257

Local plastic surfaces for cracking and crushing

in concrete

A D Jefferson
Cardiff School of Engineering, Cardiff University, Newport Road, Cardiff CF24 0YF, UK

Abstract: The experimental failure envelope for concrete has the form of a curved Coulomb failure
surface, which can reasonably be assumed to be related to a principal plane of shearing. This assumption is
used in the development of a multiple surface plasticity model for concrete, in which the same local yield
function is used to simulate directional fracture and isotropic compressive behaviour. A function is derived
from experimental data that provides the transition from directional cracking behaviour to isotropic
compressive behaviour. The local responses from all active plastic surfaces are coupled in a multisurface
plasticity formulation, which provides the interaction between compressive and tensile behaviour in a natural
way. The model is validated against experimental data from uniaxial and multiaxial compressive tests as well
as tensile fracture tests. The model has been implemented in the
nite element (FE) program LUSAS, and a
summary of the implementation algorithm is provided. A number of stress path examples are considered, and
the results from two FE analyses are presented.

Keywords: concrete, plasticity, cracking, fracture, multisurface plasticity

NOTATION N stress transformation matrix

rf cohesion-friction function in the local form
AD directional hardening parameter of the yield function
AI isotropic hardening parameter ru rf at the peak position
b normal component operator vector ˆ ‰1 0 0ŠT rc equivalent cohesion-friction constant in the
D elastic constitutive matrix local plastic potential function
DT tangent constitutive matrix r0 rf at the
rst yield position
ei local plastic strain vector for plane of r unit vector normal to the POD
degradation (POD) i ˆ ‰er es et ŠT R ˆ 12…s1 ¡ s3 †
E Young’s modulus si local stress vector ˆ ‰sr ss stŠT
fc uniaxial compressive strength Tp transformation matrix from principal to
fD directional tensile strength Cartesian axes
fI isotropic tensile strength
fs current tensile strength ac angle between the critical plane and the major
ft uniaxial tensile strength principal axis
fy reinforcement yield stress b directional–isotropic transition function
F local yield function ec uniaxial strain at the peak uniaxial compression
Fhb Hoek–Brown strength function e0 strain at the end of the uniaxial tension
G local plastic potential function softening curve
Gf fracture energy per unit area e Cartesian strain vector ˆ ‰exx eyy ezz gxy gyz gxzŠT
hD directional hardening rate Z relative plastic parameter
hI isotropic hardening rate kD directional plastic parameter
hr cohesion-friction hardening rate kI isotropic plastic parameter
H reinforcement hardening constant l plastic multiplier
n number of PODs L vector of plastic parameters
n Poisson’s ratio
The MS was received on 7 January 2002 and was accepted after revision X matrix of transformed ow vectors
for publication on 2 April 2002. rf constant in the friction hardening function ˆ 5
L00202 # IMechE 2002 Downloaded from pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Proc Instn Mech Engrs Vol 216 Part L: J Materials: Design and Applications
258 A D JEFFERSON

sj principal stress components … j ˆ 1±3† characteristics. Nevertheless, considerable progress has

