DISTURBED STRESS FIELD MODEL FOR REINFORCED

CONCRETE: IMPLEMENTATION
By F. J. Vecchio1

ABSTRACT: The Disturbed Stress Field Model is a smeared delayed-rotating-crack model, proposed recently
as an alternative to fully fixed or fully rotating crack models, for representing the behavior of cracked reinforced
concrete. It is an extension of the modified compression field theory; advancements relate to the inclusion of
crack shear slip in the element compatibility relations, the decoupling of principal stress and principal strain
directions, and a revised look at compression softening and tension stiffening mechanisms. In this paper, a
procedure is described for implementing the formulations of the Disturbed Stress Field Model into a nonlinear
finite-element algorithm. The procedure is based on a total-load secant-stiffness approach, wherein the crack slip
displacements are treated as offset strains. Computational aspects of the formulation are shown to be simple and
numerically robust. The hybrid crack slip formulation used is found to accurately model the divergence of stress
and strain directions, providing an improved representation of behavior. Predictions of shear strength and failure
mode are significantly influenced in some cases.

INTRODUCTION formulations previously presented for the MCFT, which in

themselves were based on the results of extensive experimental
The Disturbed Stress Field Model (DSFM) was introduced investigations. The MCFT relations were revised somewhat,
in Vecchio (2000) as an alternative formulation for describing however, to reflect the new view of crack slip taken in the
the behavior of cracked reinforced concrete elements. The the- DSFM. Thus, given the net (elastic) principal strains in the
ory is an extension of the modified compression field theory concrete, calculations of the concrete principal stresses can be
(MCFT) (Vecchio and Collins 1986), with advancements made made. Material and element stiffness factors can then be de-
primarily with respect to modeling of shear slip along cracks. fined according to the analysis approach being adopted.
The impetus was to address the diminished accuracy seen from This paper shows how the conceptual model and analytical
existing procedures under certain conditions, particularly for relations of the DSFM can be incorporated into a nonlinear
beams or wall elements containing no shear reinforcement. finite-element algorithm. An example analysis is used to dem-
The new formulation combines aspects of rotating-crack and onstrate the formulation of the model and numerical aspects
fixed-crack models, giving an improved representation of of the solution algorithm. A general discussion is then given
crack mechanisms and thereby resulting in increased accuracy. identifying important behavior models implicit in the theory
In the DSFM formulation presented in Vecchio (2000), ma- and how they differ from alternative formulations. Finally, ex-
terial response is described at a fundamental (element) level ample analyses are presented illustrating application of the
for membranes subjected to general 2D stress conditions. Par- analysis procedure and the significance it has on the computed
ticular attention is given to including crack shear slip in the response.
description of the element’s distortion; new equilibrium, com-
patibility, and constitutive relations are defined accordingly. FINITE-ELEMENT IMPLEMENTATION
Compatibility relations are defined such that, given the ele-
ment’s total (apparent) strain condition [ε], the concrete de- Consider the state of stress and strain at a point in a rein-
formations are resolved into components representing elastic forced concrete continuum, as shown in Fig. 1. Given a stress
strain due to stress [εc ], equivalent average strain due to rigid [␴] acting at the point, the resulting total strain [ε] will be
slip along the crack surfaces [εs ], elastic offset strains [ε 0c ] due such as to satisfy the condition
to mechanisms such as thermal expansion or shrinkage, and
plastic offset strains [ε pc ] due to mechanisms relating to loading [␴] = [D][ε] ⫺ [␴ 0] (2)
history and material damage. Thus, for the concrete compo- where [D] = composite material stiffness matrix. The element
nent, the compatibility relation is pseudo-prestress [␴ 0], due to elastic and plastic strain offsets
in the concrete and reinforcement (including crack shear slip),
[ε] = [εc ] ⫹ [ε s] ⫹ [ε 0c ] ⫹ [ε pc ] (1)
is defined as follows:

冘
From the elastic strains due to stress [ε c ], standard strain trans- n
formations are used to determine concrete principal strains, [␴ 0] = [Dc ]{[ε s ] ⫹ [ε 0c ] ⫹ [ε pc ]} ⫹ [Ds ]i [ε 0s ]i (3)
angle of inclination of the stress field, and crack direction. i=1
Note that the angle of inclination thus determined will differ
from the angle of inclination of the total strains, eliminating For the implementation procedure described here, the ma-
the requirement previously held in the MCFT that the incli- terial stiffness matrix [D] for a reinforced concrete element
nations of principal stress and principal strain coincide. will be constructed in the context of a secant-stiffness for-
Also described in Vecchio (2000) are constitutive relations mulation. Stiffness matrices for the concrete [Dc ]⬘ and each of
for cracked reinforced concrete. These are largely based on the the reinforcement components [Ds ]⬘i are first defined with re-
spect to their principal axes. The total stiffness is then deter-
1
Prof., Dept. of Civ. Engrg., Univ. of Toronto, Toronto, Canada M5S mined by combining the contributions from each of the com-
1A4. ponents, using appropriate transformations to account for
Note. Associate Editor: Julio Ramirez. Discussion open until June 1, anisotropy.
2001. To extend the closing date one month, a written request must be Cracked concrete is treated as an orthotropic material with
filed with the ASCE Manager of Journals. The manuscript for this paper
was submitted for review and possible publication on November 19,
its principal axes (1,2) corresponding to the direction of the
1999. This paper is part of the Journal of Structural Engineering, Vol. average principal stresses (i.e., crack direction). In defining
127, No. 1, January, 2001. 䉷ASCE, ISSN 0733-9445/01/0001-0012– secant stiffness values, net strains [εc ] are used; that is, total
0020/$8.00 ⫹ $.50 per page. Paper No. 22148. strains [ε] less strains due to crack slip [ε s ], elastic offsets
12 / JOURNAL OF STRUCTURAL ENGINEERING / JANUARY 2001
FIG. 1. Reinforced Concrete Element: (a) Element Properties and Applied Stresses; (b) Total Average Strain Conditions in Element;
(c) Average Stresses in Concrete

[ε 0c ], and plastic offsets [ε pc ]. Hence, the concrete material stiff-

ness matrix [Dc ]⬘, evaluated relative to the principal stress di-
rections (1,2), is

[Dc ]⬘ = 冋Ēc1
0
0
E
0
¯ c2
0
0
0
¯c
G
册 (4)

where E¯ c1 , E¯ c2 , and G
¯ c = secant moduli. At a particular stress-
strain state, the secant moduli are evaluated as follows:
¯ ¯
¯ c1 = fc1 ;
E ¯ c2 = fc2 ;
E ¯ s = Ec1 ⭈ Ec2
G (5)
εc1 εc2 ¯Ec1 ⫹ E¯ c2

where εc1 and εc2 = net principal strains in the concrete; and
fc1 and fc2 = corresponding principal stresses [Fig. 2(a)]. Pois- FIG. 2. Definition of Secant Moduli: (a) Concrete; (b) Rein-
son’s effects and lateral expansion, if being considered, are forcement
treated in the manner of elastic offsets [ε 0c ].
For each reinforcement component, a corresponding matrix where ␺ = ␪ for the concrete (i.e., inclination of principal stress);
[Ds ]⬘i is evaluated and ␺ = ␣i for the reinforcement (i.e., orientation of bars).

冋 册
In this formulation, it is necessary to account for the strain
¯s
␳i E 0 0
i offsets in determining prestrain nodal forces. For the crack slip
[Ds ]⬘i = 0 0 0 (6) strains [ε s ], free joint displacements [r sc ] are determined from the
0 0 0 element geometry; that is

冕
where ␳i = reinforcement ratio. The secant modulus Ēsi is de-
fined [r sc ] = [ε s ] dA (12)
fsi
Ēsi = (7) Given the free displacements, the prestrain joint forces [F sc ]
εsi
can be evaluated
where εsi and fsi = average strain and average stress, respec-
[F sc ] = [k c ][r sc ] (13)
tively, in the reinforcement [Fig. 2(b)].
The component material stiffness matrices are transformed where [k c ] = concrete component of the element stiffness ma-
to the global reference system and then summed. The total trix. Prestrain forces are similarly calculated for the elastic and
material stiffness matrix [D] is evaluated plastic offset strains. The prestrain forces are then added to

冘
n
the nodal load vector. A full description of the approach is
given in Vecchio (1992).
[D] = [Dc ] ⫹ [Ds ]i (8)
i=1 A total-load, iterative secant-stiffness routine is then used to
where perform a nonlinear analysis for a reinforced concrete struc-
ture. Through each iteration, the material stiffness [D] and el-
[Dc ] = [Tc ]T [Dc ]⬘[Tc ] (9) ement stiffness [k] matrices are progressively refined until sat-
T
isfactory convergence is achieved. The convergence criteria
[Ds ]i = [T ] [Ds ]⬘[T
s i
i s ]i (10) can be based on either the secant moduli or the element dis-
The transformation matrix [T] is given by placements achieving stable values. A flow chart and descrip-

冋 册
tion of the algorithm was given by Vecchio (1990) for ele-
cos2␺ sin2␺ cos ␺ sin ␺ ments with elastic offsets only; the plastic and shear slip
[T] = sin2␺ cos2␺ ⫺cos ␺ sin ␺ (11) offsets are treated in an analogous manner. A subportion of
⫺2 cos ␺ sin ␺ 2 cos ␺ sin ␺ (cos2␺ ⫺ sin2␺) the algorithm that will be useful in performing a sample cal-
JOURNAL OF STRUCTURAL ENGINEERING / JANUARY 2001 / 13
 EXAMPLE SOLUTION

