Implementation of the Modified Compression Field Theory in a Tangent

Stiffness-Based Finite Element Formulation

Wilkins Aquino* and Ibrahim Erdem
School of Civil and Environmental Engineering, Cornell University, Ithaca, NY 14853, USA
* Corresponding author. Tel: (607) 255-3294, Fax: (607)-255-9004, email: [email protected]

Abstract
A finite element implementation of the modified compression field theory (MCFT) using
a tangential formulation is presented in this work. Previous work reported on
implementations of MCFT has concentrated on secant formulations. This work describes
details of the implementation of a modular algorithmic structure of a reinforced concrete
constitutive model in nonlinear finite element schemes that use a Jacobian matrix in the
solution of the nonlinear system of algebraic equations. The implementation was verified
and validated using experimental and analytical data reported in the literature.

Keywords
Finite elements; Reinforced concrete; Compression field theory; Nonlinear analysis.
1. Introduction
The finite element method is the most widely used numerical tool for analyzing complex
reinforced concrete structures such as tanks, offshore structures, curved bridges, shear
walls with irregularities, etc. In order to analyze such structures and obtain their full
behavior (cracking, yielding, failure, failure type, stiffness, and ductility), robust material
models are essential.

Many different concrete material models have been reported in the literature. Material
models for finite element analysis are formulated depending on the type of application

and range from simple linear elastic to complex viscoplastic-damage models [1-17].
Although the theory of plasticity is very suitable for modeling path-dependent behavior,
oftentimes models based on this theory need a very large number of parameters without
clear physical significance and that are difficult to obtain. In many practical situations, it
is of interest to estimate the monotonic nonlinear behavior of reinforced concrete
structures. For this cases, theories such as rotating and fixed crack models such as the
modified compression field theory (MCFT) by Vecchio and Collins [4], and the fixedangle softened-truss model (FA-STM) by Hsu and Zhang [16, 17] present attractive
alternatives because of their simplicity in implementation and ease of use.

The implementation of MCFT has been addressed in secant finite element formulations
by Vecchio [6], but a detailed treatment of its implementation in tangent stiffness-based
formulations has not been reported in the literature to the best knowledge of the authors.
Many available commercial and in-house finite element codes use tangential formulations
to solve the nonlinear system of equations arising from the finite element discretization.
This formulation requires the calculation of a material tangent stiffness matrix from the
constitutive model during the iterative solution process, which may not be a trivial step in
many circumstances.

In this paper, a detailed implementation of MCFT in a tangent stiffness-based finite

element scheme is presented. First, the fundamental concepts from MCFT are described.
Then, the finite element formulation is presented, addressing the issues of stress updating
and computing the tangent stiffness matrix from the material model. The different

components of a reinforced concrete material model based on MCFT are explained

followed by numerical examples and the validation of the implementation.
2. The Modified Compression Field Theory (MCFT)
MCFT was introduced in 1986 by Vecchio and Collins [4] and it was developed from the
original compression field theory developed by Mitchell and Collins [18]. In these
theories, relationships between average stresses and strains were postulated based on
experimental observations. Cracks in these theories are treated in a distributed sense.

The following are the assumptions made in MCFT.

There is a one-to-one correspondence between stresses and strains. That is, the model
is non-linear elastic.

Average stresses and strains are used for all calculations.

There is perfect bonding between reinforcing bars and concrete (i.e. no slip).

The longitudinal and transverse reinforcing bars are uniformly distributed.

The principal strain directions are coincident with the principal stress directions.

Three main components defined MCFT: equilibrium equations, constitutive relationships,

and load transmission conditions at cracks. These components are extensively described
in Reference [4] and are briefly outlined herein for completeness.
2.1

Constitutive Relationships in MCFT

The constitutive relationships involved in MCFT are presented in principal stress-strain

space. Vecchio and Collins [4] reported that the principal compressive stress at a point in

concrete depends on both the principal tensile and compressive strains, while the
principal tensile stress was only dependent on the principal tensile strain (i.e. decoupled
from the compressive strain).