sm 1…s ‡ s †
2 1 3 been made in the development of this type of material
sn normal stress on a principal plane or a POD model and in the application of such models to non-linear
s Cartesian stress
nite element (FE) codes [2]. The power of such codes does,
vector ˆ ‰sxx syy szz txy tyz txz ŠT at least in theory, provide the potential for the accurate
t shear stress on a principal plane or a POD simulation of complex concrete structures, and the increas-
Y directional softening function ing availability of these computer programs provides the
f Coulomb friction angle motivation for the development of more accurate models.
w plastic ductility function Many existing constitutive models simulate compressive
cm initial dilatancy parameter and tensile behaviour with different theoretical components,
c0
nal dilatancy parameter and often have rather poor coupling between these two
W coupled plastic parameter matrix components. In the late 1970s and 1980s, an approach
G operator in coupled plastic parameter equation became widespread that involved the use of plasticity
system. theory for the simulation of compressive crushing and a
total (or incremental) strain theory for fracture behaviour
Subscripts [3, 4]. However, a particular problem with this type of model
was that the coupling between the fracture and plasticity
c indicates quantities relating to the uniaxial components was sometimes rather ill de
ned, and also the
compressive peak stress, unless noted otherwise theories were not, in general, linked in a rigorous model
D indicates a directional POD quantity framework. Some more recent models do have properly
e denotes an elastic quantity, unless noted otherwise integrated theories, such as certain plastic damage models
i indicates a POD number, unless noted otherwise [5, 6] and the current version of the microplane model [7],
I indicates an isotropic quantity but these models tend to deal more with distributed fracture
m indicates a parameter value at a limiting strain than with discrete directional cracking. Furthermore, many
p indicates a plastic quantity, unless noted otherwise of the more recent models do use different functions to
r indicates a normal direction to the POD, when simulate tensile and compressive behaviour, but, since
applied to a vector component experimental evidence suggests that there is not a clear
s indicates the
rst POD shear direction, when divide between tensile and compressive behaviour [1], it is
applied to a vector component considered desirable to use a single theory to describe both.
t indicates the second POD shear direction, when To realize this aim, a model based on a single form of local
applied to a vector component yield function is explored.
The author believes that, for accurate simulation of all the
1 INTRODUCTION important characteristics of the mechanical behaviour of
concrete, a combination of plasticity and damage theories is
Concrete is a complex composite material that contains required. However, there is value in exploring a model based
hardened cement and
ne and coarse aggregate components, on plasticity theory that uses a single surface for both
as well as
ne pores. The size and strength of the individual crushing and fracture behaviour.
particles are quite different from each other, and the strength The present model employs local yield functions in a
of the bond between the cement and aggregate components multisurface plasticity format that are applied to planes of
is different from that of the parent materials. The behaviour degradation (PODs). In tension zones these planes are made
of this material is also complex and involves mechanisms permanent and simulate directional loss of strength, whereas
such as micro- and macrocracking, particle crushing and in compression zones the planes are not permanent but
pore closure, as well as friction and contact between cracked rather may rotate and result in an isotropic loss of strength.
surfaces [1]. The characteristic features of the multiaxial The local yield function was developed from the well-known
behaviour of concrete include directional cracking in Hoek–Brown function [8], which is known to provide a
tension, increased deviatoric strength with compressive reasonable approximation to the triaxial strength envelope of
con
nement, increased ductility with compressive con
ne- concrete [9]. The present model is unique, but the basis for
ment, interaction between tensile and compressive strengths, the model is a combination of the multicrack model of Carol
stiffness degradation of differing degrees with both and Bazant [10], the plasticity model of Este and Willam [9]
compression and tension loading, hysteresis behaviour and the model of Kroplin and Weihe [11].
under cyclic loading, multiple non-orthogonal crack forma-
tion, crack closure and reopening with markedly different
2 MODEL DESCRIPTION
stiffnesses, dilatant post-peak behaviour under shear load-
ing, shrinkage and creep deformation under sustained load-
2.1 Failure envelope and yield functions
ing [1]. This degree of complexity is undoubtedly one of the
primary reasons that, even after many years of research, no The Mohr–Coulomb (M–C) strength criterion is based on
one constitutive model is able to simulate all of these the assumption that a limiting condition is reached when the
Proc Instn Mech Engrs Vol 216 Part L: J Materials:Downloaded
Design andfrom pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Applications L00202 # IMechE 2002
 LOCAL PLASTIC SURFACES FOR CRACKING AND CRUSHING IN CONCRETE 259

shear stress on a critical plane reaches a limiting value, where

which depends on the normal stress on the plane. The
standard form of the M–C criterion assumes a linear fc p




























