the design load

under pure compression. According to the theory of moment distribution, half the

applied bending moment (BM) must be transferred to the fixed end with an opposite

sign (see Fig. 6.7 (a)). Under the impact loading the column generate cracks

distributed through the column while shear is critical at the bottom. Consequently the

column failed under flexural shear critical conditions.


6-166

(a) Moments at serviceability stage (b) Ultimate stage after impact

Figure 6.7: Contours of effective plastic strain

Figure 6.8 shows the moment generated some of the selected cross sections (CS) at the

top, bottom and at the point of impact. It can be seen that the numerical simulation

accurately follows the conventional theories of moment distribution and the

fluctuation of the applied moment can be minimised by choosing a desirable duration

for the ramped up loading function. The reason for the slight difference between

theoretical moment transferred and the actual moment present at the top is the elevated

position of the centre of gravity of the bulk head where rotation takes place and hence

external moment is fully applied. That is, the moment at the top of the column slightly

deviates from the applied moment. However, this minor deviation can be negligible

under low elevation impacts where the moment at the bottom is the governing factor

which influences the overall behaviour of the column under impact.

Figure 6.8: Time histories of BM of 300mm eccentrically loaded half column with 1% steel


6-167

6.4 Deformation characteristics of the impacted column

Figure 6.9: Deflection of 300mm diameter column with 4% steel at the near collapse stage

Figure 6.9 compares the deflection characteristics of the 300mm column under load

eccentricities and pure compression. It is observed that the deflection characteristic at

the ultimate stage increases with application of the moment. Residual deflection also

increases with the associated fluctuation under eccentric loading conditions. The

maximum deflection was observed in the 300mm column, and the 600mm column has

the least residual deflection as the mode of failure changes from flexure to flexural

shear as the column diameter increases.

6.5 Impact behaviour of the eccentrically loaded column

Figure 6.10: Resultant bending moment at different locations on the 450mm column

As far as the impact generated bending moments are concerned, the resultant bending

moment diagrams follow the same triangular loading pattern which simulates vehicle


6-168

impact. Under pure axial loading the residual moment of the impacted column reaches

zero. However, it was observed that the moment at the top of the column remains

steady without changing while the moment generated at the bottom and at the point of

impact increases to a greater extent during the impact and reaches its initial valve at the

residual stage. These findings are summarised in Figure 6.10. This observation implies

that the post impact behaviour of the column remains unchanged. If adequate shear

reinforcement is provided, an impacted column can display the same static lateral

capacity of an ordinary column (Loedolff’s 1989). Nevertheless, under eccentric loads,

the impact capacity reduces due to the secondary moments applied by the enhanced

residual deflection generated by the impact at the ultimate stage. However it can be

seen that the dynamic moment (325kNm) generated under the impact does not have a

substantial influence even though it exceeds two times the ultimate (nominal) static

moment capacity of the 450mm column under the given axial load. Thus,

implementation of SDoF system to quantify the impact behaviour is doubtful as

localised strain rate effects and the corresponding strength enhancement may not be

detected using a SDoF system. Also, considerable shear force variation generated

between the point of impact and bottom support cannot be captured by such a

simplified method.

Figure 6.11: Resultant shear forces at different locations on the 450mm column

The shear force generated at the bottom support under pure axial loading is larger than

the shear force at the point of impact as shown in Figure 6.11. Consequently there is a

gradual variation of shear force between the point of impact and the bottom support,

which is unlikely to occur under static loading conditions. Thus the critical stress


6-169

distribution changes linearly in between those two sections and combination of

moment and shear force are involved in determining the failure mode of the columns.

Additionally, the initial deflection of the eccentrically loaded columns greatly reduces

the ultimate shear force generated at the bottom. The stiffness characteristics of the

column change with the eccentric load and thus eccentrically loaded columns are

increasingly vulnerable and collapse under low velocity impacts. In other words, shear

capacity under pure axial loads greatly reduced with the application of the moment.

However, the residual value of the shear force remains unchanged from its original

value after the impact, confirming the observation of Loedolff (1989).

6.6 Behaviour of eccentrically loaded confined columns under impact

a) Concentric loading b) Eccentric loading c) Near Failure conditions

Figure 6.12: Lateral pressure distribution and the corresponding stress-strain relationships

Even though the validation process mainly focuses on the vulnerability assessment of

axially loaded columns, it can be extended to assess the vulnerability of eccentrically

loaded columns. The main difference between these two loading conditions is the

formation of a stress gradient across the sections of the eccentrically loaded column as

shown in Figure 6.12(b) & (c), which influences the lateral strain distribution based on

the Poisons ratio of concrete. The lateral strain distribution will determine the intensity

of the confinement in the lateral direction and consequently it decides the failure point

due to the hoop fracture (Mander 1988). Thus it may not be always possible to apply

the theories generated under concentrically loaded (uniform strain) conditions to


6-170

eccentric loading conditions in impact simulations. However, based on the

observations of (Mander 1988, Lokuge 2003, saatcioglu 1995), it can be concluded

that the results of this numerical simulation can be extended to assess the vulnerability

of the eccentrically loaded columns made of the lower grade concrete.

In fact, it was considered that the stress-strain curves in the strips closer to the neutral

axis may not be substantially different to the one that under fully confined conditions

particularly in the preloading strain range (Figure 6.12(b)) (Saatcioglu 1995). On the

other hand, the strength decay is essentially a function of confinement stress and does

not vary with the strain gradient (Sargin 1971). Moreover, it is important to note that

the flexural cracks appearing on the column at the ultimate stage under the impact will

minimise the stress differences in various layers across the section. Consequently, the

vulnerability assessment techniques are extended to the eccentrically loaded columns

by assigning uniform confined compressive characteristics to the core section based on

the equations proposed by Mander et al. (1988).

6.7 Selection of the load combinations

Figure 6.13: Interaction diagrams for 450mm column according to AS3600 and ACI: 318

Figure 6.13 shows the P-M interaction diagrams generated for circular columns that

are subjected to bending moment and axial compression for two extreme lateral

reinforcement ratios. Figure 6.14 represents specific points, including pure axial

compression, balanced point and pure bending. Axial tension is neglected as it has

limited applications. The interaction diagram depends on the configuration of the

longitudinal reinforcement particularly for a circular column. However, for


6-171

convenience, the amount of reinforcement is expressed as the percentage of gross

concrete area which generally varies from 1% to 4% according to AS 3600. The

number of reinforcement bars in a cross section is 8, 12 and 16 for 300, 450 and

600mm column respectively. Nominal interaction diagrams for 1% and 4% steel ratios

are represented in Figure 6.13 along with their design values and strength reduction

factors which vary from 0.6 to 0.8 according to the AS3600 standard. It is interesting

to note that allowable load at the serviceability stage is significantly high in ACI: 318

(1999) compared to AS 3600 (2004). As the Accidental Limit State was defined in

between the ULS and SLS, the structural design according to the ACI: 318 (1999) (and

for other codes) must be investigated separately under the impact loading.

Figure 6.14: locations of the selected loading points on the interaction diagrams

6.8 Parametric studies and discussion of the finite element results

During their service life, concrete columns experience numerous loading conditions

due to load eccentricities, differential settlements or external loads such as wind. The

impact behaviour of such columns could be complex due to various possible load

combinations and hence difficult to predict. To simplify the numerical simulation,

columns with single axis bending are considered in the first phase and the effects are

further subdivided into positive and negative moments depending on the direction of

the moment application. The negative moment will deflect the column in the direction

of the impact and reduce the impact capacity, while positive moments do the opposite.


6-172

6.8.1 Impact behaviour of eccentrically loaded columns under maximum allowable capacity

The load combinations applied on the impacted columns are selected based on the

corresponding interaction diagrams. Columns under pure axial load are represented as

P=Pd and no moments are acting on those columns. The other load combinations are

represented by the ratio of the axial load to the pure design axial load capacity of the

column and the corresponding moment. A point on an interaction diagram is

represented using a notation such as P=0.2Pd ; M=M20 which means the column carries

20% of its pure (design) axial load and the total moment corresponding to that axial

load according to the AS3600 (2004) standards (see Fig. 6.14). The selected loading

points are marked on the interaction diagram and sensitivity of the moments is further

investigated by reducing the moment by half. The selected load combinations

represent to the load carrying capacities of columns with 1% and 4% steel.

Figure 6.15: Eccentrically loaded columns with 1% steel ratio The vulnerability of impacted columns with 1% steel is illustrated in Figure 6.15. The

peak force represents the critical (maximum) impact force that can be withstood by the

impacted columns. It is clear that the bending moment present in the columns

substantially reduces the impact capacity compared to its pure axial load capacity,

despite the columns carrying low axial loads. The 20% loaded (P=0.2Pd) column with

the corresponding full moment will reduce the impact capacity by about 50%, while

the impact capacity reduction of columns with 50% axial load (P=0.5Pd) with the

corresponding full moment is approximately 33%. Therefore load combinations


6-173

beyond this point may not be recommended. It is also evident that the 450mm columns

with 20% loading are now under threat of the low velocity (<40km/h) vehicle impacts.

6.8.2 Impact behaviour under reduced load eccentricities

Figure 6.16: Peak force under different load combinations

One of the reasons for the impact capacity reduction under the flexural loading

conditions could be the full moment (eg. M50) applied on the columns under the

corresponding 0.5Pd axial loading. However, the effects of reduced eccentric loading

were yet to be determined. Substantial impact capacity improvement may be expected

due to the reduction of the depth to the neutral axis under the reduced moments where

extreme concrete fibre is subjected to lower compressive strains.

Consequently, the analysis extended to investigate the impact capacity of eccentrically

loaded columns with partial moments. In the process, the maximum allowable

moment on the column was reduced by 50% (eg. 0.5M50) while maintaining the same

axial load (0.5Pd) as shown in Figure 6.14. However, it was evident that reducing the

moment does not have a considerable effect on the impact behaviour of the columns.

This was partly due to the small moment present (transferred) in the column close to

the bottom support where lateral impact force was applied. Based on this observation,

columns with reduced eccentric loading were omitted in further analysis. Thus the

empirical equations were generated by applying axially load with the maximum

allowable moment. This reduces the large number of load combinations that are

involved in the analysis process.


6-174

As the eccentricity of the axial load decreases, the impact capacity of the columns

increases as shown in Figure 6.16 except for the 50% loaded 300mm column. The

moment reduction is more effective for columns with low axial loads rather than

columns with high axial loads, as far as the resultant impact capacity improvement is

concerned. On the other hand it is important to note that the 20% and 50% loaded

columns carry almost equal bending moments even though there are significant

differences in their impact capacities (see Fig 6.14). That is, when the applied moment

remains constant, the impact capacity of the columns significantly increases with axial

loading. A simple calculation based on load eccentricity revealed that the compressive

stress generated on the column is significantly high with the axial load increase

compared to the moment reduction. The enhanced axial load increases the stiffness of

the column and hence increases the impact capacity to a greater extent. In particular,

the column’s sensitivity to the axial load depends on the duration of the impact and this

relationship is effective only if the duration of the transverse impact is equal or greater

than the time taken by the buckling process particularly under flexural failure

conditions.

6.9 Impact behaviour of columns under positive eccentric loading

Positive eccentric loading is a hypothetical term used to describe a moment on a

column which deflects the column towards the impacted side. It was generated by

applying pressure on the same bearing plates in the upward (opposite) direction. It is

important to note that most of the columns in building edges, supporting eccentrically

loaded beams located perpendicular to the edges, are subjected to similar (positive)

load eccentricities and most of the columns therefore generate counter moments

against the impact. The aim of this study was to investigate the effects of the counter

moment in the vulnerability analysis.

Figure 6.17 presents the effect of the positive moment on capacity enhancement. The

positive moments always enhance the impact capacity of the columns compared to the

negative moments irrespective of the diameter of the column. The initial deformation

was in the impacted face of the column and as a result the column resists the impact


6-175

induced deflection. Consequently, compressive stresses dominate the impact

behaviour of the columns by increasing the stiffness. Thus, deformation during the

transverse impact can be controlled by applying a positive moment on the column and

the resultant changes to the stiffness can increase the capacity against lateral impact.

Figure 6.17: Comparison of the Impact capacities under positive and negative moments In the comparison point of view, flexure is predominant throughout the column under

negative bending moments compared with under positive moments, in which limited

area is subjected to flexural conditions. The resultant maximum deflection of the

column is also high under the negative moments. These factors imply that the strength

enhancement due to the strain rate effects is high under negative moments compared

with under positive moments. However the strength enhancement due to the strain rate

effects is not reflected in the results. Therefore the main factor behind this strength

enhancement under the positive moment should be the change in stiffness due to

formation of an arch. Therefore the results will still be conservative if the positive load

combinations are excluded from future analyses. However, the capacity reduction due

to the bending stresses cannot be compensated by the strength enhancement due to the

formation of the arch. Consequently the maximum impact capacity is obtained under

the pure axial loading conditions (Pd, Mo).

Typical cracks that appeared at three consecutive time steps in the impacted columns

with positive moment are shown in Figures 6.18(a) and (b) along with 6.19(a). This

behaviour is different from double curvature bending and the continuous oscillation of

the impact generated moments along the column leads to this crack pattern. The


6-176

motivation of this oscillation is the axial load fluctuation of the impacted 300mm

column from 180% to 20% compared to the applied axial load. The intensity of the

load fluctuation decreases with the diameter of the columns. According to Papazoglou

and Elanashai (1996) the shear failure is likely when accounting for axial load

increment followed by vertical ground motion particularly under earthquake loading.

However, the axial load fluctuation and considerable preliminary deflection initiate

the flexural failure conditions in the 300mm column. Consequently, impact generated

shear force at the bottom of the 450mm column is 3.5 times higher than that of the

300mm column even though the ratio of the cross sectional areas and the ratio of

critical amplitudes of the impact are 2.25 and 1.67 respectively. This observation

implies that the impacted columns tend to fail in shear as their diameter increases,

despite the eccentric loading conditions. This hypothesis is confirmed by the impact

characteristics of the 600mm diameter column which failed in shear as shown in

Figure 6.19(b).

(a) During impact (b) Post impact (a) 450mm column (b) 600mm column Figure 6.18: Cracks on 20% loaded 300mm Figure 6.19: Cracks on 20% loaded 450mm and column with 1% steel 600mm columns with 1% steel 6.9.1 Impact response under positive eccentric moments

As far as the impact induced shear forces are concerned, there are no specific changes

due to the directional change of the moment except the impact generated shear forces

act in the opposite direction compared with the shear forces applied at the service

stage. The resultant shear force at the service stage is comparatively small and shear

Primary cracks

Secondary cracks

Tertiary cracks


6-177

forces generated by the impact are concentrated close to the bottom support. These

forces vary linearly from 20kN at the point of impact to 105kN at the bottom support,

at the near collapse stage for the 300mm column with 20% loading. It is important to

note that the shear force as well as the bending moment at the bottom of the support

simultaneously increases with the impact (see Fig. 6.20). Therefore a combination of

these two should be taken into account when determining the failure mode.

Figure 6.20: Resultant bending moments of the 20% loaded 300mm column

6.10 Confinement effects on eccentrically loaded columns under impact

Figure 6.21: Capacity of eccentrically loaded confined columns under impact The main objective of this section is to investigate the capacity (peak force) of

eccentrically loaded confined columns under the impact loading conditions. As the

peak forces generated by the positive eccentric moments are conservative, positive

load combinations are excluded from here onwards. Mainly two eccentric loading


6-178

conditions namely (0.5Pd, M50) and (0.2Pd, M20) are investigated and compared with

axially loaded conditions (Pd, Mo) under nominal confined conditions (see Fig. 6.21).

The nominal confined condition is achieved by proving 6mm hoops at 250mm spacing

as the strength enhancement of the concrete due to this arrangement can be negligible.