The compressive stress is calculated in MCFT as

c2

2

c2
c2
= c 2 max 2

,
o o

(1)

where c 2 is the minimum principal stress (compression in MCFT), o is the strain at the
peak stress in a uniaxial compression test, and c 2 is the minimum principal strain. The
factor c 2 max accounts for the state of biaxial tension-compression state and is calculated
as

c 2max =

f c'

0.8 0.34 c1
o

(2)

where f c' is the uniaxial compressive strength of the concrete.

The stress-strain curve of concrete under tension is defined as linear elastic up to cracking
as

c1 = Ec c1 ,

(3)

where Ec is the modulus of elasticity of concrete, c1 is the maximum principal tensile

stress, and c1 is the maximum principal strain. After cracking, the tensile stress in the
concrete is taken as

 c1 =

f cr
.
1 + 200 c1

(4)

The stress-strain behavior of the steel is assumed to be elastic-perfectly plastic in this

paper. That is, the material is assumed to be linearly elastic up to yielding and then
remain at this stress level for strains greater than the yield strain.
2.2

Equilibrium Equations

Equilibrium equations are used to calculate the average stresses at a point from the
concrete and steel contributions as

x = cx + sx sx
y = cy + sy sy ,

(5)

xy = cxy
where x , y and xy are the average stress in the X-direction, average stress in the Ydirection, and average shear stress calculated at a material point in reinforced concrete.
The quantities cx and cy represent axial stresses in the concrete, while sx and sy
represent axial stresses in the steel. Neglecting the average shear stress contribution of the
steel, the shear stress xy is assumed to be equal to the shear stress carried by the
concrete, cxy . Reinforcement ratios in the X and Y directions are represented
by sx and sy , respectively. These ratios are given as area of steel to area of concrete.

2.3

Load Transmission Conditions at Cracks

The stress-strain relationships described above are valid in an average sense. However,
stresses in the steel at cracks will be higher than their average values. Therefore, it is

necessary to ensure that the steel reinforcement is capable of transmitting the demanded
average tension stresses across cracks.

Vecchio and Collins [4] derived the following conditions which are used to ensure that
enough capacity exists in the concrete and steel to properly transmit tension across
cracks.

sxcr = sx + ( c1 + ci + vci / tan ) / sx

yield
yx

sycr = sy + ( c1 + ci vci / tan ) / sy yield

(6)

yy

In the above equations, sxcr and sycr are the axial stresses in the steel in the X and Y
directions at the crack face, respectively, xxyield is the yield stress in the X direction, yyyield
is the yield stress in the Y direction, ci is the compressive stress acting on the crack, ci
is the shear stress acting along the crack, is the angle of inclination of cracks with
respect to the X-axis.

The maximum shear that can act along the crack is calculated as [4]
vci max =

f c'
0.31 + 24 w /(a + 16)

(7)

Where w is the estimated crack width, a is the maximum aggregate size. The crack
width, w , is calculated as
w = c1s

(8)

The spacing of the cracks along the principal direction tensile direction, s , can be
calculated from the spacing of the cracks perpendicular to the reinforcement in Xdirection and Y-direction as
s =

(9)

sin cos
+
smx
smy

where smx and smy are the average spacing of the cracks perpendicular to the X and Y
directions, respectively.
3. Implementation of MCFT in Finite Element Analysis

This section describes how MCFT can be implemented in a displacement-based finite

element formulation that uses a tangent stiffness matrix. The crucial aspects for this
implementation are how stresses are updated and how the material tangent stiffness is
computed at integration points in iterative incremental solution schemes. Only material
nonlinearity is considered herein.
3.1

Finite Element Formulation of the Nonlinear Problem

The principle of virtual work is commonly used to derive the finite element formulation
of nonlinear stress analysis problems. The principle of virtual work can be expressed as

{ } { } d = { u} { } d + { u} {b} d + P u
T

(10)

where { u} is an admissible virtual displacement field in the domain and its boundary

, { } is the corresponding virtual strain field, { } is the stress vector, { } represent

the tractions applied over part of the boundary, {b} represent body forces, and Pi are

applied point forces . The domain is divided into n finite elements and the solution

