Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

TENSION-STIFFNESS M O D E L FOR REINFORCED

CONCRETE BARS
By Ajaya K. Gupta, 1 Member, ASCE, and Sergio R. Maestrini 2
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

ABSTRACT: An analytical formulation for modeling the tension-stiffness phenom-


enon in cracked reinforced concrete bars is presented. It is assumed that the bond
stress-slip relationship is bilinear, and remains linear throughout the bar in the
initial stages of the cracking. At the later stages, the bond stress becomes constant.
Unlike in the reinforced concrete models currently in use, it is shown that the
tension-stiffness phenomenon is a function of the parameters of the problem, viz.,
the area parameter, the bond parameter, and the strength parameter. A simplified
model is also presented in which the bond parameter is eliminated. Results from
the present model are compared with those from some test results. Various im-
plications related to the assumptions and the conclusions are reviewed. The ex-
perimental results show a marked reduction in the tensile capacity of concrete at
higher deformations that cannot be explained by the proposed relationships. A pos-
sible solution is to use a variable concrete tensile strength that diminishes expo-
nentially as a function of the average tensile strain and a damage parameter. Future
investigations may show that this parameter is related to the property of the ma-
terials, geometry of the structural member, the loading level, and so on. The pres-
ent formulation represents a significant improvement over the models currently in
use.

INTRODUCTION

When a reinforced concrete member is subjected to a sufficiently high


tension force, the concrete cracks. The concrete between cracks continues
to carry tension stresses and offer stiffness. This phenomenon is called ten-
sion-stiffness or tension-stiffening. Modeling of this phenomenon is impor-
tant in studying the load-deformation characteristics of reinforced concrete
structures in the post-concrete-cracking range. In simple models, the con-
crete is allowed to retain a finite tensile capacity after cracking. Three con-
crete tensile stress-strain curves are shown in Fig. 1. Scanlon and Murray
(1974) proposed a stepped stress-strain curve shown in Fig. 1(a). Lin and
Scordelis (1975) used a gradual unloading curve, Fig. 1(b). Gilbert and War-
ner (1978) tried several variations of the Scanlon-Murray and the Lin-Scor-
delis curves. In addition, they employed a new curve consisting of a small
drop in strength immediately after cracking followed by piecewise linear
unloading, which is shown in Fig. 1(c).
The relationship between the bond stress and the crack patterns in rein-
forced concrete bars has been experimentally studied by many researchers
(Lutz and Gergely 1967; Nilson 1971; Goto 1971; Houde and Mirza 1979;
Mirza and Houde 1979; Tepfers 1979; Somayaji 1979; Hwang and Rizkalla
1983; Jiang et al. 1984). Several analytical models have been developed
'Prof, of Civ. Engrg., North Carolina State Univ., Box 7908, Raleigh, NC 27695-
7908.
2
Adjunct Prof, of Arch. Engrg., Federal Univ. of Pelotas, Pelotas, RS 96015,
Brazil; formerly, Grad. Student, Dept. of Civ. Engrg., North Carolina State Univ.,
Box 7908, Raleigh, NC.
Note. Discussion open until August 1, 1990. To extend the closing date one month,
a written request must be filed with the ASCE Manager of Journals. The manuscript
for this paper was submitted for review and possible publication on October 28,
1988. This paper is part of the Journal of Structural Engineering, Vol. 116, No.
3, March, 1990. ©ASCE, ISSN 0733-9445/90/0003-0769/$1.00 + $.15 per page.
Paper No. 24466.
769

J. Struct. Eng. 1990.116:769-790.


<Vf't
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

7.8 e/e
crack
(a)

<Vf't

crack
(b)

V f 't

10 £/£
crack
(c)

FIG. 1. Concrete Tensile Stress-Strain Curves: (a) Stepped Response after


Cracking (Scanlon and Murray 1974); (b) Gradual Unloading after Cracking (Lin
and Scordelis 1975); and (c) Discontinuous Unloading after Cracking (Gilbert and
Warner 1978)

(Edwards and Picard 1972; Tepfers 1980; Somayaji and Shah 1981; Tassios
and Yannopoulos 1981; Floegl and Mang 1982; Bazant and Oh 1984; Riz-
kalla and Hwang 1984; Yankelevsky 1985; Hegemier et al. 1985; Murakami
and Hegemier 1986; Hageman et al. 1986). Somayaji and Shah (1981) pro-
posed an exponential bond stress distribution, which in turn is related to the
slip at the steel-concrete interface. Floegl and Mang (1982) proposed an
approximate relationship between the average bond stress, the average ten-
sion strain in the reinforcing bars, the ultimate compressive strength of con-
crete, and the hydrostatic compressive stress in concrete. They used this
relationship in a nonlinear finite element analysis program. Bazant and Oh
(1984), considering the tension-stiffening effect of concrete as an increase
in the stiffness of the reinforcing steel, calculated the extensional stiffness