As an illustration of the calculations involved, consider

Panel PV19 (Vecchio and Collins 1982). The material prop-
erties of the panel are as follows: f ⬘c = 19.0 MPa; f ⬘t = 1.72
MPa, ε0 = ⫺0.00215, a = 10 mm, ␳x = 0.01785, ␳y = 0.00713,
fyx = 458 MPa, fyy = 300 MPa, and Es = 200,000 MPa. An
analysis will be performed for the uniform loading condition
of ␴x = 0 MPa, ␴y = 0 MPa, and ␶ = 3.0 MPa. For the sake
of simplicity, it will be assumed that there are no elastic off-
sets, including no postcracking Poisson effects, and no plastic
offsets. From available crack spacing models (e.g., CEB-FIP
model code), the spacing of cracks in the x- and y-directions
(sx and sy ) are estimated to be approximately 50 mm.
In the first cycle of calculations, using the stiffness factors
and shear slip strains carried over from the previous load stage
(␶ = 2.9 MPa), total strains are determined to be [ε] =
{0.607 1.159 2.302} ⫻ 10⫺3. After 10 iterations of the solu-
tion algorithm, total apparent strains in the element change to
[ε] = {0.597 1.208 2.338} ⫻ 10⫺3. Given in Appendix I are
the calculations that result during the 11th iteration. Note, in
particular, the calculation of the crack slip strains; at this stress
level, the stress-based criterion governs. Strains calculated dur-
ing this iteration are found to be [ε] = {0.596 1.209 2.339}
⫻ 10⫺3. Although the strains appear to have converged, there
remains some change in the computed inclination of the stress
field; hence, additional iterations are required. After 20 itera-
tions, all values have stabilized and the resulting strains are
[ε] = {0.594 1.216 2.342} ⫻ 10⫺3. Note that all calculations
involved are simple in nature and convergence is rapid and
stable.

SHEAR SLIP MODEL

In the DSFM formulation, various alternatives were consid-

ered for modeling crack shear slip. One was to employ an
explicit constitutive model to relate the amount of shear slip
along the crack to the magnitude of the shear stress acting on
the crack. A second alternative was to fix the degree of ‘‘lag’’
between the rotation of the stress field in the concrete and that
of the strain field. For reasons previously discussed, a hybrid
approach was taken combining the two.
For the stress-based approach, the relationship adopted was
that of Walraven (1981), taking the stiffness portion of this
model as follows:

vci
␦s = (14)
1.8w ⫺0.8 ⫹ (0.234w ⫺0.707 ⫺ 0.20) ⭈ fcc

where ␦s = slip displacement along the crack (mm); vci = shear

stress acting on the crack (MPa); w = average crack width
(mm); and fcc = concrete cube strength (MPa). An alternative
formulation is obtained by drawing on the work of Okamura
and Maekawa (1991). Defining vc max as the maximum shear
stress that can be resisted on the crack and a as the aggregate
size in millimeters, the following shear slip relation is ob-
tained:

␦s =
3
4
w 冑 ␺
␺⫺1
(15)

FIG. 3. Nonlinear Analysis Algorithm

where
culation is shown in Fig. 3. In this regard, it should be noted
that, as the stiffness matrices change through each iteration, ␺ = vci /vc max (16)
so too will the prestrain force vectors due to crack slip strains,
elastic offsets, and plastic offsets. Hence, the total force vec- 兹f c⬘
vc max = (MPa) (17)
tors must be recalculated through each iteration. 0.31 ⫹ 24w/(a ⫹ 16)

14 / JOURNAL OF STRUCTURAL ENGINEERING / JANUARY 2001

FIG. 4. Comparison of Concrete Shear-Stress–Shear-Slip
Models: (a) for 25 MPa Concrete; (b) for 60 MPa Concrete