{u} within each element is approximated using interpolation functions [ N ] as

{u} = [ N ]{u e }

(11)

where {u e } are the element nodal displacements. Strains are computed as

{ } = [ B ]{u e }

(12)

where [ B ] is a matrix representing the partial derivatives of the shape functions with
respect to position. Using the same interpolation scheme for the virtual displacement and
virtual strain fields (i.e. Galerkin approach), substituting Equations (12) and (11) into
Equation (10) , and adding all elements contributions, yields

{ U }T [ B]T { }d e {Q} = 0 ,
elements e

(13)

where { U } is the virtual displacement vector containing all degrees-of-freedom in the

domain,{Q} is the external load vector defined as
{Q} =

[ N ] { } d + [ N ] {b} d

elements

+ {Ppo int } ,

(14)

and { Ppo int } is a vector of point loads.

Since Equation (13) must hold for all choices of { U } , then the expression in the square
brackets must vanish. Then,

[ B] { }d
T

= {Q} .

(15)

elements e

Defining the internal force resisting vector, {I}, as

{I } =

[ B] { }d
T

(16)

elements e

the solution of the boundary value problem reduces to the solution of a non-linear system
of algebraic equations. This system can be described as
{Q} {I } = 0 .

(17)

The Newton-Raphson or modified Riks methods [19-21] are commonly used to solve the
system of equations described in Equation (17). These iterative solution procedures
require the calculation of a tangent stiffness or Jacobian matrix defined as

{I }
.
U

{
}

[ KT ] =

(18)

The global Jacobian matrix is assembled from the contributions of the local material
tangent stiffness as

[ KT ] = [ B]T [ D ][ B ]d ,

(19)

elem e

where the local material tangent stiffness has been defined as

[ D] =

{ }
.
{ }

(20)

The main tasks that need to be performed by the constitutive model implementation in the
finite element scheme shown above are updating the stresses to compute the internal
force resisting vector using Equation (16), and computing the Jacobian matrix using
Equation (20). These quantities are computed at each Gauss integration point in the finite

element mesh. The calculation of a Jacobian matrix is many times a non-trivial task. This
is the case in MCFT mainly because of the check for load transmission conditions at
cracks.
3.2

Implementation of MCFT in a Finite Element Scheme

This section shows how the equations of MCFT enter the finite element formulation
described above. The algorithm described in this section is composed of two main parts:
stress updating and computation of the material Jacobian matrix.

The layout of the reinforced concrete model implementation is shown in Figure 1. The
main finite element program supplies the material model subroutine with strains at
integration points for computing stresses and the material Jacobian matrix. Stress
updating occurs separately for steel and concrete and these tasks are implemented as
separate subroutines. Similarly, the steel and concrete contributions to the Jacobian
matrix are computed is a separate subroutine. Box 1 shows the organization of the
reinforced concrete material subroutine.
a) Main Material Model Subroutine

Steel and concrete stresses are computed in separate subroutines called by the RC
Material Subroutine. The steps involved in this process are described in Box 1. The
implementation in this paper assumes that bars are oriented in the global X and Y
directions. If bars with arbitrary orientations are used, Equations (5) and (6) must be
modified to take directionality of steel into account.

10

Steel stresses are computed as shown in Box 2. This implementation assumes elasticperfectly plastic behavior of the steel. More sophisticated steel models that take strain
hardening into consideration can also be easily implemented.
b) Concrete Stress Subroutine

Concrete behavior is described using three different states of stress: biaxial compression,
biaxial tension, and biaxial tension-compression. In all three cases, the stress state is
determined in the principal stress space. Therefore, only principal strains are computed
and used to determine principal stresses. Once principal stresses are determined, they are
transformed back to the original coordinate system and passed back to the main material
subroutine.

The subroutine for computing concrete stresses is described in Box 3. Once principal
strains are computed, the state of stress (i.e. biaxial compression, biaxial tension, or
tension compression) is inferred from the signs of the principal strains. Principal strains
are sent to different subroutines depending on the stress state for computing principal
stresses.
c)

Biaxial Stress Subroutine

The structure of the biaxial stress subroutine is illustrated in Box 4. If both principal
strains are positive, then a state of biaxial tension exists. For this case, the stress-strain
behavior is assumed to be uniaxial in each direction. Principal stresses are computed
separately from strains for each axis using Equations (3) and (4).