At the next stage, the hoop spacing ‘s’, is decreased down to 100mm and 50mm

respectively and the resultant confined conditions are simulated by assigning

enhanced strength characteristics to the core concrete according to the equations of

Mander et al. (1988).

It is observed that the 20% loaded (0.2Pd, M20) concrete columns are always highly

vulnerable under these circumstances compared to the 50% loaded columns (0.5Pd,

M50), irrespective of the hoop spacing. Consequently, it is not recommended use larger

diameter columns than required to mitigate the impact damage (Note: M20 ≈ M50). In

fact, impact capacity can be significantly enhanced by providing hoops at 100mm

spacing. The impact capacities of the 50% loaded column and axially loaded (Pd, Mo)

column are equal when hoops are provided at 50mm spacing for the 50% loaded

column. Consequently, when the axial load is greater than 50%, the impact capacity of

eccentrically loaded columns can be increased beyond the impact capacity of fully

axially loaded (nominally confined) columns by reducing the hoop spacing s. This is

not the particularly true for 300mm column due to flexure initiated failure conditions.

Thus, if the hoops can be provided at 50mm spacing, further analysis may not be

required for columns carrying over 50% of their full axial load and the corresponding

moment under single axis bending.

6.11 Effects of the longitudinal steel ratio on the impact behaviour of columns

In this investigation, the longitudinal steel ratio is increased by increasing the diameter

of the longitudinal reinforcement while keeping the diameter of the column and steel

configuration constant. Two load combinations namely (0.8Pd, M80) and (0.2Pd, M20)

were selected in the analyses along with the axially loaded column. The effective load

carrying capacity of the column increases with the longitudinal steel ratio even though

the axial stresses on the columns are maintained constant. It is observed that the impact

capacity of concrete columns increases as the longitudinal steel ratio ‘R’ increases


6-179

from 1% to 4%. This behaviour is contrast to that under (pure) axial loading

conditions.

Figure 6.22: Effect of longitudinal steel ratio on impact capacity

In detail, the columns with 1% steel fail prior during the impact compared to the

columns with 4% steel. The effects of steel ratio are considerable for load combination

below the balanced point (0.2Pd, M20) where flexure is predominant and hence steel

can yield under lateral impacts. On the other hand the longitudinal steel ratio has very

little effect closed to the pure axial loading conditions (0.8Pd, M80) where columns fail

primarily in shear. The lateral deflection of the concrete column increases with applied

moment and hence the moment has considerable effects on the failure mode of the

impacted columns. However the axial load on the column is the governing factor of the

impact capacity of these columns. Therefore it is concluded that the impact capacity

enhancement of the eccentrically loaded columns relates not only to tensile strength

but also to shear characteristics and fracture toughness. Consequently DIF of tension,

compression, shear and fracture energy must be taken into account for predicting the

impact capacity of eccentrically loaded columns.

6.12 Confinement effects on the impacted columns with high steel ratio

To investigate the effects of confinement, columns having 4% steel ratio with three

different lateral steel spacing (s = 250, 100, 50) were considered. As the 80% loaded

column (0.8Pd, M80) fail in shear critical conditions and the 20% loaded (0.2Pd, M20)

columns fail in flexural failure conditions, the effects of the confinement on each


6-180

column could be different. Therefore 20% and 80% loaded columns were selected in

the investigation. It is evident that the impact capacity increment due to the

confinement effects is 17% and 19% on average for 20% and 80% loaded columns

with hoops at s=50mm intervals respectively as shown in Figure 6.24. When the hoop

spacing increases up to 100mm the corresponding improvements are 8% and 11%

respectively. This means that the hoop spacing can improve the impact capacity of the

columns equally irrespective of the failure mode. In other words strain rate effects

accompanied with flexural failure modes have equivalent effects compared to the

shear capacity enhancement of the columns. Consequently DIF for strain rate and

shear may be equal under eccentric loading conditions where critical sections of the

column section are under low compression. Having provided that the higher mode of

vibration does not take place due to a vehicle impact, the enhanced capacity of the

columns may be directly calculated by multiplying the impact capacity under nominal

confined conditions by a factor within 1.08 to 1.19 within this axial loading range.

Figure 6.23: Impact capacity enhancement due to confinement

6.13 A comparison of the confined columns with different steel ratios

According to the above analyses the impact capacity of columns are minimum for the

axial load combinations below the balanced point. However the following analyses

revealed that the impact capacity could be even low when the steel ratio drops to 1%

from 4%. For instance, considerable capacity drop encountered for 250mm hoop

spacing under 20% axial loading (0.2Pd, M20). The columns with 4% steel ratio have

the highest capacity particularly for 20% loaded columns. The main reason for this


6-181

observation is the flexure initiated failure conditions of the 20% loaded column despite

the steel ratio where steel can yield and contributes to the capacity enhancement.

Figure 6.24: Effects of the longitudinal steel ratio on impact capacity enhancement

6.14 Effects of the slenderness ratio and intensity of loading on impact capacity


0.250.751.251.752.252.753.253.75

3.0 5.0 7.0 9.0 11.0 13.0Slenderness ratio

Pea

k fo

rce

(MN

)

300mm; P=20% 300mm; P=50%450mm; P=20% 450mm; P=50%600mm; P=20% 600mm; P=50%


0.00

0.50

1.00

1.50

2.00

2.50


Pea

k fo

rce

(MN

)


(a) 50MPa concrete with 1% steel (b) 30MPa concrete with 1% steel

Figure 6.25: Effects of the slenderness ratio on capacity enhancement for columns

The impact capacity improvement with the concrete grade (50MPa and 30MPa) and

axial load intensity are shown in Figure 6.25. The 20% and 50% loaded columns

{(0.2Pd, M20) and (0.5Pd, M50)} were selected in the analyses. It is evident that there is

no appreciable improvement of the impact capacity with reduction of the slenderness

ratio irrespective of the concrete grade. A similar conclusion has been made for

circular columns with low steel ratio (2.1%) subjected to the combined action of

bending and torsion under earthquake loading conditions (Prakash et al. 2010).

Despite the torsion the failure was predominantly flexural. According to the present

analyses the capacity enhancement remains almost the same even when the steel ratio

increases to 4%, although the failure mode changes from flexure to shear with the


6-182

improved longitudinal steel content. Therefore substantial capacity improvement

cannot be expected by reducing the slenderness ratio. However, substantial

improvement can be obtained with the enhancement of the axial load intensity on the

columns. For instance, the impact capacity substantially increases when the axial load

increases from 20% to 50% even though the enhancement slightly reduces with the

effective height.

6.15 Strain rate sensitivity of eccentrically loaded columns

Figure 6.26: Strain rate sensitivity of a ductile column Strain rate sensitivity of the eccentrically loaded columns could be different from the

axially loaded columns due to the change of failure mode from shear to flexural shear

with the application of the moment. To investigate this hypothesis, impact pulses

ranging from 50ms to 150ms were applied on axially and eccentrically loaded (0.2Pd)

300mm columns (Figure 6.26). Amplitudes of the impact at near failure conditions

were recorded over the 50ms and 150ms duration impacts. The improvement of the

impact capacity is 18% for the column under eccentric loading conditions, compared

with only 7% for the axially loaded column. This is a significant improvement of the

impact capacity as there is initially a 70% reduction of the capacity due to the eccentric

loading conditions. Even though the overall capacity is reduced, eccentrically loaded

columns are less vulnerable to hard impact conditions and thus the impact capacity

will be proportionately enhanced with the stiffness of the vehicle. Additionally the

kinetic energy stored in the column decreases from 1200J to 700J due to the initial

deflection.


6-183

6.16 Linear equations for 20% and 50% loaded columns

Having provided the impact capacities of columns for a range of different parameters,

empirical relationships, based on the last square method, are developed in this section.

These relationships can be used to quantify the critical impact force and the associated

impulse for eccentrically loaded columns. Residual analysis is used to demonstrate the

accuracy of the predicted equations and for further improvement of the accuracy of the

outcomes.

Residuals vs Pred Log Py = -0.4064x

2 + 4.8565x - 14.455

R2 = 0.4651

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15

5.0 5.5 6.0 6.5 7.0Predicted Log P

Re

sid

ua

ls

Residuals vs Pred Log Py = -0.5038x

2 + 6.1004x - 18.401

R2 = 0.4542

-0.15-0.10-0.050.00

0.050.100.15

5.0 5.5 6.0 6.5 7.0Predicted Log P

Re

sid

ua

ls

Figure 6.27: Residuals of the predicted Figure 6.28: Residuals of the predicted values (20%) values (50%) In fact, residuals were used in many procedures designed to detect various types of

disagreement between data and an assumed model. In early days, the residual analysis

was used only to produce stronger compelling conclusions. However, interest in

residual analysis was renewed later by developing methods for assessing the influence

of individual parameters. For example, the scatter plot of residuals versus fitted values

that accompanies a linear least square fit is a standard tool used to diagnose

nonconstant variance, curvature, and outliers. In order to identify such deficiencies

associated with the data points, different residual based expressions were used. In this

analysis, the residuals are calculated as (Observed (Log P) - Predicted (Log P)) and

hence the positive and negative residuals indicate an under and over-prediction of the

data points respectively (see Fig. 2.27 & 6.28). The important point is, a polynomial

plot to the residual valves can be used to improve the accuracy of the results further

and consequently, the corrected Log Pc and Log Ic are given by the Equations 6.2 and

6.3.


6-184

6.16.1 Linear equations for 20% loaded columns The linear equations were derived based on least square method to assess the

vulnerability of the 20% loaded columns subjected to impacts under negative eccentric

moment in the plane of bending.

Eq. 6.1

Eq. 6.2

Eq. 6.3

Where Log P is the logarithm of the Peak Force P (uncorrected), Log Pc is the

(corrected) logarithm of the Peak Force P, Log Ic is the (corrected) logarithm of

Impulse I, D is the diameter of the column in m, ρv is the longitudinal steel ratio, H is

the height in m and Ah is the area of hoop in mm2. Even though the yield strength of

hoops, 'syf and hoop spacing, s included in the parametric analysis they are not appear

in the final equation. This means that effect of the above two terms are negligible

compared to the other parameters. With the introduction of the corrected equation, the

over and under prediction of the Peak Force, Pc is reduced up to ±10%.

6.16.2 Linear equations for 50% loaded columns

Similar set of equations can be derived for 50% loaded columns with the

corresponding moment as given in Equations 6.4 to 6.6. Once the critical impulse for

20% and 50% loaded columns are calculated, linear interpolation can be used to

calculate the critical impulse for columns loaded in-between. Impact capacity of the

column under pure axial loading already discussed in Chapter 5 so that comprehensive

vulnerability assessment for all the loading points located on the interaction diagram

are possible.

Eq. 6.4

Eq. 6.5

Eq. 6.6

The valid range of the equations is given as follows;

6.4001.0856.2051.0'008.0721.0 +++−+=h

ADHcfvPLog ρ

30.1

401.1810.7)(504.0 2

−=−+−=

cc

c

PLogILog

PLogPLogPLog

53.4001.0885.2051.0'007.0645.1 +++−+=h

ADHcfvPLog ρ

30.1

455.14856.5)(406.0 2

−=−+−=

cc

c

PLogILog

PLogPLogPLog


6-185

6.17 Conclusions

The main aim of this chapter was to analyse the impacted columns under single axis

bending in order to identify the severity of the moment present in the columns. The

following conclusions can be drawn from the findings. A sequence of axial load

combinations were used in the analyses and it was observed that the impact capacity

reduction (or enhancement) is gradual for the considered loading range. This allows

linear interpolation to be implemented in the vulnerability assessment process.

1. Consequently, linear equations are generated to predict the critical impulse for two

consecutive points on the interaction diagram namely; (0.5Pd, M50) and (0.2Pd,

M20). Having provided the impact capacity of axially loaded columns, a

comprehensive vulnerability assessment is possible for all the intermediate load

combinations (maximum allowable design) on the interaction diagram.

2. Reducing the moment down to 50% (eg. 0.5Pd, 0.5M50) without changing the

axial load (0.5Pd) has only minor effects on the impact capacity of the columns.

On the other hand, columns subjected to positive eccentric loading generate

conservative results compared to columns subjected to negative eccentric loading.

Therefore reduced eccentric loading conditions and columns under positive

eccentric loading can be excluded from the analysis process.

3. It was observed that the impact capacities of 50% loaded (0.5Pd, M50) confined

columns with a corresponding moment, and 100% axially loaded (Pd) columns are

equal if hoops are provided at 50mm spacing. Thus the hoop spacing alone can

recover the capacity drop due to eccentric loading by up to 50% of the design axial

msm

mmAmm

MPafMPa

mHmMPafMPa

mDm

b

sy

c

v

25.0050.0

1.11327.28

500250

425030

04.001.0

60.03.0

22

'

'

<<<<<<

<<<<<<

<<ρ


6-186

load. However, it is not recommended to use larger diameter columns than that

required to mitigate the impact damage.

4. The impact capacity of the columns increases with the longitudinal steel ratio of

the columns. The effects of the longitudinal steel ratio are predominant in lightly

loaded columns where considerable deflection takes place. Overall the results

have shown that the yielding of the steel as well as the shear characteristics and

fracture toughness of the concrete have considerable influence on the impact

behaviour of the eccentrically loaded columns. Additionally, the DIF for strain

rate and shear may be equal under eccentric loading conditions where the critical

sections of the columns are under low axial compression.

5. The impact induced bending moment is predominant close to the bottom support

and reaches its initial values at the residual stage. However, the peak moment

generated in the columns during the impact does not have a significant effect even

though it exceeds two times the ultimate static moment capacity of the columns.

The resultant shear force also follows a similar behaviour and therefore it is

concluded that the impacted column can display the same static lateral capacity as

an undamaged column. However, the initial deformation due to load eccentricity

greatly reduces the ultimate dynamic shear capacity of the impacted columns.

6. The cracks generated under a positive moment and the second mode of vibration

are identical for smaller diameter columns even though the mechanisms behind

the two processes are totally different from each other. Additionally, the

combination of moment and shear force must be taken into account when

determining the failure mode. This cannot be identified by analysing a simplified

SDoF system under equivalent static loads.

7. Higher axial load intensity on columns would be a better solution for impact

capacity enhancement compared to height (slenderness) reduction. Even though

the overall capacity is reduced under eccentrically loaded conditions compared to

axially loaded columns, eccentrically loaded columns are less vulnerable to hard

impact conditions where the duration of the impact is less than 100ms.


7-187

7. IMPACT ON COLUMNS UNDER BIAXIAL BENDING 7.1 Introduction

Generally, structural columns are loaded with biaxial bending, rather than pure axial

compression. Biaxially loaded columns in underground car parks, overpass bridges

and medium to low rise buildings located close to major roads are extremely

vulnerable to vehicle impacts due to the initial deformation present in the columns.

The bi-axially loaded columns behave in a complex manner in response to the impact

loading at a shear critical height due to the complexity of the load combinations.

However, there exits minimal data on dynamically tested columns under biaxial

bending (Zahn et al. 1989) while the existing experiments on bi-axially loaded

columns are limited to quasi-static or pseudo-dynamic loading such as earthquake

(Wong et al. 1993). On the other hand, most of the experiments focused on nonlinear

flexural deformation which is usually decoupled from shear (Xiao and Zhang 2007)

and torsion (Prakash et al. 2010) even though columns in mild tension or reduced

compression have the tendency to reduce the shear capacity (Papazoglou and Elnashai

1996). In fact, presence of torsion along with shear and bending initiates shear failure

(Prakash et al. 2009). These observations implied that the impact capacity of bi-axially

loaded columns may be substantially different from the columns under pure axial

loading. On the other hand, impact induced torsional moments can affects the flow of

internal forces and deformation capacity of impacted columns particularly when the

impact force is applied perpendicular to the direction of bending by changing the

failure mode. Therefore equation developed to quantify the impact capacity under

pure axial loading or uni-axial bending may not be adequate in routine column design.