relationship between shear and normal stresses. The local tc ˆ 1 ¡ cos2 …2ac †
2
stresses on the critical plane may be evaluated from the
global stresses using a standard tensor transformation,
fc
which may be accomplished using Mohr’s circles. The sn c ˆ ‰1 ‡ cos…2ac †Š
direction of the critical plane relative to the major principal 2
stress axis, ac , is computed from the friction angle, f, as
follows: 1
cos…2ac † ˆ
ft
1‡4
1 p fc
ac ˆ ¡f …1†
2 2
The surface illustrated in Fig. 1 has the important character-
The M–C criterion is also based on the assumption that all istic of being perpendicular to the normal stress axis at the
failure planes are parallel to the intermediate principal axis. tension limit, which is consistent with the critical plane
The M–C failure criterion may be expressed as a triaxial coinciding with the major principal stress plane in tension.
failure surface in terms of global stresses or as a local failure It is assumed that permanent PODs form when the major
surface in terms of local stresses. This suggests that the principal stress exceeds the tensile strength limit. Once a
criterion may be used as a basis for dealing with isotropic POD has formed, the current tensile strength, fs , replaces ft
strength changes which affect the entire strength envelope as in equation (3), and fs decreases as the normal strain
well as with directional strength changes which affect the increases.
local directional strength only. To allow for the type of frictional hardening behaviour
The original M–C failure criterion is not accurate for exhibited by concrete in compression, the constant ru in
concrete, but a different M–C-type criterion, originally equation (3) is replaced by the parameter rf , which varies
developed for rocks by Hoek and Brown [8], has been according to an isotropic plastic strain parameter de
ned
shown to be suitable for concrete [9]. The Hoek–Brown later in this paper.
(H–B) failure surface, in terms of principal stresses, may be The normal to the POD, de
ned by the unit vector r, is at
expressed as follows: angle 2a to the direction of the major principal stress and is
in the principal plane I–III. The angle is computed as
follows (Fig. 1):
2
s1 ¡ s 3 s1
Fhb ˆ ‡ ¡1 …2† s





















fc ft 2
¡2rf3 2rf3
sin…2a† ˆ ‡ ‡1 …5†
9R 9R
where si are the principal stresses, fc is the uniaxial
compressive strength and ft is the uniaxial tensile strength.
This form of the equation is derived by assuming that ft2 /fc2 where R ˆ 12…s1 ¡ s3 †, and the normal to the POD
is negligible, which, with a value in the region of 0.01 for
standard structural concretes, is reasonable. r ˆ Tp ‰cos…a† 0 sin…a†ŠT …6†
An approximate local form of the H–B function, similar
to that suggested by the original authors, is given by
where Tp is the transformation matrix from principal to
Cartesian axes; Tp is a 3 £ 3 matrix with the three rows
3/2
t formed from the eigenvectors of the principal axes.
‡sn ¡ ft ˆ 0 …3†
ru

where sn and t are the normal and shear stresses on the local
plane respectively.
The cohesion-friction factor, ru , is calculated such that the
surface exactly matches the Hoek–Brown function at the
peak uniaxial compressive strength point. This is given by

tc
ru ˆ …4†
… ft ¡ snc †2/3 Fig. 1 Local failure surface
L00202 # IMechE 2002 Downloaded from pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Proc Instn Mech Engrs Vol 216 Part L: J Materials: Design and Applications
260 A D JEFFERSON

2.2 Local and global stress and strain systems The plastic potential, which takes a similar form to the
yield function except that the dilatancy parameter, rc , is
The local and global coordinate systems of a POD are used in place of rf , is as follows:
shown in Fig. 2, and the relationship between the local
and global stresses is as follows: Áp













!3/2
s2s ‡ s2t
G…s; rc ; fs † ˆ sr ‡ ¡ fs ˆ 0 …11†
si ˆ N i s …7† rc

where i represents the POD number and and the ow rule is given by

qG _
e_ i ˆ l …12†
s ˆ ‰sr ss st ŠT and s ˆ ‰sxx syy szz txy tyz txzŠT qsi i

where li is the plastic multiplier for POD i.

These local stresses are related to the normal and principal
The strain is decomposed into elastic and plastic compo-
shearing stresses on the POD as follows:
nents as follows:
q














s n ˆ sr and tˆ s2s ‡ s2t …8† e_ ˆ e_ e ‡ e_ p …13†

Increments of local plastic strains contribute to the global X

n

inelastic strain increment as follows: e_ p ˆ NTi e_ i …14†

iˆ1

X
n
where the overdot denotes the time derivative.
Dep ˆ NTi Dei …9†
iˆ1
The elastic stress–strain relationship is given by
Á ! Á !
X
n X
n
qG _
where s_ ˆ D e ¡ NTi ei ˆ D e¡ NTi l …15†
iˆ1 iˆ1
qsi i

e ˆ ‰er es et ŠT and e ˆ ‰exx eyy ezz gxy gyz gxz ŠT where D is the elastic constitutive matrix.
The isotropic plastic parameter is de
ned by
and n denotes the number of PODs. …
kIi ˆ k_ Ii …16†
t
2.3 Model components
and
The local yield function is given by
Xn
1 ¡ b…s† p