11

If both strains are negative, the material point is in a state of biaxial compression. The
implementation given in this paper uses the equations described in Reference [8]. These
equations are given as


K1 = 1 0.92 2 0.76 2
fc '
fc '

p1 = K1 f c '
p1 = K1 co '

(21)

2
1 = p1 2 1 1 , 0>1 > p1
p1 p1

2
1 = p1 1 1 p1 , p1 >1 > 2 co
2 co p1

The same procedure is used for computing 2 , but the roles of 1 and 2 in (21) are
reversed. Notice that the above system of equations is nonlinear, so it needs to be solved
iteratively as shown in Box 4. The direct substitution method shown in Box 4 produced
fast convergence for the examples investigated in this research. Other solution methods
for nonlinear algebraic equations such as the Newton-Raphson scheme can also be used.
d) Modified Compression Field Theory (MCFT) Subroutine

MCFT is used to determine the stress state when biaxial tension-compression stress states
exist as illustrated in Box 5. The compressive and tensile stresses 2 and 1 , respectively,
are determined using the principal strains and Equations (1)-(4).

A fundamental aspect of MCFT is that the tension field transmitted across cracks is
limited by two main criteria: yielding of reinforcement crossing a crack and the
12

maximum shear stress that can be carried at the surface of a crack. The admissible tension
field conditions across the cracks are checked using Equations (6)-(9). The
implementation of the enforcement of these conditions is shown in Box 5. Notice that
represents the crack orientation, while represents the principal direction angle [4].

The first condition considers yielding of the reinforcement at the crack in the X-direction,
while the reinforcement in the Y-direction remains below its yield limit. The shear
stress, vci , needed to sustain the average tensile stress, 1 , is computed using the first
Equation in (6). If the maximum shear stress admissible at a crack, vci max , is exceeded,
then a new tensile stress, 1 1 , is calculated so that vci = vci max . The same procedure is
repeated for a second condition in which the steel in the Y-direction reaches yielding at
the crack face, while the steel in the X-direction remains below its yield point. If the
maximum shear is exceeded, a new admissible tensile stress, 2 1 , is computed in order to
maintain the shear stress at the crack below its maximum value. The third condition
computes the maximum tensile stress, 3 1 , that can be developed when the reinforcement
in the X and Y directions reaches its yield limit. Finally, the average concrete tensile
stress selected is the minimum of the above four possible choices ( 1 , 1 1 , 2 1 , and 3 1 ).
e)

Jacobian Matrix Subroutine

Once the stresses are updated, the material stiffness or Jacobian matrix needs to be
determined as explained previously. The material stiffness matrix in Equation (20)
involves the derivative of the stresses with respect to the strains. Although closed-form
expressions for the concrete Jacobian matrix can be calculated using the stress-strain
13

relationships described in this paper, the resulting expressions are cumbersome and their
implementation is lengthy, increasing the possibilities of introducing errors in the code. A
simpler approach is to use a finite difference approximation of the material Jacobian
matrix. Because of the modular structure of this implementation, the process of
computing this finite approximation is fairly simple as illustrated in Box 6. The
computational cost of following this approach proved to be low in comparison with the
benefits of code simplicity.

To compute the concrete contribution of the material Jacobian matrix, first an increment
of strain is defined as shown in Box 6. Then, forward and backward strain steps are
created using this increment of strain. The Jacobian matrix is formed column by column,
as shown in Box 6, by progressive calls to the concrete stress update subroutine using the
incremented strains. The steel contribution can be easily computed in closed form as
shown in Box 6. The total material stiffness matrix is then formed by adding the concrete
and the steel contributions.

It is important to realize that the Jacobian matrix computed using the above procedure
will be in general non-symmetric. Therefore, it is important to bear this in mind when
invoking solvers that may try to exploit symmetry to increase computational speed. Often
finite element software use symmetric solvers by default and non-symmetric solvers have
to be activated by the user.