Moreover, design codes (AASHTO-LRFD 1998; EN 1991-1-7:2006) and existing

guidelines (Tsang et al. 2005) do not account for the bi-axially loaded columns under

lateral impact loading. Consequently, there is a pressing need for the development of

some simplified yet rational guidelines to quantify the impact capacity under biaxial

bending.

To address this issue, extensive numerical simulation have been conducted to study the


7-188

response of biaxially loaded impacted columns and the results used to develop

empirical equations to quantify the critical impulse. Having provided that the

vulnerability assessments of the eccentrically loaded columns are independent of

confinement characteristics resulting from strain gradient, the validation process was

extended to assess the vulnerability of biaxially loaded columns. The parametric study

is limited to 30MPa to 50MPa concrete by avoiding High Strength Concrete (HSC).

The 300mm to 600mm diameter columns adequate in capacity for 5 to 15 storey

buildings with the longitudinal steel ratio ranging from 1% to 4% are investigated in

detail. The structural design was based on the Australian Standard AS3600 (2004).

Additionally, dependency of the duration of the impact on the enhanced stiffness

characteristics of the columns due to axial load variation was neglected in the analysis.

In particular, impact angle was taken into account in the analyses and full column was

used in the numerical simulations where the moment application lead to

unsymmetrical loading.

One of the main objectives was to identify the severity of the moments present in those

columns. However, the analyses process for biaxially loaded concrete columns under

lateral impact is difficult due to large number of load combinations involved in the

analyses process. If the influence of the configuration of the longitudinal steel is

negligible, the bi-axially loaded columns can be numerically simulated by using two

eccentric axial loads acting in two orthogonal directions. Thus, the effects of the

direction of the impact within 0o-90o can be accounted by varying the eccentric load on

the two perpendicular planes. The remaining impact angles were treated separately

and the non-critical cases excluded from the analysis to simplify the process.

Consequently, the entire analyses were focused on two load combinations on the

interaction diagram namely; (0.5Pd, M50) and (0.2Pd, M20) where Pd is the design axial

load capacity of the column and M50 and M20 are the corresponding moments as shown

in Figure 7.1. The aim was to define three consecutive points on the interaction

diagram along with respective critical impulses so that linear interpolation can be used

to quantify the critical impulses for points in-between (see Fig. 7.1). In the process,

effects of longitudinal and lateral steel ratios, concrete grade, direction of the impact,

strain rate sensitivity of columns and slenderness ratio were also quantified.


7-189

Figure 7.1: Impact capacity prediction for intermediate load combinations

7.2 Numerical simulation of biaxial loaded columns

A typical interaction diagram derived for circular columns under biaxial bending is

shown in Figure 7.2. Cases (a) and (b) are the uni-axial bending about the two

principle directions, X and Y. The interaction curve represents the failure envelope for

various combinations of axial loading and bending moments. As far as the biaxial

bending of the column is concerned, the methods of equivalent uni-axial eccentricity

will give a better understanding of the depth and inclination to the neutral axis. Once

the depth and inclination are determined, the corresponding interaction curve can be

easily established. By establishing such curves for various radial distances L, the

failure or allowable stress surface for biaxial bending can be constructed. The

interaction diagram may be constructed by interpolation of the uni-axial bending cases

if the differences introduced by the configuration of the lateral steel are neglected.

Based on this assumption, the neutral axis can be taken as perpendicular to the

direction of eccentricity of the resultant force. Thus different values for the bending

moments are selected along the maximum allowable service stress contour for one

particular axial load. Two of the selected values (Cases (a) and (b)) are located directly

on the uni-axial bending planes X and Y, while the other one is located in-between

(Case (c)). The Case (c) represents the resultant allowable moment which requires one

particular steel ratio about the X or Y axis. According to these observations, the

influence of the longitudinal steel configuration on the orientation of the neutral axial

is higher for small diameter columns, and reduces as the diameter increases.


7-190

Figure 7.2: Typical interaction diagram for circular columns under biaxial bending

The biaxial loading was introduced to the columns by using two eccentric loads acting

in two perpendicular directions X and Y (see Fig. 7.3(a) and (b)). The plate placed at

the centre of the bulk head was used to apply the axial load, while the eccentrically

placed plates were used to apply the moments. Since the end block was modelled to

have excessive shear and flexural strengths, rigid material characteristics were

assigned to the bulk head. The plates had known surface areas and eccentricities, thus

allowing known eccentric loads to be applied about each axis. Fully fixed conditions

between the head and plates were assumed, to avoid loss of contact during the lateral

deformation of the column during impact. The moments were applied after the axial

load by using two separate ramp functions to avoid flexural failure conditions at the

load application stage. The impact force on the column was applied parallel to the X

direction after the vertical loads stabilised. The impact can be considered as oblique as

far as the resultant eccentric load on the column is concerned. The uni-axial load

combinations were also taken into account along with the biaxial moments for

comparison purposes.


7-191

(a) Front elevation of the biaxially loaded column (b) Plan view of the top

Figure 7.3: Numerical model of the column under biaxial bending Due to non symmetry of the axial loading, the entire column was modelled in the

numerical simulation instead of using a half model (see Fig. 7.3(a)). A 50mm cover

was provided for the reinforcement and closed hoops having yield strength 350MPa

were varied from 250mm to 50mm intervals. Complete strain compatibility was

assumed between the embedded steel bars and concrete, while an elastic perfectly

plastic material model was used for both the longitudinal and transverse

reinforcements. Axial loads were applied as ramp up surface pressures while

translations and rotations were constrained at the bottom nodes of the columns to

simulate fixed end conditions. Movement of the bulk head was constrained only in the

lateral directions X and Y, while allowing rotations about both these directions so that

it could move in the vertical direction and deflect as the column deformed.

7.3 Characteristics of the simulated columns

Columns having diameters from 300mm to 600mm with 1% steel were considered in

the initial stage of the analysis while 6, 8 and 12 longitudinal steel bars were equally

distributed along the perimeter of the 300mm, 450mm and 600mm columns

respectively. The effective height of the columns was 4m and 50MPa concrete with

nominal transverse reinforcement was used. Approximately 25mm long hexagonal

solid elements with one point integration used for concrete while 25mm long beam


7-192

elements with 2×2 Gauss integration were used for longitudinal steel. The bulk head

was discretized so that the mesh generation of the head was compatible with that of the

column along the lateral direction. The direction of the displacement of the column is

not exactly in the X direction as in pure axial loading. Thus an additional set of nodes

in the perpendicular direction is needed to obtain the torsion based deformation

pattern.

Table 7.1: Biaxial load combinations on the 300mm column under 50% axial load

Load combination

0.5Pd (kN)

Mxs

(kNm) Mys

(kNm) Steel ratio

(X-X) Steel ratio

(Y-Y) 1 1000 0 47 0.40 1.12 2 1000 18 40 0.40 1.14 3 1000 37 37 0.40 1.25 4 1000 40 18 1.14 0.40 5 1000 47 0 1.16 0.40

The structural design is based on the Australian Standard AS3600 (2004). Moments

Mxs and Mys were applied about the X and Y directions of the columns so that the

longitudinal steel requirement along the major axes remained close to 1%. During the

initial analysis, 50% of the design axial loading capacity (0.5Pd) was maintained with

the corresponding maximum moment (M50) so that eccentricities in loading

comprehensively agree with the interaction diagram. The design axial load capacity Pd

was calculated from Eq. 4.1. Even though this analysis was conducted by assuming

500MPa for longitudinal steel, the results can be extended to other steel grades by

using the equivalent steel area method proposed by Shi et al. (2008). This allows the

impact behaviour of the columns to be investigated under conditions in which flexural

failure predominates. Typical load combinations and steel ratios for the 300mm

column are given in Table 7.1. The load combinations may occur even in a three storey

building column and hence a comparison based on the number of storeys is not

considered.

7.4 Selection of load combinations

7.4.1 Impact capacity of columns under biaxial bending

Figures 7.4(a) to (c) compare the impact capacities of the columns for the five different

combinations of moments about the X and Y axes, which bring the columns to the

ultimate limiting stress under lateral impact conditions. As far as the 300mm column is


7-193

concerned, the impact capacity is highest when the moment acts about the Y axis.

However, for the 450mm and 600mm columns the maximum capacity is obtained

when the moment acts about their X axes even though the difference is small. In

addition, the minimum capacity is obtained when the moment is applied perpendicular

to the direction in which the maximum capacity is obtained for each column,

particularly with 0.5Pd loading. Under the remaining allowable biaxial moment

combinations, the impact capacities vary almost linearly between the maximum and

the minimum. This is evidence of the accuracy of the theoretical approach described

in Figure 7.2 for determining the capacity of the columns under biaxial loading. Once

the impact capacity of the column is determined in two perpendicular directions, the

impact capacity under biaxial bending can be interpolated from the known capacities

under single axis bending. Here the impact force has to be applied along either the X or

Y axes.

1 2 3 4 5

05

101520253035404550

Impact capacity of 300mm column under biaxial bending

Peak force (x25kN) Moment about X axisMoment about Y axis

1 2 3 4 5

0

50

100

150

200

250

300


Peak force (x25kN) Moment about X axis

Moment about Y axis

(a) Impact capacity of the 300mm column (b) Impact capacity of the 450mm column

1 2 3 4 5

0

100

200

300

400

500

600

700


Peak force (x25kN) Moment about X axisMoment about Y axis

(c) Impact capacity of the 600mm column

Figure 7.4: Impact capacities of the columns under biaxial bending In fact, the impact capacities of the columns under each load combination do not


7-194

deviate considerably, contrary to what would be expected. The shear failure

characteristics do not allow the impacted columns to deform substantially, particularly

for larger diameter columns. Hence the secondary moments generated by the load

eccentricities due to the buckling effects are minimised. Therefore, the 450mm and

600mm columns behave identically under the impact loading. Consequently, their

maximum and minimum impact capacities under single axis bending in the two

perpendicular directions differ by only 6% for both columns. Thus, only the impact

capacity under single axis bending would be sufficient for preliminary design

calculations of the circular columns susceptible to shear failure conditions. In fact, this

is an indication of the dilute coupling action between the biaxial moments, in addition

to the coupling effects between axial force and bending moments, particularly for

shear critical impacted columns.

However, the response is substantially different for the 300mm column where

significant bending takes place with all the moment combinations. The impact force

enhances the lateral deformation of the column which had already initiated under the

single axis bending. With the substantial flexural deformation, the longitudinal steel

yields and enhances the impact capacity while activating the tensile strength of the

concrete. Consequently, the impact capacity of the column is greatest when the

moment is applied about Y axis. The minimum deflection takes place when the impact

force is perpendicular to the direction of bending (X) and the ratio between the

maximum and the minimum impact capacities is around 16% for the 300mm column.

The torsional moment generated during the impact was responsible for this deviation

other than the coupling action between the biaxial moment and axial load. Thus,

comprehensive investigation is recommended for the columns with flexural

characteristics.

It is also worth noting that the technique used to simulate the eccentric load on the

300mm column also contributes to the capacity reduction when the excessive lateral

deformation takes place. For instance, even though the lever arm used to generate the

moment is nearly the same for all columns, it produces secondary moments that would

increase the deflection of small diameter columns. Therefore this can be considered as

error induced by the modelling techniques and the error can be minimised by


7-195

increasing the lever arm. However despite all the deviations, three load combinations

including single axis bending about two perpendicular axes along with one

intermediate load combination are sufficient to predict the impact capacities of the

columns for the entire range of load combinations. Thus the number of load cases

reduces to three, which includes the one with equal moments about both perpendicular

axes.

As far as critical velocities are concerned, the 300mm column with the biaxial bending

will be vulnerable for velocities ranging from 12.5ms-1 to 14.5ms-1 (45 to 52 km/h)

generated by typical car impacts, while 450mm and 600mm columns in urban areas

will not be vulnerable to car impacts under these conditions.

7.5 Effects of the direction of the impact

So far the analysis has been conducted by assuming that the direction of the impact is

along either one of the major axes X or Y (see Fig. 7.5(a)). However a crucial situation

is encountered when the direction of the impact is in the same direction as the resultant

biaxial moment (see Fig. 7.5(b)). This is particularly the case when the equal (biaxial)

moments are applied about the two perpendicular directions where the vector

summations of the two moments may exceed the maximum allowable moment under

single axis bending. On the other hand, no firm prediction can be made due to the

variations in initial deflection and stress distribution under the biaxial bending. Hence

the analysis was extended to investigate this isolated case and Figure 7.5 compares the

plan view of the two column heads.

(a) Impact is along the X axis (b) Impact is along the resultant moment

Figure 7.5: Simulation of the effects of the direction of the impact


7-196

Columns made of 50MPa concrete having an effective height of 4m with 1% steel

were considered in the analysis. The initial simulations were conducted on 300mm,

450mm and 600mm diameter columns by assigning equivalent axial loads on the

eccentric plates to simulate the equal moments. The plates were placed at 45o on either

side of the impact so that the resultant force acted along the direction of the impact.

Comparison was made with single axis bending about the Y axis (see Fig. 7.5(a)), as

the two scenarios are almost identical. The results revealed that the impact capacity

under the symmetric biaxial moments is always higher than the capacity under single

axis bending about the Y axis. The resultant capacity increase varies from 3% to 1.2%

for the 300mm and 600mm columns respectively. It is also evident that all the columns

fail in flexure, and the resultant deflection characteristics are identical for each column

under single axis bending (Y axis) and biaxial bending. Based on these results it is

concluded that the oblique impacts do not significantly contribute to the vulnerability

of the columns. Consequently the vulnerability under oblique impact loads can be

predicted by applying the impact force along the principle (X or Y) axes. Thus this

observation significantly reduces the number of cases that must be taken into account

in the analysis process, particularly for circular columns under biaxial bending.

7.6 Effect of reduced axial load on biaxial bending

In the next stage the axial load was reduced from 0.5Pd to 0.2Pd and the impact

behaviour under biaxial loading was investigated with corresponding moments (0.2Pd,

M20) (see Fig. 7.1). According to the results, except in the 300mm diameter column,

there is considerable capacity reduction in the 450mm and 600mm columns. Moreover,

the impact capacities under biaxial bending and single axis bending about the Y axis

are equal in the 20% loaded 450mm column. In contrast, the capacity under biaxial

bending exceeds the capacities under single axis bending for the 300mm and 600mm

columns. Here the biaxial moment consists of equivalent maximum moments applied

about both axes. On the other hand, the 600mm column is slightly more vulnerable

under uni-axial bending when the moment is applied about the Y axis while the

300mm column is more vulnerable when the moment was applied about the X axis.

Consequently it is quite difficult to explain the exact reasons for this behaviour

particularly with the 20% axial loading, due its interaction with torsion, secondary


7-197

moments and axial load fluctuation during the impact.

In fact, when the stress reaches the elastic limit under an axial force and biaxial

bending, the columns tend to generate curvatures about each principle axis. The

initiation for the translation and rotation is more significant for small diameter

columns than for larger diameter columns, particularly after cracking occurs under the

impact. Once the cracking begins, the axial force and the bending moments acting on

that section will tend to shift towards the gross section with an additional secondary

moment resulting from the eccentric loading. The secondary moment will depend on

the amount of deformation and consequently the deformation about each principle axis

of the gross section is affected by the respective moment about each axis. As the

biaxial load combinations depend on the axial load level, the impact behaviour of the

columns is changed with the axial load level. On the other hand, the hypothetical

neutral axis of the deformed column due to the impact does not coincide with that of

the axial force and biaxial moments. Hence there is no simple technique which can

predict the impact behaviour of the column under biaxial bending by interpolating the

capacities under single axis bending, particularly for the small diameter columns

where the effects of secondary moments and torsion are severe due to the larger

deformations and relatively small gross area remaining in the column after cracking.