Áp













!3/2 k_ Ii ˆ e_ T e_ …17†
w…s†
s2s ‡ s2t iˆ1
F…s; rf ; fs † ˆ sr ‡ ¡ fs ˆ 0 …10†
rf
and the directional plastic parameter by
…
where ru 5 rf 5 r0 . k Di ˆ k_ Di …18†
t

and

X
n
k_ Di ˆ b…s†_er …19†
iˆ1

where b is a directional–isotropic transition function, which

takes a value of 1 in pure tension regions and zero for
regions of signi
cant compression and has a transition
region between the two extremes when both isotropic and
Fig. 2 POD coordinate systems directional plastic straining occur, and w is a ductility
Proc Instn Mech Engrs Vol 216 Part L: J Materials:Downloaded
Design andfrom pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Applications L00202 # IMechE 2002
 LOCAL PLASTIC SURFACES FOR CRACKING AND CRUSHING IN CONCRETE 261

function, which accounts for the fact that ductility increases yield surfaces. For local yield surface i, the consistency
with the con
ning stress. These functions are as follows: condition may be written as
8 9
< s p























T
> > qF qF
2 = s_i ‡ h ‡ Y …kD † k_ I ‡ hDi k_ Di ˆ 0 …27†
r / sr ‡ ss ‡ st
2 2
bˆ1¡ q















…20† qsi qr r
>…0:9s †/ s2 ‡ t2 >
: ;
nc nc c
where hr ˆ qr/qkI , hI ˆ qfI /qkI and hDi ˆ qfDi /qkDi .
Using equations (12), (17) and (l9), equation (27) may be
rearranged as follows:
¡sm
wˆ4 …21†
fc qF
T X
n
s_i ¡ AIj l_ j ¡ ADi l_ i ˆ 0 …28†
qsi jˆ1
where sm ˆ 1…s
2 1
¡ s3 †.
In order to simulate prepeak hardening and postpeak
where
softening, the function for rf is divided into two phases as
follows: s















qF 1¡b qG T qG
AIi ˆ ¡ h ‡ YhI
rf ˆ r0 ‡ …ru ¡ r0 † qr r w qsi qsi
Á !
rf …kI /kp † qG
£ if 1 4 kI 4 kp …22a† ADi ˆ ¡bi bT h
…rf ¡ 1†…kI /kp † ‡ 1 qs Di
0 1 i
1
B C
b ˆ @0A
rf ˆ r0 ‡ …ru ¡ r0 †…1 ¡ ZI † if kI > kp …22b†
0
where rf ˆ 5, kp ˆ 2:3…ec ¡ … fc /E††, ZI ˆ …kI ¡ kp †/…km ¡ kp †,
Using the transformation (7) and the elastic stress–strain
ec is the axial strain at peak uniaxial compression (as a
relationship (15) in equation (28) gives
positive number), E is the initial Young’s modulus, r0 is

xed at 0:5ru and km ˆ 5ec . " Á ! #
qF
T X
n
qG _
The tensile strength fs takes the initial value of ft and then Ni D e_ ¡ NTj l
either the value fI for isotropic surfaces or fD for directional qsi jˆ1
qsj j
surfaces, with fI and fD de
ned as follows: X
n
¡ AIj l_ j ¡ ADi l_ i ˆ 0 …29†
rf …1 ¡ ZI † jˆ1
fI ˆ ft ; 0 4 ZI 4 1 …23†
…rf ¡ 1†…1 ¡ ZI † ‡ 1
This may be arranged into the form of a matrix equation
fD ˆ fI Y …kD † …24† with l_ i as the unknowns. These coupled equations are
shown in (30) for the case where the
rst two PODs are
Y …kD † ˆ erD ZD …25† active:
2 T T 3
qF qG qF qG
where rD ˆ 5, ZD ˆ kD /kDm and kDm ˆ e0, the end of the 6 N1 DNT1 N1 DNT2 7
uniaxial tensile softening curve. 6 qs qs qs qs 7
6 1 1 1 2
7
The dilatancy function depends upon the isotropic plastic 6 ‡AI1 ‡ AD1 ‡AI2 7
6 7
strain parameter as follows: 6 T T 7
6 qF qG qF qG 7
6 N2 DNT1 N2 DNT2 7
4 qs 2 qs 1 qs 2 qs 2
5
kI ‡AI ‡AI2 ‡ AD2
rc ˆ c0 ‡ …cm ¡ c0 † r …26†
2km u 2 1 3
" # qF
_l 6 qs NT1 D_e 7
where c0 ˆ 0:25 and cm ˆ 0:001, except in the tensile 6 7
£ 1
ˆ6 1
7 …30†
_l 4 qF 5
region when rc ˆ ru . 2 NT2 D_e
qs 2