14

4. Validation and Verification Examples

The implementation described above was incorporated into a user-defined subroutine in a

tangent stiffness-based finite element code. Shear panels tested by Vecchio, Collins, and
co-workers at the University of Toronto were used to validate the implementation [4, 22,
23]. In addition, the results obtained with the described implementation were compared to
solutions obtained with other software implementation of MCFT [24]. Four node
quadrilateral elements with full integration were used for all the analyses.

Square reinforced concrete panels with a side dimension of 890 mm were used for the
validation. The thickness of the panels was 70 mm. The maximum aggregate size used
for the concrete was 6 mm. The panels were reinforced with two layers of 6 mm diameter
reinforcing bars in each direction and the steel had a Youngs modulus of 200 GPa.
Other material properties are given in Table 1.

Three different panels were investigated: two solid panels, PV20 and PB5, and one with a
square opening at its center, PC5. Details of the panel geometry, reinforcement ratios,
and loading conditions are also given in Table 1. Panel PV20 was loaded under pure
shear, Panel PC5 was loaded under biaxial compression and shear, and Panel PB5 was
loaded under biaxial tension-compression and shear.

Shear stress-strain response curves for panels PV20 and PB5 are shown in Figure 2 and
Figure 3, respectively. The plots show results from laboratory experiments, the current
MCFT implementation, and the software Membrane 2000 [24]. As can be seen from
Figures 2 and 3, the results obtained with the proposed implementation of MCFT follows
15

closely the results obtained with Membrane 2000. In addition, the initial stiffness and
initial cracking predicted by the current MCFT implementation are in close agreement
with the experimental values obtained for both specimens. The predicted maximum shear
stress satisfactory agrees with the experimental values in both cases, but the error was
more pronounced in the case of Panel PB5. The largest discrepancy between the
predicted and the experimental responses occurred in the postpeak branch. This result
was expected as the failure mechanisms that occurred in the actual panels such as
slippage at cracks or localized crushing cannot be captured by MCFT.
4.1

Panel with Square Opening (PC5)

In order to test the performance of the current MCFT implementation with a panel
structure that results in a non-uniform stress distribution, a reinforced concrete panel with
a square opening at its center was analyzed. This configuration results in stress conditions
that span the four states considered by the implementation: tension-tension, compressioncompression, and tension-compression. The results of the finite element analysis were
compared to experimental and numerical analysis results reported in Reference [22].

Figure 4 shows plots of the average shear stress-strain behavior of Panel PC5. Figure 5
shows the finite element mesh used for this problem and the non-uniform stress
distribution that results around the opening. It can be observed in the experimental
results that unloading occurred at some instances during testing. This unloading was the
result of test stops for removal and reading of the instrumentation. As can be seen from
Figure 4, the initial stiffness, cracking stress, and ultimate shear strength of the specimen
were satisfactorily estimated by the current finite element implementation whereas the
16

ultimate displacement was lower than the experimental counterpart. The finite element
results reported by Chan [22] are also shown in Figure 4.
5. Comments on Mesh Objectivity

It is well known that finite element implementations of softening material behavior result
in non-objective meshes. That is, results are dependent upon the mesh size used for the
analysis if no steps are taken to prevent this phenomenon. Several approaches have been
proposed to alleviate mesh dependency problems such as non-local formulations,
enriched interpolation functions, and the use of a characteristic element length in the
constitutive models [25-27]. The implementation of MCFT presented in this paper
displays mesh dependency pathologies since no regularization was incorporated into the
formulation. This can be fixed by using the approach proposed in References [25, 27] in
which constitutive laws are modified to incorporate a characteristic length deduced from
energy constraints. This approach ensures that energy dissipation per unit volume will not
depend on the element size selected.

Another approach that could be used with the implementation described in this paper is to
select the size of the element to be approximately equal to the expected crack spacing.
This strategy was used by the authors and produced adequate results. This spacing is
consistent with the experiments from which MCFT equations were derived. The authors
plan in the future to investigate different regularization schemes and their incorporation
into the current MCFT implementation in order to devise a more robust strategy for
alleviating mesh dependency.