770

J. Struct. Eng. 1990.116:769-790.


of a cracked panel in terms of an equivalent steel ratio. Hegemier and co-
workers (Hegemier et al. 1985; Murakami and Hegemier 1986; Hageman et
al. 1986) used the bond stress-slip relationship as a constitutive property,
and formulated the load-deformation and the related parametric relationships.
In the present study, we have also assumed that the bond stress-slip re-
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

lationship can be used as a constitutive property. Three idealized forms of


this relationship have been used, viz., linear, bilinear, and constant bond
stress. The equilibrium and compatibility equations in conjunction with the
constitutive properties give the differential equation for the problem. Closed-
form solutions of these equations give the distribution of various stresses,
strains and displacements. The closed-form equations have also been used
to obtain effective tension stress-strain relationships for concrete similar to
those shown in Fig. 1. It is shown that such relationships are functions of
the parameters of the problem. Simplified force-deformation and effective
concrete tensile stress-strain relationships are proposed that eliminate one of
the parameters, viz., the bond parameter, whose value is relatively uncer-
tain.
The proposed model is used to evaluate the steel force distribution in the
bars tested by Houde and Mirza (Houde and Mirza 1972; Mirza and Houde
1979), and to obtain the force-deformation and the effective concrete tensile
stress-strain relationships for the specimens tested by Hwang and Rizkalla
(1983). Implications of the comparisons of the test and theory results, of
various assumptions and of various conclusions are discussed.

BOND STRESS-SLIP RELATIONSHIPS

A typical bond stress-slip curve obtained experimentally by Nilson (1971)


is shown by the solid line in Fig. 2. The experimental curve can be ap-
proximated by a bilinear curve, shown in Fig. 2 by the dashed lines. The
use of the idealized curve is a major assumption of the present study. Given
the bilinear shape of the curve, the slope of the inclined line and the max-
imum bond stress associated with the horizontal line both vary with the ma-
terial, geometry, and the loading parameters. The working assumption here

w
U)
a)
l_
•4—*

CO
o •m
co
m <u&
T>
^
-6
aN> s--
-h
10
F
L_

o
:z
0 1.0 2.0 3.0
3
Slip, A*10 in

FIG. 2. Bond Stress-Slip Curve (Nilson 1971)


771

J. Struct. Eng. 1990.116:769-790.


is that the bond stress-slip relationship for a reinforced concrete bar can be
expressed by the same bilinear curve throughout the length of the bar at a
particular stage of loading. At the moment, it is not necessary to assume
that the bilinear curve does not change with the axial loading applied to the
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

bar.
According to the idealized curve, the bond stress varies linearly with the
slip in the beginning of the cracking process. The behavior of the reinforced
concrete bar at this stage can be obtained by using a linear bond stress-slip
relationship. On the other hand, when extensive cracking has occurred we
may assume that the bond stress is uniform along the bar. The intermediate
cases are represented by the bilinear curve.

EQUATIONS FOR STEEL AND CONCRETE COMPONENTS

Fig. 3 shows part of the reinforced concrete bar between two cracks. The
length of this part is 2a, which is the same as the crack spacing. The origin
of the axial coordinate x is taken midway between the two cracks, at point
0 shown in the figure. The bar is subjected to a tension force T. Let us take
the section at distance x from the origin. The free-body diagram of the right
side is shown in Fig. 3(b). We have the following equilibrium equation for
this free body:

~T,x

(a)

F 0 ^ fb
F —-^ ^ ^•£% - T F s —W— Fs + dFc
Fo — X
—a-x- dx
(b) (c)

Fc— —*F0 + dFc


"—*t
fb
Fc — Fc + dFc

dx

(d)

FIG. 3. Free Body Diagrams for Cracked Reinforced Concrete Bar: (a) Bar be-
tween Cracks; (b) Bar Showing Arbitrary Section; (c) Steel Interface; and (d) Con-
crete Interface