2.4 Elastoplastic constitutive tensor This is written in compact matrix form as follows
The relationship between incremental stresses and strains is
derived using the consistency condition, applied to all active W L_ ˆ G e_ …31†

L00202 # IMechE 2002 Downloaded from pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Proc Instn Mech Engrs Vol 216 Part L: J Materials: Design and Applications
262 A D JEFFERSON

Hence, the vector of plastic parameters can be expressed as Compute the principal stresses, spr,
from s0 ,
L_ ˆ W ¡1 G e_ …32† Compute the preferred POD plane and
check if new POD(s) should form
Substituting equation (32) into relationship (15) gives the If the principal stress order has
following relationship between the stress and strain rates: changed,corner violation has occurred,
introduce a new POD adjacent
s_ ˆ ‰D ¡ XW ¡1 GŠ_e ˆ DT e_ …33† to the existing POD at a corner
If the number of active PODs has changed,
where reset plastic variables and carry out
steps B to E with a new set of active PODs
qG qG Perform a further check on corner
X ˆ NT1 NT2 …34† violation and introduce a new surface
qs 1 qs 2
if required
Record orientations of directional and
isotropic PODs along with kI ; kDi
3 STRESS RECOVERY ALGORITHM AND
The model has been implemented in a constitutive driver
IMPLEMENTATION
program and in the FE program LUSAS. The constitutive
driver program is arranged such that the same material
A tangent cutting scheme [12] is employed for the stress routines are employed in both the driver and the FE
recovery algorithm. It is recognized that implicit algorithms programs.
[13] can give better overall convergence, but the present
algorithm is attractive for its simplicity and generally good
performance.
The algorithm checks for the creation of new permanent 4 STRESS PATH TESTS
and non-permanent PODs and updates the plastic variables
as well as the stresses. The algorithm is described by the In this section the model is considered for experimental data
following steps: from uniaxial tension, uniaxial compression and multiaxial
s0 ˆ sr¡1 ‡ DDe compression tests. The material properties used for the
For i ˆ 1; n; s 0i ˆ N i s0 ; End loop (i.e. for analyses are given in Table 1.
all existing PODs)
If F…s0i ; kI ; kDi † > 0; di ˆ 1; Else di ˆ 0
(Identify active surfaces) 4.1 Stress path tests
Enter an iteration loop for the active
existing surfaces Example 1 provides a comparison with a uniaxial tension
curve from Hordijk [14], which was derived from a wide
L ˆ W ¡1 F…s0 † range of experimental data from direct and indirect tension
tests. Example 2 uses data from a uniaxial compressive test
undertaken by Kupfer, Hilsdorf and Rusch (KHR) [15]. The
X
n
qG
sn ˆ s0 ¡ D NTi l_ i di results from examples 1 and 2 are presented in Fig. 3, which
iˆ1
qs0 i shows that for both cases the model results match the
…Only active surfaces included † experimental curves closely.
In Example 3, comparison is made with data from a test
series by Kotsovos and Newman [16]. Data from two
s0 ˆ sn
triaxial tests are used for comparison, one with a con
ning
Update kI ; kDi stress of 35 N/mm2 and the second with a con
ning stress
If F…s0 † < tol, Exit loop, otherwise of 70 N/mm2 . The graphs shown in Fig. 4 have been
continue plotted in the compression positive convention of the
Close loop experimental data. The comparison shows that the model