17

6. Conclusions

In this paper, a simple implementation of the Modified Compression Field Theory

(MFCT) has been presented. The algorithm can be used with any tangent stiffness-based
finite element formulation. The implementation was validated using experimental data
from tests performed on reinforced concrete panels. It was found that the results obtained
with the current implementation of MCFT are in agreement with experimental and
analytical results reported in the literature. The algorithm described in this paper was
implemented using FORTRAN 77. The code is available for research and educational use
from the authors.
7. Acknowledgements

Wilkins Aquino thanks Profs. David Pecknold and Daniel Kuchma from the University
of Illinois at Urbana-Champaign (UIUC) for their thorough and caring advice on
computational mechanics and the modified compression field theory, which led to this
paper.
8. References

1.

Imran, I. and Pantazopoulou, S.J., Plasticity Model for Concrete under Triaxial
Compression. Journal of Engineering Mechanics-ASCE, Vol. 127, No. 3, 2001,

pp. 281-290.
2.

Lee, J.H. and Fenves, G.L., Plastic-Damage Model for Cyclic Loading of
Concrete Structures. Journal of Engineering Mechanics-ASCE, Vol. 124, No. 8,

1998, pp. 892-900.

3.

Lee, J. and Fenves, G.L., A Return-Mapping Algorithm for Plastic-Damage

Models: 3-D and Plane Stress Formulation. International Journal for Numerical

Methods in Engineering, Vol. 50, No. 2, 2001, pp. 487-506.

18

4.

Vecchio, F.J. and Collins, M.P., The Modified Compression-Field Theory for
Reinforced-Concrete Elements Subjected to Shear. Journal of the American

Concrete Institute, Vol. 83, No. 2, 1986, pp. 219-231.

5.

Vecchio, F.J., Nonlinear Finite-Element Analysis of Reinforced-Concrete

Membranes. Aci Structural Journal, Vol. 86, No. 1, 1989, pp. 26-35.

6.

Vecchio, F.J., Reinforced Concrete Membrane Element Formulations. Journal of

Structural Engineering-ASCE, Vol. 116, No. 3, 1990, pp. 730-750.

7.

Vecchio, F.J. and Selby, R.G., Toward Compression-Field Analysis of

Reinforced-Concrete Solids. Journal of Structural Engineering-Asce, Vol. 117,

No. 6, 1991, pp. 1740-1758.

8.

Vecchio, F.J., Finite-Element Modeling of Concrete Expansion and Confinement.

Journal of Structural Engineering-Asce, Vol. 118, No. 9, 1992, pp. 2390-2406.

9.

Polak, M.A. and Vecchio, F.J., Nonlinear-Analysis of Reinforced-Concrete Shells.

Journal of Structural Engineering-Asce, Vol. 119, No. 12, 1993, pp. 3439-3462.

10.

Palermo, D. and Vecchio, F.J., Behavior of Three-Dimensional Reinforced

Concrete Shear Walls. ACI Structural Journal, Vol. 99, No. 1, 2002, pp. 81-89.

11.

Zhou, C.E. and Vecchio, F.J., Nonlinear Finite Element Analysis of Reinforced
Concrete Structures Subjected to Transient Thermal Loads. Computers and

Concrete, Vol. 2, No. 6, 2005, pp. 455-479.

12.

Hsu, T.T.C., Nonlinear-Analysis of Concrete Membrane Elements. ACI Structural

Journal, Vol. 88, No. 5, 1991, pp. 552-561.

13.

Pang, X.B.D. and Hsu, T.T.C., Behavior of Reinforced Concrete Membrane

Elements in Shear. ACI Structural Journal, Vol. 92, No. 6, 1995, pp. 665-679.

14.

Belarbi, A. and Hsu, T.T.C., Constitutive Laws of Softened Concrete in Biaxial

Tension-Compression. Aci Structural Journal, Vol. 92, No. 5, 1995, pp. 562-573.

15.

Pang, X.B.D. and Hsu, T.T.C., Fixed Angle Softened Truss Model for Reinforced
Concrete. ACI Structural Journal, Vol. 93, No. 2, 1996, pp. 197-207.

16.