As shown in Fig. 4, the two stress-based formulations provide

essentially similar stiffnesses, although they differ somewhat
in representing the initial slip.
To assess sensitivity to the various slip models and provide
an opportunity for comparison to the MCFT formulation, con-
sider again the response of Panel PV19. Separate analyses
were performed using, alternatively, the rotation lag, crack
shear stress, and hybrid criteria for crack shear slip. Shown in
Fig. 5(a) are the resulting computed shear-stress–shear-strain
responses. There are only minor differences among the three
DSFM simulations, with the rotation lag approach giving very
slightly increased deformations at the early stages and the
stress-based formulation giving a reduced strength and stiff-
ness at the intermediate stages of loading. The hybrid formu- FIG. 5. Comparison of Experimental and Theoretical Re-
lation provides a smooth transition between the two, producing sponses for Panel PV19: (a) Shear Stress-Strain Response; (b)
a result that agrees reasonably well with the experimental val- Inclination of Principal Stress Direction; (c) Inclination of Appar-
ues. The MCFT formulation also gives a good representation ent Principal Strain Direction
of response, although the deformation at the ultimate load
stage is significantly underestimated. More importantly, there parent strain field begins to increase rapidly. The other two
is a subtle but meaningful difference in the predicted failure slip formulations also provide good agreement. The MCFT
mode. The MCFT predicts failure by the concrete stress at- formulation, on the other hand, assumes that the two angles
taining its peak strength, fc2 > fp (i.e., concrete crushing). The of inclination are equal; the values computed lie approximately
DSFM, on the other hand, indicates behavior governed by ex- midway between the inclination of the principal strains and
cessive slip along the crack surface leading to a concrete shear that of the principal stresses, significantly misrepresenting
failure. The latter is more consistent with the observed failure both.
mode.
For this same set of analyses, the predicted angles of incli- SHEAR STRENGTH
nation of the stress and strain fields are shown in Figs. 5(b
and c), respectively. Here, the differences between the various Analyses were performed for a hypothetical panel in which
formulations become more apparent. The hybrid formulation the longitudinal reinforcement was fixed at ␳x = 2.5% and the
for the crack shear slip provides a reasonably good represen- transverse reinforcement ratio was increased from ␳y = 0% to
tation of the response measured in the test panel, particularly ␳y = 3%. The remaining material properties were held constant:
at load stages near ultimate where the inclination of the ap- f ⬘c = 25 MPa, f t⬘ = 2.0 MPa, ε0 = ⫺0.002, a = 15 mm, sx = sy
JOURNAL OF STRUCTURAL ENGINEERING / JANUARY 2001 / 15
 FIG. 7. Comparison of Alternative Models for Compression
Strength Reduction due to Transverse Cracking

Shown in Figs. 6(b and c) are the corresponding orientation

of the stress and strain fields. At low transverse reinforcement
levels, the DSFM predicts angles of inclination for the prin-
cipal tensile strain approximately 8⬚ higher than those pre-
dicted by the MCFT. For the inclination of the principal tensile
stress, normal to the crack direction, the DSFM angle of ori-
entation is about 4⬚ lower. These differentials can be signifi-
cant in situations where the flow of force from the load ap-
plication point to the supports may or may not form a direct
strut (i.e., in beams with short shear spans). As the transverse
reinforcement level increases, differences diminish to zero.
A comment is also in order regarding the substantially re-
duced degree of compression softening included in the new
formulation. Inclusion of crack shear slip in the compatibility
relations of the DSFM results in reduced element stiffness.
Hence, less stiffness degradation need be ascribed to the soft-
ening of the concrete in compression due to the influence of
transverse cracks. In the MCFT, where deformations due to
crack slip are not explicitly considered, the necessary degra-
dation in element stiffness is represented entirely by an in-
creased compression softening factor. In this regard, the
DSFM approach is more rational and more consistent with the
degree of compression softening reported by other researchers
(Fig. 7). However, the two approaches yield substantially sim-
ilar results over a wide range of problems, as demonstrated in
Fig. 6(a) discussed above. Difficulties will arise only if the
two formulations are incorrectly combined [for example, using
the softening equations of the MCFT with the compatibility
relations of the DSFM (or other crack slip models or fixed-
FIG. 6. Comparison of Predicted Responses for Hypothetical crack models) or using reduced softening formulations in fully
Panel: (a) Shear Capacity; (b) Inclination of Principal Stresses at
Ultimate Load; (c) Inclination of Principal Strains at Ultimate rotating crack (i.e., MCFT-like) models].
Load
SAMPLE APPLICATION
= 50 mm, fyx = fyy = 450 MPa, and Es = 200,000 MPa. Shown To illustrate the theoretical model’s application to the anal-
in Fig. 6(a) are the ultimate shear strengths of the panel as ysis of reinforced concrete structures through nonlinear finite-
predicted by the DSFM (hybrid model) and MCFT. At low element procedures, an analysis was made of a concrete box
values of transverse reinforcement ratio, the DSFM predicts structure recently tested at the University of Toronto by Kuz-
shear capacities approximately 10% lower than those obtained manovic (1998). The structural details of the test specimen are
using the MCFT. Note in particular that, for transverse rein- given in Fig. 8; note that the walls of the structure contain no
forcement levels of 0.05% or less, capacity is dictated by the shear reinforcement. The model was loaded at several points
stress at first cracking (i.e., shear strength is constant). At high by actuators acting on the top and bottom walls, simulating
reinforcement ratios and in situations where there is little slip soil pressures acting on the external surfaces. The load was
induced on the crack (e.g., ␳x ⬇ ␳y ), shear strengths predicted monotonically increased until a brittle shear failure occurred
by the DSFM are significantly higher. This is more consistent in the top slab, near the slab-wall joint, at a load equivalent
with observed test results; for example, with panels PV23 and to a uniformly distributed load of 350 kN/m2.
PV25 (Vecchio and Collins 1982) and panels tested by Kol- The finite-element model constructed to represent the test
legger and Mehlhorn (1990). specimen is shown in Fig. 9(a). A total of 788 8-degrees-of-
16 / JOURNAL OF STRUCTURAL ENGINEERING / JANUARY 2001
freedom (DOF) rectangular elements and 4 6-DOF triangular load-deflection responses of the specimen are compared in Fig.
elements were used to present the concrete sections. Longi- 9(b). It is seen that the predicted response is somewhat
tudinal reinforcement was mostly modeled as smeared within stronger, more flexible, and more ductile than that observed.
appropriate narrow bands of elements, although 44 truss bar However, given that this is a large-scale shear-critical structure
elements were used to augment the modeling to properly rep- containing no shear reinforcement, the correlation is of ac-
resent the reinforcement details. Analyses were conducted us- ceptable accuracy; such structural elements are prone to wide
ing a load increment of 10 kN/m2. The nonlinear finite-element scatter in their responses (Vecchio 1999). It should be noted
analysis (NLFEA), utilizing the formulations of the DSFM, that the calculated flexural capacity of the structure is 635 kN/
found the structure to withstand an ultimate load of 370 kN/ m2 and the American Concrete Institute code calculated shear
m2. Failure occurred by shear failure of the top slab, near the capacity is 455 kN/m2.
corner, as in the test specimen. The observed and calculated A more typical application of the analysis procedure would
be one where the structure contains at least minimum rein-
forcement (for example, a shear wall structure). Considered
here is Wall SW16, tested by Lefas et al. (1990), shown in
Fig. 10. The wall web region contained 1.1% horizontal and
2.1% vertical reinforcements. The concealed column elements
at the edges of the wall contained 3.1% vertical reinforcement
and also hoop ties representing an out-of-plane reinforcement
level of 1.2%. The test specimen was subjected to a monoton-