Table 1 Material properties for stress path tests

2
Example E (N/mm ) n fc (N/mm2) ft (N/mm2) ec e0
1 35 000 0.15 40 3.2 0.002 0.0016
2 32 000 0.22 32 2.4 0.0021 0.003
3 37 000 0.15 46.9 3.0 0.0022 0.003

Proc Instn Mech Engrs Vol 216 Part L: J Materials:Downloaded

Design andfrom pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Applications L00202 # IMechE 2002
 LOCAL PLASTIC SURFACES FOR CRACKING AND CRUSHING IN CONCRETE 263

Fig. 3 Uniaxial tension and compression response

Fig. 4 Comparison with triaxial test data from Kotsovos and Newman

predicts the peak stresses with reasonable accuracy, but FE example 1

that the numerical response shows less ductility with
increasing con
nement than that observed in the experi- The
rst example is an unreinforced notched beam tested by
ments. Since it appears that this inaccuracy only becomes Petersson [17]. The experimental beams had cantilever
pronounced for relatively high con
ning stresses, which sections to compensate for the beam weight. The experi-
are relatively unusual in practice, this inaccuracy is consid- mental arrangement and the mesh of eight-noded elements
ered acceptable. used for the analysis, which makes use of symmetry, are
illustrated in Fig. 5. The material properties used for the
analysis are shown in Table 2. The numerical and experi-
mental responses are compared in Fig. 6, in which a
numerical crack plot for the elements adjacent to the
5 FINITE ELEMENT ANALYSIS EXAMPLES notch is also presented.

The results from two FE analyses are presented. The
rst is FE example 2
of an unreinforced notched fracture beam test and the
second is of a reinforced concrete beam test without shear The second example is the analysis of a rectangular rein-
reinforcement. The examples are relatively standard but they forced concrete beam tested by Bresler and Scordelis [18].
do illustrate that the model performs reasonably in an FE The beam is simply supported and is loaded with a central
context, and in both cases the overall responses and crack point load. It has two layers of longitudinal reinforcement
plots are reasonable when compared with the experimental but no shear reinforcement. The testing arrangement and the
results. reinforcement are shown in Fig. 7, which also shows the
L00202 # IMechE 2002 Downloaded from pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Proc Instn Mech Engrs Vol 216 Part L: J Materials: Design and Applications
264 A D JEFFERSON

Fig. 5 Petersson’s notched beam

Table 2 Material properties for FE examples

2
FE example E (N/mm ) n fc (N/mm2) ft (N/mm2) ec e0 Gf (N/mm)
1 30 000 0.2 45 3.33 0.0023 — 0.125
2 25 000 0.2 22.6 3.35 0.0018 0.0032 —

half-mesh used in the analysis. The concrete is simulated properties are E ˆ 205 000 N/mm2 , fy ˆ 553 N/mm2 and
with eight-noded elements and the reinforcement with three- H ˆ 15 000 N/mm2 . The numerical and experimental
noded bar elements. The concrete material properties used responses are compared in Fig. 8, in which a numerical
in the analysis are given in Table 2. The reinforcement crack plot is also given.

Fig. 6 Results from notched beam analysis

Proc Instn Mech Engrs Vol 216 Part L: J Materials:Downloaded
Design andfrom pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Applications L00202 # IMechE 2002
 LOCAL PLASTIC SURFACES FOR CRACKING AND CRUSHING IN CONCRETE 265

Fig. 7 Reinforced concrete beam test by Bresler and Scordelis

Fig. 8 Results from reinforced concrete beam analysis

6 CLOSING REMARKS analysis of structures under cyclic loading. A plastic direc-

tional damage model that addresses this problem is being
The paper illustrates that a useful constitutive model developed by the author and will be described in a future
for concrete can be developed using simple local yield publication.
functions to represent both compressive and tensile beha- The two FE analysis examples are insuf
cient to provide
viour. The major device necessary to accomplish this is a a full evaluation of the model and serve mainly to illustrate
directional–isotropic transition function that distinguishes that the model works in a reasonable manner when imple-
compressive crushing behaviour from tensile softening mented in an FE program. However, further analyses, not
behaviour. reported here, have been undertaken, and these show that the
The chief aw with the model is that unloading behaviour model is able to represent the behaviour from a range of
in tension is elastic, which is unrealistic for concrete, and experimental tests on reinforced and unreinforced concrete
therefore the model, as presented, is not suitable for the specimens with reasonable accuracy.
L00202 # IMechE 2002 Downloaded from pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Proc Instn Mech Engrs Vol 216 Part L: J Materials: Design and Applications
266 A D JEFFERSON