In fact, the 300mm column under biaxial bending with 1% steel has almost the same

capacity under 20% and 50% loading even though there is a substantial difference in

the ultimate deflection between each case The ultimate deflection increases from

20mm to 55mm as the axial load reduces from 50% to 20%, and this is the expected

behaviour in flexure dominated columns. However, the enhanced strain rate due to the

deflection does not always compensate the capacity reduction resulting from the

diminished axial load. For instance, the impact capacity is reduced in the 20% loaded

300mm column with single axis bending (either about the X or Y axis) despite its

enhanced deflection characteristics. However, despite all these differences, the impact

capacity of the column under biaxial bending can be conservatively taken as the

minimum of the capacities under single axis bending without further investigation. In

particular, this comment applies when the direction of the impact is along one of the

principle axes of bending, and it can be applicable to both 50% and 20% loading.


7-198

7.7 Effects of longitudinal steel ratio on biaxial bending

The effects of the longitudinal steel ratio were investigated by increasing the steel ratio

from 1% to 4% without changing the configuration. The axial load on the column was

0.5Pd and the corresponding biaxial moment was applied on columns having

diameters of 300mm, 450mm and 600mm. Increasing the steel ratio to 4% enhances

the axial load capacity by 20% to 30%, and the moment capacity by 40% to 50%

compared to the columns with 1% steel.

When the steel ratio increases to 4%, the impact capacity of the 300mm column

increases around 17%, while the 450mm and 600mm columns show 3% and 5%

reduction respectively. As far as the 300mm column with 1% steel is concerned, the

impact capacity primarily results from the flexural-shear behaviour of the column

where the longitudinal steel can yield under flexure and contribute to the capacity

enhancement (see Fig. 7.6(a)). However, the 300mm column with 4% steel fails in

shear due to the extra stiffness introduced by the additional steel content. Similarly, the

450mm and 600mm columns also stiffen with the 4% steel, and hence during the

impact the longitudinal steel buckles due to shear deformation and the columns fail

soon after this event (see Fig. 7.6 (d)).

Figure 7.6: Failure characteristics of 50% loaded 300mm columns under biaxial bending

The 450mm and 600mm columns always failed by shear despite the biaxial moment

induced deflections or the effects of the longitudinal steel ratio. The longitudinal steel

buckles prematurely due to shear failure induced lateral deformations. Consequently,

there is no substantial contribution from the longitudinal steel to the impact capacity of


7-199

the columns. The results also show that the resultant deflection characteristics also

reduce with the increase of the longitudinal reinforcement. Consequently, the extra

axial load and moment applied on the column, which are supposed to be carried by the

additional steel content, will transfer to the concrete once the longitudinal steel fails.

Thus the columns with high longitudinal steel ratios are more susceptible to shear

failure and hence their impact capacity will be further reduced. Therefore, increasing

the longitudinal reinforcement ratio will not always improve the impact capacity of the

columns under biaxial bending.

In general, all the columns with 4% steel fail without developing their full flexural

capacity due to premature buckling of the longitudinal steel at an early stage of the

impact. Consequently, the 17% capacity enhancement observed in the 300mm column

is due to the change of the failure mode from flexure to shear. However it may not

always be true to expect that there is a capacity enhancement when the failure mode

changes from flexure to shear as in the 300mm column. The motives are different in

this particular case where there is extra axial load and moment applied on the column

with 4% steel compared to the column with 1% steel, which may alter the internal

stresses at the ultimate stage even though theoretically the axial stress on the concrete

is the same.

7.8 Effects of biaxial bending on 20% loaded impacted columns with a 4% steel

ratio

The mode of failure also depends on the amount of the axial load present in the column

at the time of impact. For instance, columns with 20% axial load may tend to fail in

flexure while columns with 50% axial load may tend to fail in shear. Therefore it is

worth identifying the impact behaviour of the column with low axial loading. However,

enhancing the longitudinal steel will further stiffen the columns as discussed in the

previous paragraph. Consequently, the impact behaviour of lightly loaded columns

with a high steel ratio leads to a complex situation. In addition, higher concrete grades

are also responsible for brittle failures and hence the effects of biaxial bending on 20%

loaded impacted columns made of 50MPa concrete are investigated in the following

section.


7-200

The same columns having diameters from 300mm to 600mm with 4% steel are

selected and their impact capacities are compared with 50% loaded columns. In

general, the shear failures of the 50% loaded columns with 4% steel transform into

flexural failures under 20% axial loading and corresponding moment. The flexural

failure of the 300mm column occurs without any longitudinal steel buckling. Even

though the longitudinal steel yields at the time of failure, the impact capacity reduces

by 30 to 40% compared to the columns with 50% loading. The reason for this is the

substantial reduction of the axial stress in the 20% loaded columns, and the conditions

further worsen due to the corresponding moment which is nearly two times that under

50% loading. The initial deflection of the column is an indication of its vulnerability,

and is more predominant under single axis bending about the Y axis where the

deformation is in the same direction as the impact. Thus it represents the most critical

case. As far as 300mm column is concerned, applying the moment about the X axis is

also more critical than applying a biaxial moment. The secondary moments generated

by the load eccentricity that corresponds to the moment about the X axis is the reason

for this observation.

The impact capacities of the 450mm and 600mm columns also decrease under the 20%

loading. It is observed that there is longitudinal steel buckling followed by flexural

deformation of the 450mm columns. This signifies the transition stage of the failure

mode from flexure to shear depending on the diameter of the column. Thus the 600mm

column always fails in shear as expected despite the axial load reduction. Thus far the

analysis has confirmed that the impact capacity under bi-axial bending can be

conservatively taken as the minimum of the two capacities under single axis bending if

the impact force is applied along one of the principle axes of bending. The calculated

value based on this assumption may be increasingly over-conservative when the

percentage of the axial load decreases, the diameter of the column decreases or when

the longitudinal steel ratio decreases.


7-201

7.9 Damage mitigation of the impacted columns under single axis bending

Peak force vs Diameter

0.25

0.75

1.25

1.75

2.25

2.75

3.25

0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65Diameter (m)

Pea

k fo

rce

(MN

)

X-X; 1%; P=20% X-X; 1%; P=50%X-X; 4%; P=20% X-X; 4%; P=50%


0.25

0.75

1.25

1.75

2.25

2.75

3.25

0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65Diameter (m)

Pea

k fo

rce

(MN

)

Y-Y; 1%; P=20% Y-Y; 1%; P=50%Y-Y; 4%; P=20% Y-Y; 4%; P=50%

(a) Moment about X-X axis (b) Moment about Y-Y axis Figure 7.7: Peak force vs. slenderness ratio for 4m high columns made of 50MPa concrete

The impact capacities of the columns bending about two orthogonal axes discussed so

far are summarised in Figures 7.7(a) and (b). The figures reflect the effects of the

diameter on individual columns under the varying steel ratio and axial stresses. It can

be seen that the impact capacity reaches its maximum and minimum when the columns

with 1% steel carry 50% and 20% loads respectively. The impact capacity of columns

with 4% steel is in-between the capacities for 1% steel, particularly for larger diameter

columns. It is important to note that both the axial load and moment carrying capacity

of the columns increase with the longitudinal steel content. Thus, 20% loaded larger

diameter columns with 4% steel are more effective compared to the columns with 1%

steel. However the opposite is true for 50% loaded columns. At this stage it would be

better to rely on axial loading rather than the steel content for damage mitigation as the

axial stress increment generated the highest capacity enhancement.

7.10 Effects of the confinement on biaxial bending

7.10.1 Impact behaviour of 50% loaded columns with 4% steel under biaxial

bending

An investigation was conducted to improve the ductile characteristics of 50% loaded

columns with 4% steel, by providing transverse steel at a closer spacing where the

columns failed without developing their full flexural capacity due to premature

buckling of the longitudinal steel. The hoop spacing was reduced from its nominal

value of 250mm to 50mm and capacity enhancement due to the closer hoop spacing

was simulated by assigning enhanced concrete characteristics to the core concrete by

using the equation proposed by Mander et al. (1988). Single axis bending about the X


7-202

and Y axes was also considered for comparison purposes. The impact force was

applied along the X axis as shown in Figure 7.5. The impact capacities are compared

with 50% loaded columns with nominal confinement.

With the enhancement of the confinement characteristics, the impact capacity of the

300mm column increased by 20% and 30% when the moment was applied about the X

and Y axes respectively. Therefore the strength enhancement due to the confinement is

not proportionate to the initial impact capacity of the column under nominal confined

conditions. According to the numerical results, the ductile characteristics of the

impacted column are improved with the confinement particularly when the moment is

applied about the Y axis. However, the enhancement of the ductility may not be the

only reason for the observed enhancement of the impact capacity of the column. In fact,

the failure mode also changes from shear to flexural shear due to the confinement

effects. Consequently the steel buckled at a later stage only at one point compared to

the nominally confined columns where the longitudinal steel buckled at two separate

sections simultaneously. In addition, the location of the shear failure plane moved

further downwards while the remaining part of the column was subjected to flexural

conditions. Thus, the 20% capacity enhancement occurred mainly due to the

enhancement of ductility and a change of the failure mode. However, the 30% capacity

enhancement mainly results from the enhancement of ductility alone as there is no

significant change in the failure mode. Thus, in general, the reasons behind the

capacity change of the 50% loaded 300mm column are two-fold. The first reason is the

change of the failure mode from shear to flexure, and the second reason is the

enhancement of ductility due to the confinement effect.

As far as the 450mm and 600mm columns are concerned, there is no considerable

variation of the capacity enhancement compared to the 300mm column. The average

enhancement is around 15% and remains constant despite the axis of bending as there

are no changes of the shear failure mode due to the confinement effects. Therefore the

capacity enhancement solely depends on the shear capacity enhancement.

Consequently is it concluded that the ultimate shear capacities of the columns do not

depend on the orientation of the moment even though the resulting shear forces may

increase with the confinement effects.


7-203

7.11 Behaviour of 20% loaded confined columns with 4% steel under biaxial

bending

Impacted columns with 4% steel were investigated to identify the confinement effects

on partial loading conditions which are associated with biaxial bending moments. The

moments are also applied about the X and Y axes separately for comparison purposes

while the impact force is applied along the X axis. The impact capacities are compared

with 20% loaded columns with nominal confinement. The results showed that the

impact capacities of the columns significantly reduce with the 20% loading. The

capacity enhancement due to the confinement is around 16% on average for all the

20% loaded columns despite the diameter. The capacity enhancement of the 20%

loaded 300mm columns is reduced compared to that of the 50% loaded 300mm

columns. However, as far as the 450mm and 600mm columns are concerned there is a

slight increase of the capacity compared to that under 50% loading. Therefore there is

no general rule that can be used to determine the capacity enhancement due to the

confinement effects as it depends on the diameter and loading conditions of the

columns.

All these individual cases were scrutinised to identify the source of the variation. In

fact, the 300mm columns fail in flexure without significant shear deformation despite

the level of confinement and the axis of bending. Consequently the strength

enhancement is solely due to the longitudinal steel yielding under flexural deformation.

On the other hand there is no change in the failure mode for the 450 and 600mm

columns due to the confinement. Thus the capacity enhancement is constant for the

20% loaded columns with 4% steel.


bending

The effects of the confinement were investigated for columns with a low steel content.

The strength enhancement was compared with nominally confined 50% loaded

columns with 1% steel. It was observed that the results follow the conventional

theories of confinement effects. For instance, there is a gradual decrease of the


7-204

capacity enhancement as the diameter increases. However there are no proper

relationships as far as each axis of bending is concerned. The average capacity

enhancement in the 300mm column is around 24% and gradually decreases to 18%

and 16% for 450mm and 600mm columns respectively. This observation arises from

the fact that the strength enhancement due to the confinement effects does not

effectively change the mode of failure of the columns. The failure mode is

predominantly flexure irrespective of the diameter of the columns. Moreover, there is a

considerable ductility enhancement with the confinement effects particularly for

300mm columns. The considerable ductile behaviour with low steel content is a

unique feature of the bi-axially loaded impacted columns.


bending

The investigation continued with the 20% loaded confined columns with 1% steel. The

6mm hoops made of 350MPa steel were placed at 50mm intervals. The capacity

enhancement was compared with nominally confined columns with 20% loading. The

behaviour of 20% loaded columns totally differs from 50% loaded columns. For

instance, the average capacity enhancement is 6%, 17% and 29% for 300mm, 450mm

and 600mm columns respectively. This indicates an increase of the capacity

enhancement with the increase of column diameter.

The 300mm column with 1% steel achieved a substantial ductile capacity with 20%

loading which cannot be improved further by providing hoops at closer intervals.

Therefore the capacity enhancement due to confinement is only marginal. However,

the limited ductile characteristics of the 450mm and 600mm columns substantially

improved with the closer hoop spacing. Thus the enhanced ductility increases the

capacity of those columns to a considerable level.

7.14 Effects of the steel grade and diameter of the hoops

So far the investigation has been conducted by assuming either nominally confined

concrete columns with 250mm hoop spacing or confined concrete columns with

50mm hoop spacing. Instead of changing the hoop spacing, the impact capacity can


7-205

also be increased by increasing the yield strength or the diameter of the hoops. In fact,

it was identified that the collapse load rapidly increases as the hoop spacing decreases,

particularly less than 100mm. Therefore 100mm is considered as the optimum hoop

spacing. Thus the impact capacity enhancement of the columns was estimated by

changing the yield strength and diameter of the hoops while keeping the hoop spacing

at 100mm. The diameter of the hoops was changed from 6mm to 12mm while the yield

strength of the hoops was changed from 250MPa to 500MPa.

Impact capacity vs Hoop spacing

0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0 50 100 150 200 250 300 350 400

Hoop spacing (mm)

Pea

k fo

rce

(MN

)

300mm; X-X 450mm; X-X 600mm; X-X300mm; Y-Y 450mm; Y-Y 600mm; Y-Y

Figure 7.8: Impact capacity of 20% loaded columns under varying hoop spacing

A comparison of collapse loads of 300mm, 450mm and 600mm columns under

varying hoop spacing for 20% loading is shown in Figure 7.8. The capacity is highest

when the moment is applied about the X axis, which is parallel to the direction of the

impact. The difference between the collapse loads for bending about the X and Y axes

increases with the diameter of the column. Peak force is substantially higher for hoops

spacing less than 100mm and peak force is decrease as hoop spacing increase.

Moreover, the impact capacity under biaxial loading with equal moments about both

axes either falls in-between the peak forces under single axis bending (ei. X and Y), or

else exceeds the peak forces under single axis bending.

Other than the biaxial moment, the axial load plays a major role in determining the

impact capacities of the columns. The impact capacity increases with axial load and

the corresponding moment on the column. For instance, the impact capacity increases

from 16% to 52% in 600mm and 300mm columns respectively, when the axial load

increases from 20% to 50% along with the corresponding moment. The capacity


7-206

increase due to axial load is insensitive to the hoop spacing as shown in Figure 7.9.

Thus the medium biaxial loadings do not effectively enhance the confinement effects.

Impact capacity vs Hoop spacing

0.00.51.01.52.02.53.03.54.0

0 50 100 150 200 250 300 350 400Hoop spacing (mm)

Pea

k fo

rce

(MN

)

300mm; X-X; 20% 450mm; X-X; 20% 600mm; X-X; 20%300mm; Y-Y; 20% 450mm; Y-Y; 20% 600mm; Y-Y; 20%300mm; X-X; 50% 450mm; X-X; 50% 600mm; X-X; 50%300mm; Y-Y; 50% 450mm; Y-Y; 50% 600mm; Y-Y; 50%

Figure 7.9: Impact capacities of 20% and 50% loaded columns under varying hoop spacing

The continuous lines in Figure 7.10 express the impact capacity enhancement of the

20% loaded columns when the diameter of the hoops is increased from 6mm to 12mm.