FIG. 8. Details of Box Structure Tested by Kuzmanovic (1998)

FIG. 10. Details of Wall SW16 Tested by Lefas et al. (1990)

FIG. 9. Comparison of Experimental and Calculated Load-De-

flection Response of Kuzmanovic Box Structure FIG. 11. Results of Finite-Element Analysis of Wall SW16

JOURNAL OF STRUCTURAL ENGINEERING / JANUARY 2001 / 17

ically increasing lateral load in combination with a constant procedure, the total apparent strains [ε] and concrete net strains
axial load. Wall SW16 was a well-reinforced and confined [εc ] are found to be
squat wall; its ultimate capacity and deflection response were
shear critical and influenced by confinement effects. [ε] = {0.597 1.208 2.338} ⫻ 10⫺3
The wall specimen was represented by the finite-element [εc ] = {0.823 0.982 2.377} ⫻ 10⫺3
model shown in Fig. 11(a) comprising 340 8-DOF rectangular
elements. Lateral load was applied as a monotonically increas- Calculations will now be performed for the 11th iteration.
ing imposed displacement in 0.2-mm increments. The resulting Note that equation numbers prefixed with I [e.g., Eq. (I-33)]
load-deflection response at the top beam level is shown in Fig. refer to equations given in Vecchio (2000).
11(b). The computed lateral load capacity of 365 kN compares
well with the experimentally determined value of 350 kN. As Step 1: Strain Components
well, the stiffness and ductility of the wall are accurately rep-
resented. The predicted failure mode, involving a shear-crush- 1. Using Eqs. (I-10) and (I-11), determine the average prin-
ing failure in the compression-toe region of the wall, is also cipal strains and their inclination
consistent with the observed behavior. εc1 = 2.094 ⫻ 10⫺3, εc2 = ⫺0.288 ⫻ 10⫺3; ␪␴ = 46.92⬚

CONCLUSIONS 2. For inclination of apparent strains, use Eq. (I-9)