772

J. Struct. Eng. 1990.116:769-790.


FS + FC = T (1)
in which Fs and Fc = the axial forces in the steel and concrete components,
respectively.
An infinitesimal element of length dx is taken out from the bar. The free-
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

body diagrams of the steel and the concrete components are shown in Figs.
3(c) and (d). We can obtain the following equilibrium equations for the two
components:
dF„ dFc
T=pf», -r=-pf" (2)
dx dx
where fb = the bond stress at the steel-concrete interface; and p represents
the perimeter (circumference) of the steel bar. The force-strain relationships
are
dus duc
Fs = AsEses = ASES — , Fc = AcEcec = ACEC — (3)
dx dx
in which the subscripts s and c refer to the steel and concrete components;
A = the area of cross section; E = Young's modulus; e = the axial strain;
and u represents the axial displacement.
The slip at the steel-concrete interface is defined by
A = us - uc (4)
Eqs. 2-4 give
d2& pfb Es As
—r = (1 + no), n = —, p=— (5)
dx2 ASES V
" E; Ac
The terms n and p in Eq. 5 are commonly called the modular ratio and the
steel (reinforcement) ratio, respectively. We shall call the nondimensional
product no the "area parameter."
In deriving Eq. 5, both steel and concrete have been assumed to be linear
elastic. Further, it is implied that the force Fc is uniformly distributed on
the concrete cross section, and that ec and uc do not vary over the cross
section. Eq. 4 is general with respect to the bond stress-slip relationship. It
will be applied to three types of relationships in the subsequent sections.

SOLUTION FOR LINEAR BOND STRESS-SLIP RELATIONSHIP CASE

The linear bond stress-slip relationship can be written as


h = AA (6)
where A = the slope of the bond stress-slip curve, and can be called the
slip modulus. The units of fb and A are FL'2 and L, respectively. Therefore,
the unit of A is FL~3, e.g., pci (pounds per cubic in.); Nm"3 (Newtons per
cubic meter). Eq. 5 and 6 give
d2k , , Ap
— - ^A = 0, ** = -f- (1 + np) (7)
dx A.E.
773

J. Struct. Eng. 1990.116:769-790.


The unit of k'1 in Eq. 7 is L. We shall call k~ the "characteristic length."
The solution of Eq. 7 is given by A = Cx sinh kx + C2 cosh kx, in which
Cj and C2 are constants of integration. Considering the symmetry of the
problem, A(—x) = — A(x). Hence, C2 = 0, and we get
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

A = d sinh kx •. (8)
The constant C, will be determined later.
Eqs. 2, 3, 6, and 8 give
ApCi
Fs = T (cosh ka — cosh kx),
k
Tx Cx
us = (kx cosh ka — sinh kx) (9)
ASES 1 + zip
and
ApCi npCi
Fc = (cosh ka - cosh kx), uc = (kx cosh ka - sinh kx) ... (10)
k 1 + np
In the derivations of Eqs. 9 and 10, the following boundary conditions have
been used: Fs = T and Fc = 0 at x = a; us and uc = 0 at x = 0. Eqs. 4,
8, 9, and 10 yield
r 1
C, = (11)
AjEsfccosh fa?

Concrete Tensile Contribution


To obtain a concrete stress-strain relationship of the type shown in Fig.
1, we state the following:
T = A„Ese + Ac<rc, ac > 0 (12)
in which the term ac = the effective tensile stress in concrete. The average
strain in the reinforcing steel bar is given by e and can be obtained from
Eqs. 9 and 11
us(x = a) T ( tanhfca\
e= = np + (13)
a AsEs(l + «P) V ka J
Eqs. 12 and 13 give
T t tanh ka\
ac = 1 (14)
Ac(l + np) V ka )
The maximum tensile force in concrete is at x = 0. This maximum force
and the corresponding concrete stress can be obtained from Eqs. 10 and 11
T
^«nax = -— (1 - sech ka),
(1 + np)
F T
oVmax = —t— = —— (1 - sech ka) (15)
Ac Ac(l + np)
The maximum tensile stress in concrete o-cmax is directly proportional to the
774