The highest capacity enhancement is obtained by the 300mm diameter column even

though its impact capacity does not depend on the axis of bending. On average, 12% to

35% capacity enhancement is possible by increasing the bar diameter alone. The

impact capacity enhancements of the 50% loaded columns are shown by the dotted

lines in Figure 7.10. Columns carrying larger axial load and corresponding moment

have the highest capacity irrespective of the diameter of the columns. From a

comparison point of view, the relative capacity enhancement is highest when the hoop

diameter is increased compared with when the hoop spacing is reduced. Hence

increasing the diameter of the hoops is more effective than reducing the hoop spacing.

Impact capacity vs Hoop diameter

0.00.51.01.52.02.53.03.54.0

5 6 7 8 9 10 11 12 13Hoop diameter (mm)

Pea

k fo

rce

(MN

)


Figure 7.10: Impact capacities of 20% and 50% loaded columns under varying hoop diameter


7-207

Columns with 50% axial loading are always safe compared to those with 20% loading

despite the axis of bending. This behaviour is common for various load combinations

studied so far by changing the confinement effects. However, the capacity

enhancement obtained by changing the yield strength is only marginal. The average

capacity enhancement due to yield strength is lower than the enhancement obtained by

varying either the bar diameter or hoop spacing. On average it varies from 5% to 7%

(see Fig. 7.11). Therefore, increasing the yield strength is not recommended as a

damage mitigation technique for impacted columns. On the other hand, the potential

for yielding of the lateral reinforcement is reduced with the reduction of the axial load

(Watson et al. 1994). Consequently, hoops with a yield strength exceeding 500MPa are

not recommended for columns subjected to impact loads. Therefore, the maximum

capacity enhancement is gained by changing the diameter of the hoops, or

alternatively the hoop spacing can also be changed.

Impact capacity vs Yield strength

0.00.5

1.01.5

2.02.5

3.03.5

200 250 300 350 400 450 500 550

Yield strength (Nmm-2)

Pea

k fo

rce

(MN

)


Figure 7.11: Impact capacities of 20 and 50% loaded columns under varying yield strength On the whole, the direction of the impact still has considerable effect particularly for

the confined columns. When the moment is applied in two orthogonal directions, the

percentage variation of the capacity is around 15% on average for 600mm columns. In

general, the difference is marginal for 300mm columns and therefore the direction of

the impact has only minor effects on the impact capacity of the 300mm diameter

columns.


7-208

7.15 Effects of slenderness ratio on impact capacity of columns under biaxial

bending

The effectiveness of the reduction of slenderness ratio as a damage mitigation

technique is also of interest, in particular for smaller diameter columns. The effective

height of the columns was reduced over a range from 4m to 2m. Figures 7.12 and 7.13

represent the peak impact force that can be withstood by the columns with different

slenderness ratios and steel ratios, and the capacity enhancements due to increasing the

axial load intensity. The axial load is represented as a percentage of the design load

capacity of the columns and the corresponding moment is applied on the column in

two perpendicular directions. Figures 7.12(a) and (b) and Figures 7.13(a) and (b)

represent the impact capacities when the moment is applied about the X and Y axes

respectively. The load combinations consist of maximum moment about one axis and

zero moment about the perpendicular axis simultaneously. The impact pulse is applied

along the X direction as usual.

The allowable moment on the 300mm column slightly increases due to the reduction

of the effective height. However, the allowable moments on the 450mm and 600mm

columns are not affected by the effective height. As a whole, the columns are relatively

less vulnerable if the axial load and corresponding moment increase simultaneously.

Conversely, the columns are more vulnerable if the moment is applied on an axis

perpendicular to the direction of the impact (i.e. Y). In fact, the percentage increase of

the impact capacity resulting from a decrease in the slenderness ratio significantly

varies with diameter and direction of the applied moment. Therefore, a firm

conclusion cannot be drawn on the capacity enhancement due to slenderness. However,

reducing the effective height is more effective for small diameter columns. Thus, the

most reliable way to enhance the impact capacity is to maintain a high axial load and

corresponding moment on the columns. For instance, 50% loaded 450mm and 600mm

columns will enhance the impact capacity by around 24% on average compared to

20% loaded columns. Consequently lightly loaded columns may not always be safe

against impact loads. Moreover reducing the slenderness ratio will increase the shear

failure characteristics of the impacted column. Therefore, columns with high

longitudinal steel ratios are slightly less vulnerable compared to columns with low


7-209

steel ratios particularly for 2m columns.


0.250.751.251.752.252.753.253.75


Pea

k fo

rce

(MN

)



0.250.751.251.752.252.753.253.75


Pea

k fo

rce

(MN

)


(a) Moment about X axis (b) Moment about Y axis Figure 7.12: Peak force vs. Slenderness ratio for columns of 50MPa concrete with 1% steel


0.250.751.251.752.252.753.253.75


Pea

k fo

rce

(MN

)



0.25

0.75

1.25

1.75

2.25

2.75

3.25


Pea

k fo

rce

(MN

)


(a) Moment about X-X axis (b) Moment about Y-Y axis Figure 7.13: Peak force vs. Slenderness ratio for columns of 50MPa concrete with 4% steel The same data points were rearranged so that they reflect the dependency of the impact

capacity on the effective height as shown in Figures 7.14 and 7.15. In general, the

impact capacity can be increased by reducing the effective height. However, the

capacity enhancement obtained by reducing the effective height is limited for

eccentrically loaded columns compared to that of the axially loaded columns. For

instance, the respective average capacity enhancement for 600mm and 300mm

columns is around 15% to 25% under eccentric loading compared to 40 to 60%

enhancement under (pure) axial loading. Therefore, it can be concluded that the

reducing the height does not significantly reduce the vulnerability of the eccentrically

loaded columns. However, the columns with 4% steel tend to increase the impact

capacity at a higher rate as the height reduces compared to the columns with 1% steel.

Moreover, columns are more vulnerable despite the height when the moment is

applied about the Y axis, which is perpendicular to the direction of the impact.


7-210


0.250.751.251.752.252.753.253.75

0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65Diameter (m)

Pea

k fo

rce

(MN

)

X-X; 1%; P=20% X-X; 1%; P=50%X-X; 4%; P=20% X-X; 4%; P=50%


0.25

0.75

1.25

1.75

2.25

2.75

3.25

0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65Diameter (m)

Pea

k fo

rce

(MN

)

Y-Y; 1%; P=20% Y-Y; 1%; P=50%Y-Y; 4%; P=20% Y-Y; 4%; P=50%

(a) Moment about X-X axis (b) Moment about Y-Y axis Figure 7.14: Ultimate capacity of 3m columns made of 50MPa concrete


0.250.751.251.75

2.252.753.253.75

0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65Diameter (m)

Pea

k fo

rce

(MN

)

X-X; 1%; P=20% X-X; 1%; P=50%X-X; 4%; P=20% X-X; 4%; P=50%


0.25

0.75

1.25

1.75

2.25

2.75

3.25

0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65Diameter (m)

Pea

k fo

rce

(MN

)

Y-Y; 1%; P=20% Y-Y; 1%; P=50%Y-Y; 4%; P=20% Y-Y; 4%; P=50%

(a) Moment about X-X axis (b) Moment about Y-Y axis Figure 7.15: Ultimate capacity of 2m columns made of 50MPa concrete

7.16 Effects of the concrete grade on impact behaviour of columns

The impact behaviour of the eccentrically loaded columns made of 30MPa concrete is

investigated in this section by assigning 30MPa characteristics to the columns. The

load carrying capacity of the columns with 1% steel reduces by 36% compared with

the 50MPa concrete. In addition, the moment capacity of the column is also reduced by

the same amount. Interestingly, it was identified that the overall impact capacity

reduction is proportionate to the ratio of the concrete grades.


0.00.51.01.52.02.53.03.5

0.25 0.35 0.45 0.55 0.65Diameter (m)

Pea

k fo

rce

(MN

)

G30; 20%; X-X G30; 50%; X-X G50; 20%; X-XG50; 50%; X-X G30; 20%; Y-Y G30; 50%; Y-YG50; 20%; Y-Y G50; 50%; Y-Y


0.0

0.5

1.0

1.5

2.0

2.5

3.0

3.5

0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65Diameter (m)

Pea

k fo

rce

(MN

)

G30; 20%; X-X G30; 50%; X-X G50; 20%; X-XG50; 50%; X-X G30; 20%; Y-Y G30; 50%; Y-YG50; 20%; Y-Y G50; 50%; Y-Y

(a) With 1% steel (b) With 4% steel Figure 7.16: Comparison of peak force of 4m high columns made of Grade 30 and 50 concrete


7-211

Figures 7.16(a) and (b) compare the peak impact force for columns made of 30 and

50MPa concrete with longitudinal steel content of 1% and 4%. It can be observed that

the peak impact forces are slightly larger for columns with 4% steel. However, the

comparative advantages are insignificant and the bearing capacity of the columns is

also increased with the steel content. The failure load is low for columns made of the

lower grade of concrete. The highest and lowest average capacity enhancement are

observed in 50% loaded columns with 1% steel and 20% loaded columns with 4%

steel respectively when the grade of concrete is changed, irrespective of the axis of

bending. However, the difference between the maximum and the minimum rate of

enhancement is insignificant. In addition, the 300mm and 600mm columns show the

highest and the lowest capacity increase, due to the flexural and shear failure

conditions of the respective columns. Thus, the overall capacity enhancement due to

the grade 50 concrete is around 63% and this is proportionate to the ratio between the

two concrete grades. However, it is not recommended to adhere to this hypothesis

without further investigation, due to the capacity variations in individual columns.

Figures 7.17 and 7.18 represent the peak impact force that can be withstood by the

columns made of Grade 30 concrete with different slenderness ratios and steel ratios,

and the capacity enhancements obtained by increasing the axial load. It is evident that

the capacity enhancement due to the slenderness effects is significant for small

diameter columns and rapidly deteriorates for larger diameter columns. For instance,

the average capacity enhancement is around 56% for 300mm columns and 9% for

600mm diameter columns. The reason behind this enhancement is the flexural failure

of 300mm diameter columns which yields the longitudinal steel. Strain rate effects

may also have a significant contribution as the Dynamic Increasing Factor (DIF)

increases when the concrete grade reduces. Therefore reducing the height is only

effective for damage mitigation for eccentrically loaded small diameter columns made

of lower grade concrete.


7-212


0.00

0.50

1.00

1.50

2.00

2.50


Pea

k fo

rce

(MN

)



0.00

0.50

1.00

1.50

2.00

2.50


Pea

k fo

rce

(MN

)


(a) Moment about X-X axis (b) Moment about Y-Y axis Figure 7.17: Peak force vs. Slenderness ratio for columns of 30MPa concrete with 1% steel


0.00

0.50

1.00

1.50

2.00

2.50


Pea

k fo

rce

(MN

)



0.00

0.50

1.00

1.50

2.00

2.50


Pea

k fo

rce

(MN

)


(a) Moment about X-X axis (b) Moment about Y-Y axis Figure 7.18: Peak force vs. Slenderness ratio for columns of 30MPa concrete with 4% steel

7.17 Development of equations for biaxially loaded columns under lateral

impact

Structural columns are seldom designed for vehicle impacts due to inadequacies of

design guidelines. Empirical relationships are developed to bridge the gap based on

the least square method, which can be used to quantify the critical force and the

associated impulse for biaxially loaded columns. The empirical equations are

particularly valid under serviceability design load combinations. Effect of the reduced

load eccentricity can be neglected if it is within the 50% of the maximum allowable

moment under one particular axial load. In fact, direction of the impact was varied

from 0 to 900 by taking into account the most critical load combinations on two

mutually perpendicular axes. Having provided that the positive eccentric loading

conditions generated conservative results, the results can be extended to account for

the other impact angles.


7-213

7.17.1 Linear equations for 50% loaded columns (Impact angle 0o to 90o)

Partial Plot

-0.2

-0.1

0.0

0.1

0.2

-0.03 -0.02 0.00 0.02 0.03Steel ratio ρ

Log

PPartial Plot

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

-20 -10 0 10 20Compresive strength f' c

Log

P

(a): Partial regression plot of steel ratio (b): Partial regression plot of compressive strength

Partial Plot

-0.2

-0.1

0.0

0.1

0.2

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5Height H

Log

P

Partial Plot

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

-0.2 -0.1 0.0 0.1 0.2Diameter D

Log

P

(c): Partial regression plot of height (d): Partial regression plot of diameter

(e): Partial regression plot of Ah (f): Partial regression plot of 'syf


7-214

Partial Plot

-0.2

-0.1

0.0

0.1

0.2

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5

Impact angle (∆ /45)

Log

P

(g): Partial regression plot of Impact angle (h): Partial regression plot of hoop spacing

Figure 7.19(a-h): Partial regression plots of each parameter against Log P

A simple linear correlation between the steel ratiov

ρ , concrete grade 'cf (in N/mm2),

effective height H (in m), diameter of the column D (in m), area of the hoophA (in mm2),

and impact angle ∆ (in degrees) is determined by using a statistic program ‘StatistiXL’.

A total of 281 data records are used in the process. The Coefficient of Determination

(R2) of the data set indicates that 96% of the variation in Log P, is explained by

variation in the independent X variables, and the R value 0.98, which is the square roof

of R2, indicates a strong correlation between the Log P and independent X variables.

The Standard Error of Estimate, 0.085 is only 1% of the mean of Log P, 6.06 and thus

indicates that the Multiple Regression model has accurately calculated a large amount

of the Log P values.

The resultant linear regression expression is given in Equation 7.1. It can be used to

calculate the predicted Peak Force P of the critical impulse for a typical 100ms vehicle

impact.

Eq. 7.1

( ) 71.445003.0001.066.2043.0'007.0062.0 +∆−++−+−=h

ADHcfvPLog ρ


7-215

Residuals vs Pred Log P

y = -0.4077x2 + 4.9385x - 14.909

R2 = 0.3066

-0.20

-0.10

0.00

0.10

0.20

5.00 5.50 6.00 6.50 7.00

Predicted Log P

Re

sid

ua

ls

Figure 7.20: Residual of the predicted values

The accuracy of the predicted data points compared to the observed values in the

numerical simulation. Figure 7.20 represent the residuals that are calculated as

(Observed (Log P) - Predicted (Log P)). The positive and negative residuals indicate

an under and over-prediction of the data points respectively. By considering the

distribution of the residuals, second order polynomial equation was used to improve

the accuracy of the predicted Log P values. The final over and under prediction of the

Peak Force, P is within the range of ±11 %. The corrected Log Pc and Log Ic are given

by;

Eq. 7.2

Eq. 7.3

7.17.2 Linear equations for 20% loaded columns (Impact direction 0 to 90o)

Partial Plot

-0.20-0.15-0.10-0.050.000.050.100.150.20

-0.04 -0.02 0.00 0.02 0.04

Steel ratio ρ

Log

P

Partial Plot

-0.4

-0.3

-0.2

-0.1

0.0

0.1

0.2

0.3

-20 -10 0 10 20Compresive strength f' c

Log

P

(a): Partial regression plot of steel ratio (b): Partial regression plot of compressive strength

30.1

91.1494.5)(408.0 2

−=−+−=

cc

c

PLogILog

PLogPLogPLog


7-216

Partial Plot

-0.15

-0.10

-0.05

0.00

0.05

0.10

0.15

0.20

-2.0 -1.0 0.0 1.0 2.0Height H

Log

P

Partial Plot

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

-0.2 -0.1 0.0 0.1 0.2Diameter D

Log

P

(c): Partial regression plot of height (d): Partial regression plot of diameter

(e): Partial regression plot of 'syf (f): Partial regression plot of Ah

Partial Plot

-0.15

-0.05

0.05

0.15

-1.5 -1.0 -0.5 0.0 0.5 1.0 1.5Impact angle (∆ /45)

Log

P

(g): Partial regression plot of Hoop spacing s (h): Partial regression plot of impact angle

Figure 7.21 (a-h): Partial regression plots of each parameter against Log P

Similar set of expressions can be derived for 20% loaded columns with the

corresponding moment, as given in Equations 7.4 to 7.6. It was observed that 97% of

the variation in Log P, is explained by variation in the independent X variables while


7-217

Standard Error of Estimate, 0.063 is only 1% of the mean of Log P, 5.98. The accuracy

of the final over and under prediction of the Peak Force, P for the 20% loaded columns

is also within the range of ±11% with the improvement of the accuracy (see Fig. 7.22).