The DSFM, for analysis of reinforced concrete elements, is ␪ε = 52.34⬚

presented in a form adaptable to nonlinear finite-element al- 3. For average strains in reinforcement, use Eq. (I-19)
gorithms. Procedures are described for implementation into a
total-load, secant-stiffness based formulation. Crack shear slip εsx = 0.597 ⫻ 10⫺3; εsy = 1.208 ⫻ 10⫺3
as well as other elastic and plastic offset strains in the concrete
and reinforcement are handled through use of prestrain nodal Step 2: Average Stresses in Concrete
forces. The resulting algorithm is stable and robust, allowing and Reinforcement
calculation of an element’s load-deformation response up to
and beyond ultimate capacity. The conceptual model is simple 1. From Eq. (I-24), find Cd = 1.658. Given that Cs = 0.55,
and transparent, and the calculations involved are basic. use Eq. (I-23) to calculate ␤d = 0.523. Hence, using Eqs.
The sample problem and investigative studies undertaken (I-26) and (I-27), find fp = ⫺9.94 MPa and εp = ⫺1.124
yielded the following observations: ⫻ 10⫺3. Using the Popovic’s formulation [Eqs. (I-28)–
(I-30)] find
• The presented crack slip formulations allow for the di- fc2 = ⫺4.93 MPa
vergence of principal stress and principal strain directions
in the concrete, providing a better representation of typical 2. Average principal tensile stress in concrete—from the
behavior observed in test specimens. tension softening mechanisms, using Eqs. (I-33) and
• The described hybrid crack slip model is a successful (I-34) and assuming Gf = 7.5 N/m, Lref = 750 mm, and
amalgamation of stress-based and lag-based approaches; f t⬘ = 1.72 MPa, find
however, the stress-based slip model taken alone is suf-
ficient in most applications. f ac1 = 0.00 MPa
• In elements containing little or no transverse reinforce- From the tension stiffening mechanisms, using Eqs.
ment, the rotation of the concrete stress field is retarded (I-35)–(I-37), find
by up to 10⬚ relative to that of the strain field, resulting
in reduced computed shear capacities compared to those f bc1 = 1.08 MPa
obtained from a fully rotating crack model where the in-
Hence, using Eq. (I-38), find
clination of the stress and strain fields are assumed to
coincide. fc1 = max( f ac1 , f bc1 ) = 1.08 MPa
• In elements containing high amounts of shear reinforce-
ment, the lower degree of compression softening ascrib- 3. Average stresses in reinforcement—from the average
able to transverse cracking results in considerably in- strains εsx and εsy , using Eq. (I-39), find
creased shear capacities relative to those obtained fsx = 119 MPa; fsy = 242 MPa
according to the fully rotating crack model.
• The degree of compression softening required in the de-
Step 3: Local Stresses at Cracks
layed-rotating-crack model (DSFM) is consistent with that
reported by other researchers. 1. To satisfy equilibrium Eq. (I-7), using an iterative pro-
cedure, find the incremental strain at the crack to be ⌬ε1cr
A thorough corroboration of the accuracy of the proposed for- = 1.111 ⫻ 10⫺3. Hence, using Eq. (I-20), calculate local
mulation is possible (Vecchio, unpublished manuscript, 2000). rebar strains
εscrx = 1.115 ⫻ 10⫺3; εscry = 1.802 ⫻ 10⫺3
APPENDIX I. SAMPLE CALCULATIONS
Consider Panel PV19, with properties as previously de- Using Eq. (I-39), find local rebar stresses
scribed, subjected to a pure shear load of ␶xy = 3.0 MPa. Re- fscrx = 223 MPa; fscrx = 300 MPa
sponse is calculated according to the formations presented, us-
Check the equilibrium

冘
ing the secant-stiffness algorithm summarized in Fig. 3 (at the
element level). For simplicity, it will be assumed that elastic
␳i ( fscri ⫺ fsi )cos2␪ni = (0.01785)(223 ⫺ 119)cos2(46.92⬚)
expansion offsets [ε0c ] and plastic offsets due to loading history
and material flow [ε pc ] are zero. Elastic offsets in the reinforce- ⫹ (0.00713)(300 ⫺ 242)cos2(⫺43.08⬚) = 1.08 MPa = fc1
ment [ε0s ]i are also zero.
At the current load stage, after 10 iterations of the iterative 2. Find the shear stress on the crack using Eq. (I-8)
18 / JOURNAL OF STRUCTURAL ENGINEERING / JANUARY 2001
Step 4: Crack Slip
vci = 0.716 MPa
[D] = 冋 9046 3885
5792
⫺4397
⫺3875
4385
册 (MPa)