J. Struct. Eng. 1990.116:769-790.


Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

;_~ 0.8 •
o
b

crack
FIG. 4. Effective Concrete Tensile Stress-Strain Curves, Linear Bond Stress-Slip
Relationship

applied tension T. (The term o-cmax in Eq. 15 denotes the maximum of the
actual concrete stress. The termCTC,on the other hand, is defined in Eq. 12,
and is used to represent the effective concrete stress.) As the applied tension
T increases, so does crcmax. At some value of T, crcmax will become equal to
the tensile strength of concrete/,'. At that point, a new crack will be formed
at x = 0. Let us consider the state just before the new crack is formed and
^cniax —* ft- For this state, Eq. 15 gives the maximum value of T that can
be applied with the given crack spacing la
/;
T = Ac(l + wp) (16)
(1 — sech ka)
We can now eliminate T from Eqs. 13 and 14 and rewrite them in a non-
dimensional form
tanh ka tanh ka
i - 1 +
ka 6 npka
(17)
/; 1 — sech ka ^crack 1 — sech ka
The actual crack spacing, 2a, will vary discretely. The initial value 2a can
be taken equal to the length L of the reinforced concrete bar under consid-
eration, 2a = L. Subsequent values of 2a will be L/2, L/A, L/8, ..., etc.
The value of L is different for different bars. To keep the formulation gen-
eral, we may assume that a varies continuously from °° to 0 without intro-
ducing any significant error. (As we will see later, the lower limit of a will
be limited by the value of the maximum bond stress, fm, and the yield stress
of the reinforcing steel,/,). Thus, Eq. 17 represents a parametric relationship
between o-c//,' and e/ecrack, in which ka is the common parameter. This ob-
servation is particularly important because we need not know the actual value
of k and of the parameters that define k. On the other hand, we observe
from Eq. 17 that the (rc/f',-e/ecrack relationship is a function of the area par-

775

J. Struct. Eng. 1990.116:769-790.


emeter (np). This particular observation is a departure from the curves shown
in Fig. 1.
The crc/f}-e/ecmck relationship obtained from Eq. 17 is plotted in Fig. 4
for four values of the area parameter (np = 0.02, 0.04, 0.1, and 0.5). For
ka = oo, Eq. 17 gives <Jc/f't - 1, e/e crart = 1, defining the stage at which
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

the cracks are just beginning to form. If there is no limit on the bond stress
and the reinforcing steel stress values, we can see that for ka = 0, CTC//,' =
0.67 and e/ecrack = oo. As such, the 0.67 value is a lower bound for
°"c//<- We observe from Fig. 4 that the effect of tension-stiffness is most
predominant when np is relatively small. This effect diminishes as np in-
creases. At higher values of np the stiffness of the reinforcing bars dominates
the stiffness of the reinforced concrete bar. Soon after the first crack is found
(e/tcrac* = 1), the <rc/f', value drops from 1 to 0.67, the lower bound value.

SOLUTION FOR CONSTANT BOND STRESS CASE

In this case,/;, = fm, independent of the slip. Eq. 2 with appropriate bound-
ary conditions give
Fs = T - pfja - x), Fc = pfm(a - x) (18)
The first of Eq. 3 yields

T
1 Pfma
~ Pfm\ a - T- (19)
ASES
Eqs. 12 and 19 give
PfmCl
crc = (20)
2AC
Calculating o-cmax from Eq. 18 and setting it equal to/,', we get
, _Pfma
Jt — (21)

For given values of Ac, p, /,', and/ m , Eq. 21 gives a constant value of a.
Eqs. 19-21 can be written in the nondimensionalized form as follows:
_ J_/_27^_ \
(22)
/; ^•crack 2np \pfma )
Eq. 22 shows that CTC//,' remains constant at 0.5, while t/ecrack varies linearly
with the increasing value of T. Eq. 22 is valid only after the cracks have
been formed. As we will see later, CTC//,' = 0.5 constitutes the lower bound
value for the case with the bilinear bond stress-slip relationship.

SOLUTION FOR BILINEAR BOND STRESS-SLIP RELATIONSHIP CASE

We shall now consider the bilinear bond stress relationship


/„ = AA, /t</m, A<A m ; fb=fm, A>Am (23)
At the initial stages of cracking, the slip would remain less than Ara through-
776

J. Struct. Eng. 1990.116:769-790.


out the bar. The behavior of the bar is described by the first case, the linear
bound stress-slip relationship case. As the applied tension T is increased, A
first reaches Am at x = a, and then at x = xm, 0 ^ x„, :£ a. Applying the
condition A = Am = fjA at x = xm in Eq. 8 we get
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