Residuals vs Pred Log Py = -0.3539x2 + 4.2308x - 12.605

R2 = 0.2794

-0.2

-0.1

0.0

0.1

0.2

5.0 5.5 6.0 6.5 7.0Predicted Log P

Re

sid

ua

ls

Figure 7.22: Accuracy of the predicted values

Eq. 7.4 Eq. 7.5 Eq. 7.6

Once the Corrected Impulse Ic is known, the critical velocity, v can be calculated for a

known impacted mass m (kg) of a vehicle, in meters per second (ms-1) by using the

relationship given in Equation 7.7. For instance, Eurocode EN 1991-1-7 (2006)

suggested that mean mass of 1500kg for cars and 20,000 kg for trucks.

mvI c = Eq. 7.7 The valid range of the terms in the equations is given as follows;

( ) 635.445006.0001.0698.2045.0'006.0881.0 +∆−++−+=h

ADHcfvPLog ρ

30.1

605.12231.5)(354.0 2

−=−+−=

cc

c

PLogILog

PLogPLogPLog

o

h

sy

c

v

msm

mmAmm

MPafMPa

mHmMPafMPa

mDm

900

25.0050.0

1.11327.28

500'250

4250'30

04.001.0

60.03.0

22

<∆<<<<<<<

<<<<<<

<<ρ


7-218

The critical impulses for the 50% and 20% loaded columns can be calculated from the

above equations while the impact capacity of axially loaded columns with nominal

load eccentricities can be calculated by the empirical equations provided in Chapter 5.

Hence, the critical impulses for three different locations (load combinations) on the

interaction diagram can be identified for one particular column. Consequently, linear

interpolation can be used to quantify the critical impulse for the loading points that are

located in-between on the interaction diagram. Having provided a known force and

impulse pair for an average impact duration of 100ms, this method can be extended to

assess the vulnerability of columns for a general vehicle population based on the

method suggested in Chapter 4. This involved an analytical method that can be used to

quantify the critical peak forces under different impact durations.

7.18 Conclusions

Design of RC columns for vehicle impact at shear critical height is a tedious task.

Understanding of the impact behaviour of the columns and a knowledge on the order

of importance of the key design parameters will have a significant contribution

towards the routine column design. This research has provided comprehensive

equations for this purpose in terms of the geometrical and material properties of the

impacted column. The outcomes will greatly facilitate researchers and design

engineers who desire to either cross check or validate their models. The main findings

of this chapter are summarised below.

1. Consideration of three load combinations is sufficient for the vulnerability

analyses of impacted circular columns under biaxial bending which include the

single axis bending about two orthogonal directions and the one with equal

moment about both principle axes. In fact, the impact capacity under single axis

bending would be sufficient for preliminary design of larger diameter columns

where shear failures are predominant while comprehensive investigation is

recommended for the columns that initiate flexural failure characteristics. In

particular, the application of the impact force should be along one of the major

axes of bending.


7-219

2. Vehicle impacts in the direction of the resultant moment can be considered as

non-critical (see Fig. 7.5(b)). In addition, reasonably conservative results can be

generated (from the equations) for load combinations that exceed 50% of the

allowable moment (0.5Mx) for the corresponding axial loading( )dxP . The columns

that deflect against the direction of the impact force are always safe compared to

their counterpart. Based on this conclusion the results generated from the equation

can be extended to account for the other impact angles.

3. Effects of the longitudinal steel ratio are marginal for damage mitigation

particularly for larger diameter columns. Thus the design (option) with low steel

content may be the best alternative solution. Alternatively, the impact capacity

rapidly increases with the intensity of the loading particularly under shear critical

conditions. Thus, it would be better to rely on higher axial load intensities with low

steel content when it comes to damage mitigation. When the effects of different

concrete grades are concerned, the overall impact capacity variation is

approximately proportionate to the ratio of the concrete grades.

4. Effectiveness of the confinement depends on axial load, axis of bending, diameter

of column and longitudinal steel ratio. As the resultant changes will depend on the

axis of bending, the resultant capacity enhancement due to the confinement for

each axis of bending may not be proportional to the initial impact capacities of

columns under unconfined conditions. This is particularly significant in small

diameter columns where the failure mode changes with the confinement.

5. Confinement effects are mostly effective for columns with low longitudinal steel

ratios as the deformation ductility will lessen with high longitudinal steel ratios

under low elevation impacts, while the resultant axial load enhancement further

increases the shear capacity of the column. However, the resultant impact capacity

enhancement due to the confinement effects could be insignificant in lightly

loaded small diameter columns with low longitudinal steel ratios as their ductility

cannot be further improved by the confinement effects.


7-220

6. From a comparison point of view, the relative capacity enhancement is highest

when the hoop diameter is increased rather than the hoop spacing. However,

increase of the yield strength may not be effective as a damage mitigation

technique.

7. The impact capacity enhancement due to reduction of the effective height is

substantially small in eccentrically loaded columns compared to that of the axially

loaded columns. This option particularly effective only for small diameter columns

made of lower grade concrete. In addition, even if the overall capacity is being

reduced, the eccentrically loaded columns under negative moments are less

vulnerable to the hard impact conditions.

8. Prediction of the impact capacities of the small diameter columns under biaxial

bending is rather complex as it depends on the load intensity, initial deflection,

secondary moment, torsional effects, coupling action, steel ratio and fluctuation of

axial load during the impact. Conversely, larger diameter columns mostly fail in

shear and hence their impact capacity is independent to a certain extent and thus

stable.


8-221

8. CONCLUSIONS AND FURTHER DEVELOPMENTS 8.1 Introduction

The main objective of this thesis was to generate design information to quantify the

vulnerability of columns subjected to vehicle impacts. In the process, a numerical

model of an impacted column was developed and validated using experimental results.

The validated numerical model was then used for parametric studies and development

of equations, so as to minimise the cost and time of physical testing.

8.2 Main contributions of the thesis

The most significant contributions arising from this research are listed below.

1. This research indicates that triangular impact pulses can be used in an impact

simulation process based on the results of full scale impact tests. Implementing

triangular impact pulses for frontal impact simulation is innovative as it

generates outcomes in terms of critical velocity (v) of the impacting vehicle.

The validated pulse characteristics also provide a cost-effective means of

performing parametric studies by using numerical simulation techniques when

conducting large amounts of full scale impact tests is difficult, if not

impossible.

2. The parametric studies were conducted in three stages for axially loaded

columns and for columns under uni-axial and biaxial bending. Partially loaded

columns were also investigated in detail. Numerical equations were provided

to quantify the peak force and the critical impulse for all load combinations

exceeding 20% of the allowable axial load, with the corresponding moment.

The equations are particularly valid for impact angles between 00 and 900. The

guidelines are also provided to extend the outcomes of these equations for

other impact angles over the common mode of collisions. These equations are

the first of their kind to predict the critical impact force and critical impulse in

terms of the geometrical and material properties of the impacted columns.


8-222

3. An innovative technique was developed and introduced to ensure the accuracy

of the equations in predicting the critical impact force and impulse where other

techniques failed due to the shape of the error distribution under a logarithmic

scale.

4. New guidelines were introduced to determine the contact area between the

colliding objects based on realistic vehicle impact tests on a circular column.

Additionally, key design parameters were defined which can be used for

damage mitigation while allowing for optimum usage in the design process.

One other main contribution of this research is a better understanding of the

dynamic response of reinforced concrete circular columns under both axial and

eccentric loading conditions.

5. A new limit state (ie. ALS) was proposed to assure safety against impact loads.

The columns susceptible to impacts should be checked for all conventional

limit states. In particular the accidental limit state was declared in-between the

serviceability and ultimate limit states depending on the expected level of

safety. Low shear demand of structural columns under serviceability

conditions strengthen this approach.

8.3 Practical significance

The following exclusive research outcomes are recommended for routine column

design.

Numerical simulations

The numerical simulations of the column tests can be simplified by isolating the

impacted column from the connecting structure. This observation excludes the need to

simulate the entire structure in the vulnerability analysis of structural columns.

Additionally, it was shown that the material model Mat_Concrete_Damage_REL3 is

more reliable for impact simulation compared to other material models available in the

LS-DYNA library where columns are subjected to a tri-axial state of stress. Moreover,

the application of stress-strain models developed under concentric loading conditions


8-223

was proved to be valid under eccentric loading conditions and thus the numerical

simulation process itself can be largely simplified. These findings make the vehicle

impact problems manageable at the industrial level where supercomputer facilities

may not be available.

Impact reconstructions

The average duration of vehicle impact can be taken as 100ms. The duration of the

impact depends on the stiffness of the vehicle which is the result of a compromise

between passenger safety and better driving performance. Therefore, significant

changes cannot be expected, and this method maybe extended to assess the

vulnerability of columns against new generation vehicles where the impact duration

varies from 50ms to 150ms. Consequently, the outcomes of the equations can be used

as the foundation to generate a database to determine vulnerability assessment against

a general vehicle population. In the process, the effects of the shape of the pulses and

strain rate effects can be neglected and hence collision severity can be predicted by

interpolating known collision pulses.

Columns under concentric loading

The impact capacity was investigated of reinforced circular columns up to 4m in

height made of 30MPa to 50MPa concrete to suit 5 to 20 storey buildings. The

longitudinal steel ratio varied from 1% to 4%. It is concluded that the vulnerability of

axially loaded columns under vehicle impacts can be reduced by reducing the column

slenderness ratio, and by choosing the design option with a low amount of steel and

low concrete grade. Thus an equivalent (in terms of capacity) column of lower grade

concrete with a low amount of steel and low slenderness ratio will offer the maximum

protection against impact loads.

Additionally, the confinement effects are particularly effective when the hoop spacing

is closer than 100mm. The confinement has to increase particularly with the diameter

of the columns and with concrete grade to achieve the same level of capacity

enhancement. Therefore, a method based on the maximum diameter of the

longitudinal steel is not effective for determining the lateral steel spacing of the


8-224

columns susceptible to vehicle impacts. Moreover, it is recommended to increase the

diameter of the hoops rather than the yield strength when the minimum allowable

spacing of the transverse reinforcement is restricted due to practical issues.

In fact, the impact capacity reduced by 10%, 20% and 30% under the 0.6Pd, 0.4Pd and

0.2Pd axial loading respectively. On the other hand, the capacity drop can be recovered

by up to 90% by providing hoops at closer spacing if the axial load can be maintained

around 0.4Pd. Therefore, a limit should be imposed on the minimum axial load that has

to be maintained during a rehabilitation process, so that risk of a progressive collapse

can be minimised. In particular, there are no substantial warnings before collapse upon

post impact loading and the condition remains unchanged even after introducing

hoops at closer intervals even though this contributes to the overall capacity

enhancement. Under these conditions, it may be more appropriate to replace the

impacted column rather than repair it for further use. The catastrophic nature of the

(shear) failure increases with the intensity of the axial loading. However, compromise

between the axial load intensity and allowable impact force may not be of value as no

residual capacity remains in the impacted columns.

Static and dynamic shear capacities under impact have been compared. The

investigations reveal that the static capacity is not an indication of the dynamic

capacity, even though the peak force (dynamic shear capacity) has some correlation

with the static shear capacity of the columns. If the correlation factor is known it can

be used for approximate vulnerability assessments for impacted columns. However,

existing guidelines highly underestimate the dynamic shear capacity of the larger

diameter columns and hence may leads to non-aesthetic and expensive design options.

Columns under eccentric loading

Consideration of three load combinations is sufficient for the vulnerability analyses of

impacted circular columns under biaxial bending which include single axis bending

about either one or the other orthogonal direction and equal moment about both axes.

In fact, the number of load combinations can be further reduced depending on the

failure mode.


8-225

Since eccentric loading in a building cause bending in the columns about its principal

(x – y) axes, the resultant moment about any other axis will not exceed the bending

capacity of the column. Due to this vehicle impacts in the vectorial direction about

which the resultant bi-axial moment is applied can be considered as non-critical. The

columns that deflect against the direction of the impact force are always safe compared

to their counterpart. In addition, reasonably conservative results can be generated

(from the equations) for load combinations that exceed 50% of the allowable moment

(0.5Mx) for the corresponding axial loading (dxP ).

The impact capacity variation is gradual for consecutive load combinations on the

interaction diagram for one particular column. This allows linear interpolation to be

used in a vulnerability assessment. The critical impact force for the intermediate load

combinations can be easily calculated once the impact capacities of the columns are

provided at three locations on the interaction diagram. This is the most significant

practical outcome of this research as no other method in the literature allows

quantifying the critical impulse under arbitrary load combinations.

8.4 Recommendations for future work

The results of this research lead to several recommendations for future work.

1. The use of external confinement for damage mitigation is recommended where

the strength of cover concrete can be fully utilised while minimising premature

spalling which diminishes the confinement effects provided by transverse

reinforcement.

2. The Damage Index D was used to identify the residual capacity of the partially

loaded columns as it has a proven history of success against blast loads.

However collapse of the impacted columns will be brittle and sudden upon

post impact loading. Therefore the damage index is not a sensitive index for

impacted columns and should not be used in future research.

3. This research is primarily based on the vulnerability assessment under low


8-226

elevation impacts. However impact behaviour of columns under train or ship

collision could be significantly different, particularly the area of the impact

while mass, duration and velocity may also affect the impact capacity and

failure mode. Additionally a series of successive impacts can be expected

under a train collision due to the significant momentum of passenger

compartments. The number of the impacts, severity of post collisions and

duration of the impacts are yet to be determined.

4. The typical dead weight of a bridge superstructure can be 5% to 10% of the

capacity of the bridge piers (Prakash et al. 2009). Therefore a separate analysis

is recommended for bridge piers where axial load is very limited and hence

confinement effects may not be fully developed. In particular, loads and

support conditions could be substantially different depending on the method of

construction (ie. balanced cantilever, suspension, incremental launching, cable

stayed). This is another area for further research.