ACKNOWLEDGEMENT 8 Hoek, E. and Brown, E. T. Empirical strength criterion for

rock masses. J. Geotech. Div. ASCE, 1980, 9(9), 1013–1035.
The author would like to thank FEA Limited for collaborat- 9 Este, G. and Willam, K. Fracture energy formulation for
ing in and sponsoring this work and for their continued inelastic behavior of plain concrete. J. Engng Mechanics
interest in the development of accurate FE material models. ASCE, 1994, 120(9), 1983–2011.
10 Carol, I. and Bazant, Z. P. New developments in micro-plane
and multicrack models for concrete. Proceedings of FRAM-
REFERENCES COS2 Vol. II (Ed. F. H. Wittmann), 1995, pp. 841–856
(Aedi
catio, Germany).
11 Kroplin, B. and Weihe, S. Aspects of fracture induced aniso-
1 van Mier, J. G. M. Fracture Processes of Concrete, 1997 (CRC tropy. In Proceedings of 5th International Conference on
Press, Florida, U.S.A.). Computational Plasticity (COMPLAS5), Barcelona, 1997,
2 ACI Committee 446 Finite element analysis of concrete pp. 255–279.
structures. ACI Report 446.3R-97, 1998. 12 Ortiz, M. and Simo, J. C. An analysis of a new class of
3 Owen, D. R. J., Figueiras, J. A. and Damjanic, F. Finite integration algorithms for elastoplastic constitutive relations.
element analysis of reinforced and prestressed concrete struc- Int. J. Numer. Meth. Engng, 1986, 23, 353–366.
tures including thermal loading. Computer Meth. in Appl. 13 Simo, J. C. and Taylor, R. L. Computational Inelasticity, 1998
Mechanics and Engng, 1983, 41, 323–366. (Springer, New York).
4 Cervera, M., Hinton, E. and Hassan, O. Nonlinear analysis of 14 Hordijk, D. A. Local approach to fatigue of concrete.
reinforced concrete plate and shell structures using 20-noded PhD thesis, Delft University of Technology, The Netherlands,
isoparametric brick elements. Computers and Struct., 1987, 1991.
25(6), 845–869. 15 Kupfer, H. B., Hilsdorf, H. K. and Rusch, H. Behaviour
5 Meschke, G., Lackner, R. and Mang, H. A. An anisotropic of concrete under biaxial stresses. J. ACI, 1969, 66(8),
elsatoplastic-damage model for plain concrete. Int. J. Numer. 656–666.
Meth. Engng, 1998, 42, 703–727. 16 Kotsovos, M. D. and Newman, J. B. A mathematical descrip-
6 Lee, J. and Fenves, G. L. Plastic-damage model for cyclic tion of the deformable behaviour of concrete under complex
loading of concrete structures. J. Engng Mechanics ASCE, loading. Mag. Concr. Res., 1979, 31, 77–90.
1998, 124, 892–900. 17 Petersson, P. Crack Growth and Development of Failure Zones
7 Bazant, Z. P., Caner, F. C., Carol, I., Adley, M. D. and in Plain Concrete and Similar Materials, 1981 (Lund Institute
Akers, S. A. Micro-plane model M4 for concrete. I: formula- of Technology, Sweden).
tion with work conjugate deviatoric stress. J. Engng Mechanics 18 Bresler, B. and Scordelis, A. C. Shear strength of reinforced
ASCE, 2000, 126(9), 944–953. concrete beams. J. ACI, January 1963, 51–72.

Proc Instn Mech Engrs Vol 216 Part L: J Materials:Downloaded

Design andfrom pil.sagepub.com at PENNSYLVANIA STATE UNIV on September 15, 2016
Applications L00202 # IMechE 2002