Hsu, T.T.C. and Zhang, L.X., Nonlinear Analysis of Membrane Elements by

Fixed-Angle Softened-Truss model. ACI Structural Journal, Vol. 94, No. 5, 1997,

pp. 483-492.

19

17.

Wang, T.J. and Hsu, T.T.C., Nonlinear Finite Element Analysis of Concrete
Structures using New Constitutive Models. Computers & Structures, Vol. 79, No.

32, 2001, pp. 2781-2791.

18.

Mitchell, D. and Collins, M.P., Diagonal Compression Field Theory-A Rational

Model for Structural Concrete in Pure Torsion. ACI Journal, Vol. 71, No. 8,

1974, pp. 396-408.

19.

Bathe, K.J., Finite Element Procedures. Prentice Hall, 1996.

20.

Crisfield, M.A., Non-Linear Finite Element Analysis of Solids and Structures.

Vol. 1. Wiley, 1991.

21.

Belytschko, T., Liu, W.K., and Moran, B., Nonlinear Finite Elements for
Continua and Structures. Wiley, 2000.

22.

Chan, C.C.L., Testing of Reinforced Concrete Membrane Elements with

Perforations, M.S., Civil Engineering, University of Toronto, 1989, Toronto, CA

23.

Bhide, S.B. and Collins, M.P., Reinforced Concrete Elements in Shear and
Tension. 1987, University of Toronto: Toronto, CA.

24.

Bentz, E.C., Membrane 2000. 2000: Toronto, CA.

25.

Bazant, Z.P. and Cedolin, L., Stability of Structures. Oxford University Press,
1991.

26.

Ortiz, M., Leroy, Y., and Needleman, A., A Finite-Element Method for Localized
Failure Analysis. Computer Methods in Applied Mechanics and Engineering,

Vol. 61, No. 2, 1987, pp. 189-214.

27.

Bontempi, F. and Malerba, P.G., The Role of Softening in the Numerical Analysis
of RC Framed Structures. Structural Engineering and Mechanics, Vol. 5, No. 6,

1997, pp. 785-801.

20

9. Table Captions

Table 1. Details of the test specimens.

10. Figure Captions

Figure 1. Layout of a reinforced concrete material model implementation.

Figure 2. Shear stress-strain behavior of Panel PV20.
Figure 3. Shear stress-strain behavior of Panel PB5.
Figure 4. Average shear stress-strain response of Panel PC5.
11. Box Captions

Box 1. Reinforced concrete material subroutine.

Box 2. Steel stress subroutine.
Box 3. Concrete stress subroutine.
Box 4. Biaxial stress subroutine.
Box 5. Concrete MCFT subroutine.
Box 6. Jacobian matrix subroutine.

21

Table 1
Details of the test specimens
Specimen

fc
(MPa)

x (%)

xxyield
(MPa)

y (%)

yyyield
(MPa)

Loading
xx : yy : xy

0: 0: 1
PV20

19.6

0.0018

1.785

460

0.885

297

1: -1: 1
PB5

23.5

0.0018

1.085

415

0.32: 0.32: 1
PC5

27.3

0.0018

1.65

390

0.82

390

RC Material Subroutine
Call Steel Stress Subroutine
Call Concrete Stress Subroutine
Compute total stress from steel and
concrete stress contributions using Eq. (5)
Call Jacobian matrix Subroutine
Return updated stresses and Jacobian Matrix
End RC Material Subroutine
Box 1. Reinforced concrete material subroutine.

Steel Stress Update Subroutine

*Elastic-perfectly plastic*
If strain component < yield strain
xx = E xx

or

yy = E yy
Else
yield

xx = xx
or

yy = yyyield
end If
Return steel stresses
End Steel Stress Subroutine
Box 2. Steel stress subroutine.

Concrete Stress Subroutine

Compute principal strains 1 and 2
Compute principal plane angle
If 1 and 2 are both positive or negative
Call Biaxial Stress Subroutine
Else
Call MCFT Subroutine
End If
Compute concrete average stresses by
transforming principal stresses back to
original coordinate system
Return concrete average stresses
End Concrete Stress Subroutine
Box 3. Concrete stress subroutine.