8-227

References AASHTO, 2002, “LRFD Bridge Design Specifications” American Association of State Highway and Transportation Officials, 2nd ed., Washington, D.C., USA. Abrams, D. P. "Influence of axial force variation on flexural behaviour of reinforced concrete columns." ACI Structural Journal. 1987. J 84(3): p. 246–254. ACI 318-05, Building Code Requirements for Structural Concrete. American Concrete Institute, ACI Committee 318, 2005, 430 pp. ACI 318-02, Building code requirements for structural concrete. American Concrete Institute, 2002(Farmington Hills, USA): p. 443. ACI 318-83, Building code requirements for reinforced concrete. American Concrete Institute, 1983(Detroit): p. 111. ACI-ASCE Committee 441 (1997). "High-Strength Concrete Columns: State-of the-art." ACI Structural Journal 94(3): pp. 323-335. Adachi, T., T. Tanaka, et al. (2004). "Effect of transverse impact on buckling behaviour of a column under static axial compressive force." International Journal of Impact Engineering 30(5): 465-475. Adelman H.M. and R.T. Haftka , Sensitivity analysis for discrete structural systems. AIAA J. 24 5 (May 1986), pp. 823–831. Ahmad and S. P. Shah (1982). "Complete tri-axial stress-strain curves for concrete." Journal of Structural Division, Proceedings, American Society of Civil Engineers 108(ST4): pp. 728–742. AIJ 1994, Structural design guidelines for reinforced concrete buildings. Architectural Institute of Japan, 1994. Tokyo, Japan: p. P 207. American Association of State Highway and Transportation Officials, Load and Resistance Factor Design AASHTO-LRFD. 1998. LRFD, bridge design specifications - Second edition, AASHTO, Washington, D.C. Anselm, A. (2005). "A literature review on the shear capacity of dynamically loades concrete structures." Royal Institute of technology TRITA-BKN Report 89. Anzell, A., A literature review on the shear capacity of dynamically loads concrete structures. Royal Institute of technology, 2005. TRITA-BKN Report 89. AS 1170.1, AS/NZS 1170.1:1989 : Structural design actions - Permanent, imposed and other actions. 1989: p. 20. AS 3600, Concrete structures, 2004: p. 185.


8-228

ASCE-ACI, c., Suggested revisions to shear provisions for building codes. ACI Journal, American Concrete Institute, 1977: p. 458-469. Aschheim, M. and J.P. Moehle, shear strength and deformability of R/C bridge columns subjected to inelastic displacements. UCB/EERC 92/04, 1992. University of California, Berkeley: p. P 93. Azizinamini, A., S. Sharon, et al. (1994). "Seismic behavior of square high-strength concrete columns." Americal Concrete Institute 91(3): 336-345. Baker, W. E., P. A. Cox, et al. (1983). Explosion hazards and evaluation, Amsterdam, New York: Elsevier Scientific Pub. Co. Banthia, N.P., et al., Impact testing of concrete using a drop-weight impact machine. Experimental Mechanics, 1989. 29(2): p. 63-69. Bao, X. and B. Li. (2009). "Residual strength of blast damaged reinforced concrete columns." International Journal of Impact Engineering 137(3): 295-308. Barpi, F., Impact behaviour of concrete: a computational approach. Engineering Fracture Mechanics, 2004. 71(15): p. 2197-2213. Bayrak, O. and S.A. Sheikh, Plastic Hinge Analysis. Journal of Structural Engineering, 2001. 127(9): p. pp. 1092-1100. Belarbi, A., S. S. Prakash, et al. (2008). "Flexure-Shear-Torsion Interaction of RC Bridge Columns." Proceedings of the Concrete Bridge Conference, Paper No. 6, St Louis, MO. Bentur, A., S. Mindess, and N. Banthia, The behaviour of concrete under impact loading: Experimental procedures and method of analysis. Materials and Structures, 1986. 19(5): p. 371-378. Bischoff, P.H. and S.H. Perry, compressive behaviour of concrete at high strain rates. Material and Structures, 1991. 24: p. 425-450. Bischoff, P.H. and S.H. Perry, Impact Behaviour of Plane concrete Loaded in Uni-axial Compression. Journal of Engineering Mechanics, 1995. Vol. 121(6): p. 685-693. Brach, R. M. (1991). Mechanical impact dynamics, New York: John Wiley and Sons Inc. Breed, D. S., V. Castelli, et al. (1991). "A new automobile crash sensor tester" Society of automotive engineers: SAE Technical Paper. Warrendale, PA: Paper 910655. BS 8110, Part 1: structural use of concrete. British Standards Institution, 1985. London, U.K.: p. 118.


8-229

Burnett, David S., (1987). Finite Element Analysis. Reading: Addison-Wesley Publishing Company. Campbell, K.L., Energy Basis for collision severity. SAE Technical Paper, 1974. Paper 740565. CEB-FIP, CEB-FIP Model Code 1990. Redwood Books, Trowbridge, Wiltshire, UK, 1990. Chopra, A.K., Dynamics of Structures. Theory and Application to earthquake engineering. Vol. vol 1. 2001, Englewood Cliffs, New Jersey: Prentice-Hall. Chung, L. and S.P. Shah, Effects of loading rate on anchorage bond and beam-column joint. Structural Journal, 1989. 96(2): p. 132-142. Cowell, W.L., Dynamic properties of plain Portland cement concrete. Tech. Report No. R447, 1966. U.S. Naval Civ. Engrg. Lab., Port Hueneme, Calif. Cui, S., H. Hao, et al. (2002). "Theoretical study of dynamic elastic buckling of columns subjected to intermediate velocity impact loads." International Journal of Mechanical Sciences 44(4): 687-702. Cusson, D. and P. Paultre, High strength concrete columns confined by rectangular ties. Journal of Structural Engineering, ASCE, 1994. 120(3): p. 783-804. Design against accidental loads, Recommended practice, det norske veritas, DNV-RP-C204, veritas veien 1, no-1322 høvik, Norway (2004). El Tawil, S., E. Severino, and P. Fonseca, Vehicle Collision with Bridge Piers. Journal of Bridge Engineering, 2005. 10(3): p. 345-353. EN 1991-1-7:2006, Eurocode 1 - Actions on structures - Part 1-7: General Actions & Accidental actions. Irish standards, 2006. Esmaeily, A. and Y. Xiao (2004). "Behavior of Reinforced Concrete Columns Under Variable Axial Loads." ACI structural Journal, V101-1(2004.): 124-132. Ferguson, P. M. and J. N. Thompson (1962). "Development length of large High-strength reinforcing bars." Americal Concrete Institute 59(1): 887. Ferguson, P. M. and J. N. Thompson (1965). "Development length of large high-strength reinforcing bars." Americal Concrete Institute 62(1): 71-94. Feyerabend, M., Der harte Querstoff auf stutzen aus Stahl und Stahlbeton. University of Karlsruhe (TH), 1988. Flanagan, D. P. and T. Belytschko (1981). "A uniform strain hexahedron and quadrilateral with with orthogonal hourglass control." International Journal for


8-230

Numerical Methods in Engineering 17: 679-706. Forrestal, M. J., D. Y. Tzout, et al. (1994). "A counterintuitive region of response for elastic-plastic rings loaded with axisymmetric pressure pulses." International Journal of Impact Engineering 15(3): 219-223. Fu, H.C., M.A. Erki, and M. Seckin, Review of Effects of Loading Rate on Reinforced Concrete. Journal of Structural Engineering, 1991. 117(12): p. 3660-3679. Fu, H.C., M.A. Erki, and M. Seckin, Review of Effects of Loading Rate on Concrete in Compression. Journal of Structural Engineering, 1991. 117(12): p. 3645-3659. Galiev, U. Experimental observations and discussion of counterintuitive behavior of plates and shallow shells subjected to blast loading. Int J Impact Eng 18 7–8 (1996), pp. 783–802 GALIEV, S. U. (1997). "Influence of cavitation upon anomalous behaviour of a plate/liquid/underwater explosion system." International Journal for Impact Engineering 19(4): 345-359. Gebbeken, N., M. Teich, T. Linse, Numerical Modelling of High Speed Impact and Penetration in to Concrete structures. 7th international conference on Shock and Impact loads on Structurs, 2007. 1(2007): p. 241-250. Georgin, J.F. and J.M. Reynouard, Modelling of structures subjected to impact: concrete behaviour under high strain rate. Cement and Concrete Composites, 2003. 25(1): p. 131-143. Glanville, W. H., G. Grime, et al. (1938). "An investigation of the stresses in reinforced concrete piles during driving." Building Research Technical Paper. 20. Department of Scientific and Industrial Research (HMSO, LONDON). Gopalaratnam, V., S. Shah, et al. (1984). "A modified instrumented charpy test for cement-based composites." Experimental Mechanics 24(2): 102-111. Govindjee et al., (1995). "Anisotropic modelling and numerical simulation of brittle damage in concrete." International Journal for Numerical Methods in Engineering 38(21): 3611-3633. Hallquist, J.O., LS-DYNA 3D: Theoretical manual, Livermore. Livermore Software Technological Corporation, 2007. Hamouda, A.M.S. and M.S.J. Hashmi, Modelling the impact and penetration events of modern engineering materials: characteristics of computer codes and material models. Journal of Materials Processing Technology, International Conference on Advances in Material and Processing Technologies, 1996. 56(1-4): p. 847-862. Ho, J.C.M. and H.J. Pam, Inelastic design of low-axially loaded high-strength


8-231

reinforced concrete columns. engineering Structures, 2003: p. 1083-1096. Ho, J.C.M. and H.J. Pam, Influence of transverse steel configuration on post elastic behaviour of high strength reinforced concrete columns. Transaction of the Hong Kong Institute of Engineers, 2003. 10(2): p. 1-9. http://www.afrl.hpc.mil/software/info/patran/ http://www.dtei.sa.gov.au/roadsafety http://www.dynasupport.com/. http://www-nrd.nhtsa.dot.gov/database/nrd-11/veh_db.html http://www.statistixl.com http://www.wsdot.wa.gov/TA/T2Center/T2Pubs.HTM. http://zmi.dynamore.de/dynamore/software/ls-prepost Hughes, B.P. and H. AL-Dafiry, Impact energy absorption at contact zone and supports of reinforced plain and fibrous concrete beams. Construction and Building Materials, 1995. 9: p. 239-244. Hughes, B.P. and R. Gregory, Concrete subjected to high strain rate loading in compression. Magazine of concrete, 1972. 36: p. 380-391. Hughes, G. and D.M. Speirs, An investigation of beam impact problems. Cement and concrete association, 1982. London. Hungspreug, S., Local bond between a steel bar and concrete under high intensity cyclic load. PhD thesis, 1981. Cornell University(1981): p. 449. Hwee, Y. S. and B. V. Rangan (1990). "Studies on Commercial High-Strength Concretes." Materials Journal of the American Concrete Institute 87(5): 440-445. Janke, L., Czaderski, C., Ruth, J., & Motavalli, M. (2009). Experiments on the residual load-bearing capacity of prestressed confined concrete columns. Engineering Structures, 31(10), 2247-2256. Johansson M. Structural behaviour in concrete frame corners of civil defence shelters, non-linear finite element analyses and experiments. Doctoral Thesis, Department of Structural Engineering, Concrete Structures, Go¨ teborg, Sweden: Chalmers University of Technology, 2000. 204pp. Johansson, M. and K. Gylltoft, Mechanical Behaviour of Circular Steel-Concrete Composite Stub Columns. Journal of Structural Engineering, 2002. 128(8): p. 1073-1081.


8-232

Johnny, H.C.M., Inelastic design of reinforced concrete beams and limited ductile high-strength concrete columns. 2003, University of Hong Kong (People's Republic of China): Peoples Republic of China. Jones, N. (1997). Structural impact, Cambridge: Cambridge University Press. p 333-385. Jones, P. G. and F. E. Richart (1936 ). "The effect of testing speed on strength and elastic properties of concrete" ASTM J. 36 (Part II): 380-392. Jones, N. and T.E. Wierzbicki, Structural Crashworthiness. 1983, London UK: Butterworth & Co. Ltd. Karagiozova, D. and N. Jones (1996). "Dynamic elastic-plastic buckling phenomena in a rod due to axial impact." International Journal of Impact Engineering 18(7-8): 919-947. Kaplan, S.A., Factors affecting the relationship between rate of loading and measured compressive strength of concrete. Magazine of concrete research, 1980. 32(111): p. 79-88. Kim, J.K. and J.K. Yang, Buckling behaviour of slender high-strength concrete columns. Engineering Structures, 1995. 17(1): p. 39-51. Kishi, N., et al., Impact behavior of shear-failure-type RC beams without shear rebar. International Journal of Impact Engineering, 2002. 27(9): p. 955-968. Kolsky, H., P. Rush, et al. (1991). "Some experimental observations of anomalous response of fully clamped beams." International Journal of Impact Engineering 11(4): 445-456. Kreger, M.E. and L. Linbeck, Behaviour of reinforced concrete columns subjected to lateral and axial load reversal. In: Proc. Third U.S. National Conf. on Earthquake Engineering, 1986. Kulkarni, S.M. and S.P. Shah, Response of reinforced concrete beams at high strain rates. Structural Journal, 1998. 95(6): p. 705-715. Lee, J. H., K. Seong-Hyun, et al. (2003). "Shear Strength and Capacity Protection of RC Bridge Columns" Li, B., R. Park, and H. Tanaka, Effects on confinement on the behaviour of high strength concrete columns under seismic loading. Proc. Pacific conference on earth quake engineering, 1991. Auckland(Nov. 1991): p. 55pp. Li, B., Park, R., & Tanaka, H. (1994). Strength and Ductility of Reinforced Concrete Members and Frames Constructed Using HSC. Research Report No. 94-5, Department of Civil Engineering, University of Canterbury, Christchurch, New Zealand, 373 pp.


8-233

Li, Q. M., Y. M. Liu, et al. (2006). "The anomalous region of elastic-plastic beam dynamics." International Journal of Impact Engineering 32(9): 1357-1369. Liu, J. and S.J. Foster, A finite element model to study the behaviour of confined concrete columns. J. Struct. Engineering, ASCE 124 9 (1998), pp. 1011–1017. Lloyd, N.A. and V. Rangan, Studies on High-Strength Concrete Columns under Eccentric Compression. ACI Structural Journal, 1996. 93(6): p. 631-638. Loedolft, M.J., The behaviour of reinforced concrete cantilever columns under lateral impact loads. 1989, University of Stellenbosch. Lokuge, W. P., S. Setunge, et al. (2003). "Modelling eccentrically loaded high-strength concrete columns." Magazine of Concrete Research 55(4): 331-341. Louw, M.J., G. Maritz, and M.J. Loedolff, The Behaviour of RC Columns under Impact Loading. The Civil Engineer in South Africa, 1992: p. 371-378. Louw, M.J., G. Maritz, and M.J. Loedolff, RC Cantilever columns under lateral impact loads. Pretoria, South Africa, 1992. LS-DYNA user manual, LS-DYNA keyword user's manual. Livermore software technological corporation, 2003(California, Livermore Software Technological Corporation.). LSTC, LS-DYNA 950 Keyword Manual. Version 940, Non linear Dynamic Analysis of Structures in Three Dimensions. Livermore Software Technology Corporation, Livermore, June 1997. 2006. Luts, L.A. and G. Peter, Mechanics of bond and slip of deformed bar in concrete. ACI Materials Journal, 1967. 64(11): p. 711-721. Magnusson, J., Structural concrete elements subjected to air blast loadings. TRITA-BKN, Bulletin 92, 2007: p. 82. Mainstone, R.J., Properties of materials at high rates of straining or loading. Material and Structures, 1975. 8(44): p. 102-116. Majewski, T., J. Bobinski, et al. (2008). "FE analysis of failure behaviour of reinforced concrete columns under eccentric compression." Engineering Structures 30(2): 300-317. Malvar, L.J., Bond of reinforcement under controlled confinement. ACI Materials Journal, 1992. 89(6): p. 593-601. Malvar, L.J., Review of static and dynamic properties of steel reinforcing bars. ACI Structural Journal, 1998. V95(5): p. 609-616.