Jm fm smhfcx
C, = A= , \J — *v —— A'm (24)
A sinh kxm A sinh kxm
For xm ^ x < a, we can substitute/4 = / m in Eq. 2. Integration of these
equations and substitution of the appropriate boundary conditions yields
FS = T - pfm(a - x), Fc = pfjfl - x), (25)
For 0 < x =£ xm, fb = AA, where A is given by Eq. 24. Substituting these
in Eq. 2, integrating the equations and satisfying the equilibrium &tx = xm
(Eq. 25), we get
cosh kx — cosh kx„
Fs = T ( ka — kx, (26a)
k sinh kxm
coshfcc— cosh kx„
Fc = —^ ( ka - kxm (26&)
sinh fccm
Eqs. 3 and 26 with appropriate boundary conditions give the correspond-
ing displacements
Tx kx coshfccm— sinh fcc
US kx(ka fofm) + ;- .. (27a)
~AJS~
sinh fccm
Pfm kx coshfccm— sinh kx
kx(ka — kxm) , U — X — Xm (21b)
uc = — sinh fccm
#ACEC
Tx Pfm KXm ~- kx(2ka — kx) kxm cosh kxm — sinh fccm
U, = h (21c)
ASES fAsE, 2 sinh kxm
2
Pfm kx(2ka — kx)- k ^ kxm cosh kxm — sinh fccm
fAcEc 2 sinh kxm
x
m < x
*~ " • .. (21 d)
Eqs. 4, 24, and 27 give

T = — (1 + np)[(ka - kxm) + coth kxm] (28)


k
As before the average strain in steel reinforcement is obtained by dividing
us at x = a by the semi-crack spacing, a. From Eqs. 21a-d we have
Pfm *V-4) + kxm cothfcc„- 1 (29)
ASES kf'aAsE,
Eqs. 12 and 29 give
Pf (30)
(kxm cothfccm— 1) +
tfaA, 2

777

J. Struct. Eng. 1990.116:769-790.


The maximum stress in concrete is calculated from Eqs. 26. Setting the max-
imum concrete stress equal to /,' we get
Pfm /cosh kx
2m - 1•+ ka-jaj \
f< = » » (31)
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

kAc \ sinh kxm /


Eqs. 28-31 give the nondimensionalized parametric equations for ac/f't and
^ / ^-crack

CTC 1 2kxm + (&V - k2^ - 2) tanh kxm


—= — (32a)
/,' 2ka 1 — sech kxm + (ka — kxm) tanh kxm
<• [2ka(ka - fct„)(l + np) - (jfcV - <*£ - 2)] tanh kxm + 2[Aa(l + np) - fccj
(326)

"«t* 2kanp[l - sech focm + (fca - fc*„) tanh kx„]

As was done in the case of the linear bond stress-slip relationship, Eqs.
32 can be used to obtain <rc/f, = e/ecrack relationship by continuously varying
the value of ka from °° to 0. In this case, however, we have another unknown
on the right-hand sides of the equations, kxm. To evaluate kxm, we propose
to use the following modified form of Eq. 31
cosh kxm - 1 kAcf,
kxm = ka + — ~ /, /= - ^ (33)
sinh kxm pfm
in which the term / is a parameter of the problem. It is the ratio of the
concrete cracking force (Acf}) and the bond force (A:_1/?/m) transferred at the
steel-concrete interface over the characteristic length (k"1). We shall c a l l /
the "bond parameter." Maestrini and Gupta (1987) originally called / the
tension-stiffness parameter. The new name appears to describe it better.
Fig. 5 graphically shows the relationship between kxm and ka - fin ac-
cordance with Eq. 33. When kxm is large, (cosh kxm - l)/sinh kxm ~ 1, and
Eq. 33 becomes
kxm = ka - / + 1 (34)
On the other hand, when kxm is small, (cosh kxm - l)/sinh kxm ~ kxm/2,
for which Eq. 33 gives
kxm = 2(ka-f) (35)
It is clear from Fig. 5 that the curve represented by Eq. 33 is asymptotic to
the straight lines given by Eqs. 34 and 35 at the higher and the lower ends,
respectively.
In solving for xm from Eq. 33, we need to apply a practical constraint on
the equation, viz., xm £ a. Two xm = a lines are drawn in Fig. 5, one each
for / = 0 and 1, respectively. The admissible values of xm lie below the xm
= a line for the appropriate/-value. It is clear that f o r / = 0, xm = a governs
for the entire range of the curve. A zero value of the bond parameter (/)
implies that/ m is very large, and that the linear bond stress-slip relationship
applies. F o r / = 1, Eq. 33 holds good for all values of a, as it also would
when/ > 1. For values of the tension-stiffness parameter in the range, 0 <
/ < 1, the xm = a line (not shown) would intersect the curve of Eq. 33 at
some intermediate point. Above the intersection point, xm = a would apply,
and below it Eq. 33 would. The substitution of xm = a in Eqs. 32 gives Eq.