Biaxial Stress Subroutine

*Biaxial tension*
If 1 and 2 >0

Compute 1 using Eqs. (3 ) and (4)

Else
*Biaxial compression*
Initialize 1 = 2 = 0
i=1
Norm(i)=0
Error=9999
Do while Error < tolerance and i< Max iterations
Compute K1 and K2 as shown in Eqs. (21)
Compute
1 p = K1 co

2 p = K 2 co
Compute 1 and 2 as shown in Eqs. (21)
norm(i + 1) = 12 + 2 2

Error =| norm(i ) norm(i + 1) |

increment i
end Do while
End If
End Biaxial Stress Subroutine
Box 4. Biaxial stress subroutine.

Concrete MCFT Subroutine

Compute 1 and 2 using Eqs.(1)-(4)
*Check if maximum shear stress at cracks is not exceeded as folows. Neglect
compressive stress across cracks*

Compute crack angle

If 1 and 2 0

=
Else

End if
Compute , s , vci max using Eqs. (7)-(9)
* First check: use equilibrium in X-direction in Eq. (6)*
Compute vci = x ( xxyield sx ) c1 tan ( )

If vci > vci max

1

1 = x ( xxyield sx )

vci max
tan ( )

End If
* Second check: use equilibrium in Y-direction in Eq. (6)*
1 y ( yyyield sy )
Compute vci =
tan ( )
If vci > vci max
2

1 = y ( yyyield sy ) m vci max tan ( )

End If
*Third check: verify that reinforcement does not yield in X and Y directions
simultaneously*
3
1 = x ( xxyield sx ) sin2 ( ) + y ( yyyield sy ) cos2 ( )
*Select admissible tensile stress across cracks*
1 = min ( 1 ,1 1 ,2 1 ,3 1 )
End Concrete MCFT Subroutine
Box 5. Concrete MCFT subroutine.

Jacobian Subroutine

Define a small strain increment =

co
1000

*Compute incremented strains*

xx +
1
{ incx } = yy

xy
xx
2
{ incx } = yy

xy
*Compute stresses using above strains from Concrete Stress
Subroutine*
{1 incx } = Call Concrete Stress Subroutine ( {1 incx } )

{ } = Call Concrete Stress Subroutine( { } )

2

incx

incx

*Compute Column 1 of Jacobian matrix*

1
incx 2 incx }
{
Dco (1: 3,1) =
2
Repeat above procedure for Columns 2 and 3.
*Steel Contribution*
If xx < xx yield
Esx = Es
Else
Esx = 0
End If
Repeat for Y component
0
0
x Esx

[ Ds ] = 0 y Esy 0
0
0
0

[ D ] = [ Dco ] + [ Ds ]
End Jacobian Subroutine

Box 6. Jacobian matrix subroutine.

FEA Program

Material Model

Update stress

Compute stresses
in concrete

Compute Jacobian

Compute stresses
in steel

Concrete
contribution

Steel
contribution

Figure 1. Layout of a reinforced concrete material model implementation.

5
4.5

Shear stress (MPa)

4
3.5
3
2.5
2

Experimental, Ref [4]

1.5

Membrane 2000

Current MCFT Implementation

0.5
0
0

10

12

14

Shear strain (10^3 mm/mm)

Figure 2. Shear stress-strain behavior of Panel PV20.

Shear stress (MPa)

2.5

1.5

Experimental, Ref [23]

Membrane 2000
1

Current MCFT Implementation

0.5

0
0

10

12

Shear strain (x10^3 mm/mm)

Figure 3. Shear stress-strain behavior of Panel PB5.

4.5
4

Shear stress (MPa)

3.5
EXPERIMENTAL
FEA by Chan, 1989 Ref. [22]
CURRENT MCFT IMPLEMENTATION

3
2.5
2
1.5
1
0.5
0
0

10

Shear strain (x10^3 mm/mm)

Figure 4. Average shear stress-strain response of Panel PC5.

a)

b)

c)

Figure 5. a) Axial stress in Y-direction ( yy ), b) Axial stress in X-direction ( xx ), c) Shear

stress ( xy ).