8-234

Malvar, L.J. and J.E. Crawford, Dynamic increasing factor for concrete. 28th Department of defence explosive safety seminar (DD ESB), 1998. Orlando FL, USA. Malvar, L. J., J. E. Crawford, et al. (1997). "A plasticity concrete material model for DYNA3D." International Journal of Impact Engineering 19(9-10): 847-873. Malvar, L.J. and J.E. Crawford, Dynamic increasing factor for steel reinforcing bars. 28th Department of defence explosive safety seminar, 1998. Orlando FL(USA). Mander, J.B., M.J.N. Priestley, and R. Park, seismic design of bridge piers. Research report No. 84-2, University of Canterbury, 1984: p. 442. Mander, J.B., M.J.N. Priestley, and R. Park, Theoretical stress-strain model for confined concrete. Journal of Structural Engineering, ASCE, 1988. 114(8). Marsh, H. and J.D. Campbell, The Effect of Strain Rate on Post- Yield Flow of Mild Steel. Journal of Mech. and Phys. of Solids, 1963. 11(5): p. 255-263. Martinez, S., A. H. Nilson, et al. (1984). "Spirally reinforced high strength concrete columns" ACI J. Proc. 81(5): pp. 431–442. Mathisen, K.M., et al., Error estimation and adaptivity in explicit nonlinear finite simulation of quasi-static problems. Computers and structures, 1999. 72(4-5): p. 627-644. McCormick J, Nagae T, et al. (2009). Investigation of the sliding behaviour between steel and mortar for seismic applications in structures. In. Earthquake engineering and structural dynamics, John Wiley & Sons, Ltd. McDermott, J.F., Effects of steel strength and reinforcement ratio on the mode of failure and strain energy capacity of reinforcement concrete beams. Journal of American Concrete Institute, 1969. Report by ACI Committee 439: p. 165-172. Memari, A.M., et al., Ductility evaluation for typical existing R /C bridge columns in the eastern USA. Engineering Structures, 2005. 27(2): p. 203-212. Mendis, P., Plastic Hinge Lengths of Normal and High-strength Concrete in Flexure. Advances in Structural Engineering Journal, 2001. 4(4): p. pp. 189-196. Menzal, C.A., Some factors influencing results of pull-out bond tests. Journal of American Concrete Institute, 1939. 35: p. 516-543. Mindess, S., N. Banthia, et al. (1986). "The response of reinforced concrete beams with a fibre concrete matrix to impact loading." International Journal of Cement Composites and Lightweight Concrete 8(3): 165-170. Moan, T. (2009). "Development of accidental collapse limit state criteria for offshore structures." Structural Safety 31(2): 124-135.


8-235

Nemecek, J. et al. 2005. Effect of transversal reinforcement in normal and high strength concrete columns. Materials and Structures 38: 665–671. Neptune, J.A., A comparison of crush stiffness characteristics from partial-overlap and full-overlap frontal crash tests. SAE Technical Paper, 1999. Niclasson, G., Report D 20241-2.6. Swedish defence research establishment (FOA), 1994. Sunbyberg. Norris, G.H., et al., Structural design for dynamic loads. 1959: McGraw-Hill, New York, USA. NZS 3101, The design of concrete structures. Concrete Structures Standards, Part 1: Code of Practice, supplemented by Part 2: Commentary, Standard Association of New Zealand, Wellington, New Zealand, 1995. Otsuka, H., E. Takeshita, et al. (2004). "Study on the seismic performance of reinforced concrete columns subjected to torsional moment, bending moment and axial force." 13thWorld Conference on Earthquake Engineering Vancouver, Canada Paper No. 393. Palm, J. (1989). "On concrete structures subjected to dynamic loading." Fortifikations-forvaltningen Report A4 89(Eskilstuna). Papazoglou, A.J., Elnashai, A.S., 1996. Analytical and field evidence of the damaging effect of vertical earthquake ground motion. Earthquake Engineering and Structural Dynamics 25, 1109–1137. Park, R., M.J.N. Priestley, and W.D. Gill, Ductility of Square-Confined Concrete Columns. Journal of the Structural Division, 1982. Vol. 108(No. 4): p. 929-950. Patrick, P., Rami, E., Hugo, I. R., & Najib, B. (2009). Seismic Performance of Circular High-Strength Concrete Columns. Structural Journal, 106(4), 395-404. Paultre, P., F. Legeron, and D. Mongeau, Influence of Concrete Strength and Transverse Reinforcement Yield Strength on Behaviour of High-Strength Concrete Columns. ACI Structural Journal, 2001. Vol. 98(4). Polat, M. B. (1992). "Behaviour of Normal and High Strength Concrete Under Axial Compression." Department of Civil Engineering, University of Toronto Master's thesis: pp. 175. Popovics, S. (1973). "A numerical approach to the complete stress strain curve for concrete." Cement and concrete research 3(5): 583-599. Popp C. Der Querstob beim Aufprall (in German), Forschungshefte aus dem Gebiete des Stahlbaues, Deutschen Stahlbau Verband, Köln am Rhein, 1961


8-236

Prakash, S., A. Belarbi, et al. (2010). "Seismic performance of circular RC columns subjected to axial force, bending, and torsion with low and moderate shear." Engineering Structures 32(1): 46-59. Prasad, A.K., Energy dissipated in vehicle crush-a study using the repeated test technique, SAE Paper. Society of Automotive Engineers, 1990. Warrendale PA. Razvi, S.R., Saatcioglu, M. "Circular High-Strength Concrete Columns under Concentric Compression," ACI Structural Journal, Vol. 96, No.5, 1999, pp. 817-825. Razvi, S. and M. Saatcioglu, Confinement model for high strength concrete. Journal of Structural Engineering, ASCE, 1999. 125(3): p. 281-289. Reid, S.R. and T.Y. Reddy, Axial crushing of foam filled tapered sheet metal tubes. Internatinal jurnal of Mechanical Sciences, 1986. 28(10): p. 643-656. Reinhardt, H., P. Rossi, and J. van Mier, Joint investigation of concrete at high rates of loading. Materials and Structures, 1990. 23(3): p. 213-216. Reinschmidt, K. F., R. J. Hansen, et al. (1964). "Dynamic test of reinforced Concrete Columns." Journal of the American Concrete Institute 23(14): 317-330. Remennikov, A.M. and S. Kaewunruen, Impact resistance of reinforced concrete columns: Experimental studies and design considerations. Progress in Mechanics of structures and Materials, 2006: p. 817-823. Richart, F. E. (1946). "The Structural Effectiveness of Protective Shells on Reinforced Concrete Columns." Americal Concrete Institute 18(4): 353-363. Robins, P.J. and I.J. Standish, Effects of lateral pressure on bond of reinforcing bars in concrete. Proc. of the Int. conference, Paisley Collage of Technology, 1982. Applied science Publishers: p. 262-272. Rust, W. and K. Schweizerhof, Finite Element Limit Load Analysis of Thin-Walled Structures by ANSYS (Implicit), LS-DYNA (Explicit) and in Combination. Proceedings 3rd Conf. on ”Thin Walled Structures“, Elsevier Science Publ., Krakow, Poland., 2003. Saadeghvaziri, M.A., Nonlinear response and modelling of RC columns subjected to varying axial load. Engineering Structures, 1997. 19(6): p. 417-424.

Saatcioglu, M., Salamat, A. H. and Razvi, S. R. "Confined Columns under Eccentric Loading," ASCE Journal of Structural Engineering, Vol.121, No.11, November 1995, 1547-1556.

Saatcioglu, M., A. H. Salamat, et al. (1995). "Confined columns under eccentric loading, ASCE." Journal of Structural Engineering 121(11): 1547-1556.


8-237

Saatcioglu, M. and S.R. Razvi, Strength and Ductility of Confined Concrete. Journal of Structural Engineering, 1992. 118(6): p. 1590-1607. Saatcioglu, M. and S.R. Razvi, High-Strength Concrete Columns with Square Sections under Concentric Compression. Journal of Structural Engineering, 1998. 124(12): p. 1438-1447. Sargin, M., S. K. Ghosh, et al. (1971). "Effects of lateral reinfocement upon the strength and deformation properties of concrete." Mag. Concrete Res. 1971 28(75-76): 99-110. Sastranegara, A., Adachi, T., and Yamaji, A., 2005, “Improvement of Energy absorption of Impacted Column Due to Transverse Impact,” Int. J. Impact Eng., 31, pp. 483–496. Schwer, E. L. and L. J. Malvar (2005). "Simplified concrete modelling with Mat_Concrete_Damage_REL3." JRI LS_DYNA user week. Schwer, L. E., W. K. Samuel, et al. (2005). "An Assessment of the LSDYNA Hourglass Formulations via the 3D Patch Test. 5th European LS-DYNA Users Conference." Birmingham, United Kingdom, May 25-26, 2005. Scott, B.D., R. Park and M.J.N. Priestley, Stress–strain behavior of concrete confined by overlapping hoops at low and high strain rates, ACI Journal 79 (1982) (1), pp. 13–27. Sheikh, S.A. and S.S. Khoury, A performance-based approach for the design of confining steel in tied columns. ACI Structural Journal, 1997. 95(3): p. 305-317. Sheikh, S. A. and S. M. Uzumeri (1982). "Analytical Model for Concrete Confinement in Tied Column." Journal of Structural Engineering, ASCE 108(ST12): 2703-2722. Sheikh, S.A., and Uzumeri, S.M., “Strength and ductility of tied concrete columns”, J. of Struct. Engrg., ASCE, Vol. 106, No. 5, 1980, pp. 1079-1102. Sheikh, S. A. and C. C. Yeh (1986). "Flexural behaviour of confined concrete." Journal of the American concrete institute 83(3): 389-404. Sheikh S. A. and Yeh C. C. Analytical moment–curvature relations for tied concrete columns. Journal of Structural Engineering, 1992, 118, No. 2, 529–544. Shi, Y., H. Hao, and Z.X. Li, Numerical derivation of pressure-impulse diagrams for prediction of RC column damage to blast loads, International Journal of Impact Engineering, 2008. 35(11): p. 1213-1227. Stouffer, D. C., L. T. Dame, et al. (1996). Inelastic deformation of metals: models, mechanical properties, and metallurgy. New York: John Wiley and Sons. P 520.


8-238

Strong, W. J. and T. X. Yu (1993). Dynamic models for structural plasticity, London, UK. Springer: p 112-155. Suaris, W. and S.P. Shah, Strain-rate effects in fibre-reinforced concrete subjected to impact and impulsive loading. Composites, 1982. 13(2): p. 153-159. Suaris, W. and S.P. Shah, Properties of concrete subjected to impact. Journal of Structural Engineering, 1983. 109(ST7): p. 1727-1741. Sukontasukkul, P. and S. Mindess, The Shear fracture of Concrete under Impact Loading using End Confined Beams. Material and Structures, 2003. 36: p. 372-378. Summers, S., A. Prasad, and W. Hollowell, NHTSA's Compatibility research program update. SAE Technical Paper, 2001. World congress, Detroit, Michigan. Takeda, J. and H. Tachikawa, Deformation and fracture of concrete subjected to dynamic loading. Proc. of the Int. conference on mechanical behaviour of materials, 1971. 4: p. 267-277. Takeda, J., H. Tachikawa, and K. Fujimoto, Effects on straining rate on deformation and fracture of reinforce concrete members. Proc. seventh world conference on earth quake engineering, 1977. 1(New Delhi, India). Thompson, J.N. and P.M. Ferguson, Development length of high strength reinforcing bars in bond. Journal of American Concrete Institute, 1962. 59(1962): p. 887. Tirasita, P. and K. Kawashimaa (2008). "Effect of Nonlinear Seismic Torsion on the Performance of Skewed Bridge Piers " Journal of Earthquake Engineering 12, (6 ): p 980 - 998 Tokluku, M. T. and S. A. Sheikh (1992). "Behaviour of reinforced concrete columns confined with circular spirals with hoops." Research report, department of civil engineering, university of Toronto, Canada: 330. Tomosawa, F., M. M., et al. (1990). "High-Strength Concrete for High-Rise Buildings in Japan Utilization of High Strength Concrete." Second International Symposium, American Concrete Institute Detroit SP- 121: 33-45. Tsang, H.H., et al., Collapse of reinforced concrete column by vehicle impact. 6th Asia-Pacific conference on Shock & Impact Loads on Structures, 2005. Perth, Australia. Unosson, M. and L. Nilsson, Projectile penetration and perforation of high performance concrete: experimental results and macroscopic modelling. International Journal of Impact Engineering, 2001. 32(7): p. 1068-1085. Varat, M. S. and S. E. Husher (2000). "Vehicle impact response analysis through the use of accelerometer data." SAE Technical Paper Series SAE World Congress (Detroit.


8-239

Michigan). Varat, M.S., S.E. Husher, and J.F. Kerkhoff, An analysis of trends of vehicle frontal impact stiffness. SAE Technical Paper, 1994. Paper 940914. Vrouwenvelder, T. (2000). "Stochastic modelling of extreme action events in structural engineering." Probabilistic Engineering Mechanics 15(1): 109-117. Wagstrom, L., R. Thomson, and B. Pipkorn, structural adoptively for acceleration level reduction in passenger car frontal collisions. International Journal of Crashworthiness, 2004: p. 121-127. Wang, N., S. Mindess, and K. Ko, Fiber reinforced concrete beams under impact loading. Cement and concrete research, 1996. 26(3): p. 363-376. Wong, Y. L., T. Paulay, et al. (1993). "Response of circular reinforced concrete columns to multi-directional seismic attack." ACI Journal, American Concrete Institute 90(2): p. 180-91 Watson, S., F.A. Zahn, and R. Park, Confining Reinforcement for Concrete Columns. Journal of Structural Engineering, 1994. 120(6): p. 1798-1824. Watstein, D., Effects of straining rate on the compressive strength and elastic properties of concrete. ACI Structural Journal, 1953. 49: p. 729-744. Weathersby, J.H., Investigation of bond slip between concrete and steel reinforcement under dynamic loading conditions. 2003, Louisiana State University and Agricultural & Mechanical College: United States -- Louisiana. Weerheijm, J., Concrete under impact tensile loading and lateral expansion. TNO-Prins Maurits Laboratory, Netherlands, 1992. Weerheijm, J. and J.C.A.M. Van Doormaal, Tensile failure of concrete at high loading rates: New test data on strength and fracture energy from instrumented spalling tests. International Journal of Impact Engineering, 2007. 34(3): p. 609-626. Xiao, J. and C. Zhang (2007). "Seismic behaviour of RC columns with circular, square and diamond sections." 22(5): 801-810. Xie, J., J. G. MacGregor, et al. (1996). "Numerical Investigation of Eccentrically Loaded High-Strength Concrete Tied Columns." Structural Journal 93(4): 449-461. Xu, S., Y. Zhao, and Z. Wu, Study on the Average Fracture Energy for Crack Propagation in Concrete. Journal of Materials in Civil Engineering, 2006. 18(6): p. 817-824. Yan, C., Bond between reinforcing bars and concrete under impact loading. PhD thesis, 1992. University of British Colombia: p. P 369.


8-240

Yong, Y. K., M. G. Nour, et al. (1988). " Behaviour of laterally confined high-strength concrete under axial loads." J. Struct. Eng. 114: 333–351. Yonten, K., Manzari, et al. (2005). "An assessment of constitutive models of concrete in the crashworthiness simulation of roadside safety structures." Journal of International Journal of Crashworthiness 10(1): 5-19. Zahn, F. A., R. Park, et al. (1989). "Strength and ductility of square reinforced concrete column sections subjected to biaxial bending." ACI Journal, American Concrete Institute 86(2): p 123-31 Zeinoddini, M., J.E. Harding, and G.A.R. Parke, Axially pre-loaded steel tubes subjected to lateral impacts (a numerical simulation). International Journal of Impact Engineering, 2008. 35(11): p. 1267-1279. Zhenguo, T. and L. Yong. (2009). "Evaluation of typical concrete material models used in hydrocodes for high dynamic response simulations." International Journal of Impact Engineering 36(1): 132-146.

Documents

Vulnerability Assessment of Reinforced Concrete Columns ...eprints.qut.edu.au/43693/1/Herath_Thilakarathna_Thesis.pdf · Vulnerability assessment of reinforced concrete columns subjected