1. Calculate the crack spacing, from Eq. (I-21), to be s = Step 7: Offset Strains
35.4 mm. Calculate the crack width, using Eq. (I-22), to
be w = 0.073 mm [ε 0c ] = [ε pc ] = [ε 0s ]1 = [ε 0s ]2 = {0 0 0}
2. From the stress-based approach, using Eqs. (I-40) and s
[ε ] = {⫺0.227 0.227 ⫺0.038} ⫻ 10⫺3
(I-12), calculate the shear slip strain
In the NLFEA procedure, use [ε s ] to determine the prestrain
␥ as = 0.460 ⫻ 10⫺3 joint forces. Alternatively, at the element level, use (3) to de-
termine the element prestress vector
3. From the rotation lag approach, using an initial crack
direction of ␪ic = 45⬚ and the lag limit ␪ᐍ = 5⬚, find from (␴⬚) = [Dc][ε s ] = {⫺0.194 0.256 ⫺0.048} (MPa)
Eqs. (I-41)–(I-43) that
Step 8: New Estimate of Strains
⌬ ␪ε = 7.34⬚; ⌬␪␴ = 2.34⬚
From the NLFEA algorithm, determine the resulting strains
And hence, ␪␴ = 47.34⬚. Now, using Eq. (I-18), find at the conclusion of the 11th iteration
␥ bs = 0.418 ⫻ 10⫺3 [ε]⬘ = {0.596 1.209 2.339} ⫻ 10⫺3

4. At this stage of loading, the stress-based criterion gov- [εc ]⬘ = {0.823 0.982 2.377} ⫻ 10⫺3
erns Alternatively, at the element level, use (2) to determine the
⫺3 new estimate of total strains
␥s = max(␥ , ␥ ) = 0.460 ⫻ 10
a
s
b
s
[ε]⬘ = [D]⫺1{[␴] ⫹ [␴⬚]} = {0.596 1.209 2.339} ⫻ 10⫺3
5. Resolve the slip strain into orthogonal components using
Eqs. (I-13)–(I-15) [εc ] = [ε] ⫺ [ε s ] = {0.823 0.982 2.377} ⫻ 10⫺3

ε sx = ⫺0.227 ⫻ 10⫺3; ε sys = 0.227 ⫻ 10⫺3 where

Step 5: Secant Moduli

␥ sxy = ⫺0.038 ⫻ 10⫺3
冋
0.215
[D]⫺1 = 0.000
0.216
0.000
0.422
0.373
0.216
册
0.373 ⫻ 10⫺3
0.775
(MPa⫺1)

[␴] = {0.0 0.0 3.0} (MPa)

1. Concrete—given
APPENDIX II. REFERENCES
εc1 = 2.094 ⫻ 10⫺3; εc2 = ⫺0.288 ⫻ 10⫺3
Kollegger, J., and Mehlhorn, G. (1990). ‘‘Experimentelle Untersuchungen
fc1 = 1.08 MPa; fc2 = ⫺4.93 MPa zur Bestimmung der Druckfestigkeit des gerissenen Stahlbetons bei
einer Querzugbeanspruchung.’’ Rep. 413, Deutscher Ausschuss für
Find, using (5) Stahlbeton, Berlin (in German).
Kuzmanovic, S. (1998). ‘‘An investigation of the shear design of a re-
¯ c1 = 516 MPa;
E ¯ c2 = 17,096 MPa;
E ¯ c = 501 MPa
G inforced concrete box structure.’’ MASc thesis, University of Toronto,
Toronto.
2. Reinforcement—given Lefas, I. D., Kotsovos, M. D., and Ambraseys, N. N. (1990). ‘‘Behaviour
of reinforced concrete structural walls: Strength, deformation charac-
εsx = 0.597 ⫻ 10⫺3; εsy = 1.208 ⫻ 10⫺3 teristics and failure mechanism.’’ ACI Struct. J., 87(1), 23–31.
Okamura, H., and Maekawa, K. (1991). Nonlinear analysis and consti-
fsx = 119 MPa; fsy = 242 MPa tutive models of reinforced concrete, ISBN 7655-1506-0, University of
Tokyo, Tokyo.
Using (7), find Shirai, S., and Noguchi, H. (1989). ‘‘Compressive deterioration of
cracked concrete,’’ Proc., ASCE Struct. Congress 1989: Design, anal-
¯ sx = 200,000 MPa;
E ¯ sy = 200,000 MPa
E ysis, and testing, ASCE, New York, 1–10.
Vecchio, F. J. (1990). ‘‘Reinforced concrete membrane element formu-
lations.’’ J. Struct. Engrg., ASCE, 116(3), 730–750.
Step 6: Material Stiffness Matrices Vecchio, F. J. (1992). ‘‘Finite element modeling of concrete expansion
and confinement.’’ J. Struct. Engrg., ASCE, 118(9), 2390–2406.
1. Concrete—from (9) and (11), and with ␪ = 46.92⬚ Vecchio, F. J. (1999). ‘‘Analysis of shear-critical reinforced concrete

冋 册
beams.’’ ACI Struct. J., 97(1), 102–110.
5476 3885 ⫺4397 Vecchio, F. J. (2000). ‘‘Disturbed stress field model for reinforced con-
[Dc ] = 4366 ⫺3875 (MPa) crete: Formulation.’’ J. Struct. Engrg., ASCE, 126(9), 1070–1077.
4385 Vecchio, F. J., and Collins, M. P. (1982). ‘‘Response of reinforced con-
crete to in-plane shear and normal stresses.’’ Rep. No. 82-03, Dept. of
2. Reinforcement—from (10) and (11), given that ␳1 = Civ. Engrg., University of Toronto, Toronto.
0.01785, ␳2 = 0.00713, ␣1 = 0⬚, and ␣2 = 90⬚, find Vecchio, F. J., and Collins, M. P. (1986). ‘‘The modified compression

冋 册
field theory for reinforced concrete elements subjected to shear.’’ ACI
3570 0 0 J., 83(2), 219–231.
[Ds ]1 = 0 0 0 (MPa) Walraven, J. C. (1981). ‘‘Fundamental analysis of aggregate interlock.’’
J. Struct. Div., ASCE, 107(11), 2245–2270.
0 0 0

冋 册
0 0 0
APPENDIX III. NOTATION
[Ds ]2 = 0 1426 0 (MPa) The following symbols are used in this paper:
0 0 0
a = aggregate size (mm);
3. Composite [D] = total material stiffness matrix;

JOURNAL OF STRUCTURAL ENGINEERING / JANUARY 2001 / 19

[Dc ] = material stiffness matrix for concrete component; ␤d = strength reduction factor due to transverse cracking;
[Ds ] = material stiffness matrix for reinforcement component; ␥s = shear strain due to slip along crack surface;
Ec = initial modulus of elasticity of concrete; ⌬ε1cr = local incremental strain in 1-direction at crack location;
Ēc1 = secant modulus of concrete in principal 1-direction; ⌬␪ = angle difference between inclinations of principal
Ēc2 = secant modulus of concrete in principal 2-direction; stresses and apparent principal strains;
Es = elastic modulus of reinforcing steel; ␦s = slip displacement along crack surface;
Esh = strain hardening modulus of reinforcing steel; [ε] = apparent (total) average strains in element, including
[F sc ] = equivalent nodal forces representing offsets due to crack crack slip strains;
slip; [ε pc ] = plastic offset strains in concrete due to loading history;
f c⬘ = compressive strength of concrete cylinder (28 days); [ε s ] = equivalent average strains due to discontinuous slip along
f t⬘ = tensile strength of concrete; crack;
fcc = compressive strength of concrete cube; [ε 0c ] = elastic offset strains in concrete due to expansion, ther-
fc 1 = principal tensile stress in concrete (in 1-direction); mal, or prestressing effects;
fc 2 = principal compressive stress in concrete (in 2-direction); ε 0s = initial elastic prestraining reinforcement;
fp = peak compressive stress in cracked concrete (negative [εc ] = average strains in concrete, net of crack slip strains;
value); εcr = cracking strain of concrete;
fs = average stress in reinforcement; εc 1 = strain in concrete in principal tensile stress direction;
εc 2 = strain in concrete in principal compressive stress direc-
fscr = local stress (at crack) in reinforcement;
tion;
fy = yield stress of reinforcement;
εp = compressive strain at peak stress fp in transversely
Ḡc = secant shear modulus of cracked concrete;
cracked concrete (negative value);
Gf = fracture energy of concrete; εs = average strain in reinforcement;
[k] = element stiffness matrix; εscr = local strain in reinforcement at crack location;
[r sc ] = equivalent free nodal displacements due to crack slip off- εsh = strain in reinforcement at start of strain hardening;
sets; εu = ultimate strain for reinforcement;
s = average crack spacing in 1-direction; ε0 = compressive strain at peak stress f ⬘c in concrete cylinder
sx = average crack spacing in reference x-direction; (negative value);
sy = average crack spacing in reference y-direction; ␪ = inclination of normal to crack direction in concrete;
[T ] = rotation transformation matrix; ␪ic = inclination of principal stresses in concrete at first crack-
vc = shear stress on concrete relative to reference x,y-direc- ing;
tions; ␪n = angle between reinforcement component and normal to
vci = shear stress on crack surface; crack;
vc max = maximum shear stress that can be resisted on crack sur- ␪ε = inclination of apparent (total) principal strains in con-
face; crete;
w = average crack width; ␪␴ = inclination of principal stresses in concrete;
␣ = inclination of reinforcement component relative to ref- ␳ = reinforcement ratio; and
erence x-direction; [␴] = average stresses acting on element.

20 / JOURNAL OF STRUCTURAL ENGINEERING / JANUARY 2001