778

J. Struct. Eng. 1990.116:769-790.


Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

ka-f -
FIG. 5. Variation of xm with a

17, which are for the linear bond stress-slip relationship.


The lowest possible value of xm is zero, which is associated with the min-
imum crack spacing. Eqs. 33 and 35 give
/ Kf't
flmin = 7 = ~ — (36)
* Pfm
The value of arain given by Eq. 36 is identical to the value of a given by
Eq. 21, which was derived for the constant bond stress case. This is further
confirmed by substituting xm = 0 in Eq. 32, which yields Eq. 22, derived
earlier for the constant bond stress case. When a < a^, the value of xm
given by Eq. 33 or 35 would be negative, which is impossible. In such cases,
we should assume x,„ = 0. This means that a constant bond stress would be
achieved in the bar at an appropriate value of T without any further cracking.
Curves showing the <Jc/f, ~ e/ecract relationship based on Eqs. 32 and 33
were obtained for several sets of area and bond parameters (np,f) in Maes-
trini and Gupta (1987). Of those, curves for np = 0.02 and 0.5 with/vary-
ing between 0 and 5.0 are shown in Fig. 6. As discussed, when/ = 0, the
linear bond stress-slip relationship applies. This case constitutes an upper
bound for the (rc/f, values. For the constant bond stress case, o-c/,' = 0 . 5 .
The curves for the higher bond parameter ( / = 5) in Fig. 6 approach this
lower bound value (CTC//,' = 0.5) soon after the cracking initiates. Curves
for all values of the bond parameter, except those for/ = 0, should approach
the lower bound line asymptotically at sufficiently high strains.
Fig. 7 shows the comparison of the distribution of the force in the rein-
forcing bar obtained from Eqs. 25 and 26 and those obtained experimentally
779

J. Struct. Eng. 1990.116:769-790.


np = 0.020
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 6. Effective Concrete Tensile Stress-Strain Curves, Bilinear Bond Stress-


Slip Relationship

by Houde and Mirza (1972) and Mirza and Houde (1979) for three of a series
of tests they performed. The details of the calculation procedure are given
in a report by Maestrini and Gupta (1987). The agreement between the ex-
perimental curve and the numerically calculated values is reasonably good.
It should be pointed out that three of the points on the curve were used as
input to the numerical scheme. The force at the crack or at the end of the
bar (x = a) is known. Further, the experimentally measured forces in the
steel bar at x = 0 and x = a/2 were used to calculate the values of fm and
xm. The present comparison shows that the shape of the steel force distri-
bution curve given by the present model is realistic. This is a necessary
condition for any model, but it is not sufficient.

LOAD-DEFORMATION RELATIONSHIP

It has been shown that the effective concrete tensile stress-strain relation-
ship in the post-cracking range is a function of the two parameters of the
780

J. Struct. Eng. 1990.116:769-790.


o e Experimental points. (Houde and Mirza)
Present study.
Test No. 18
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

3 6 9 12 15 18 21 24 27 30

Load Level, kips

FIG. 7. Distribution of Force in Reinforcing Bar

problem identified so far, viz., the area parameter (np) and the bond param-
eter (/). Whereas the area parameter is accurately known, the same is not
true of the bond parameter. The slip modulus (A) and the maximum bond
stress are idealized constitutive properties, and are highly variable. It is therefore
proposed to study the effect of the bond parameter further considering the
load-deformation relationship.
If the effect of tension-stiffness is ignored, the maximum value of T will
781

J. Struct. Eng. 1990.116:769-790.


be AJy, which will be attained at e = ty, where fy and ey are the steel yield
stress and the corresponding strain, respectively. T/AJy and e/e^, are a set
of natural nondimensionalized load and deformation parameters: Both of these
become unity at the yielding of steel when concrete is not present. When
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

concrete is present, and tension-stiffness is included in the model, T/AJy


reaches unity at an e/e.y value less than unity. However, since the steel at
the crack cannot resist a stress greater than/,,, the maximum limiting value
of the load parameter remains unity. Eqs. 28 and 32 can be rewritten as
T _ nf, 1 + np (ka - kxm) tanh kxm + 1 __ 1 ' (36a)
,,* x
£
A
Jy fy «P (ka - kxm) tanh kxm + 1 - sech kxm
e nf, [2ka(ka - fct„)(l + np) + 1^4, - fcV + 2] tanhfcc„+ 2[ka(l + np) - kxm] / n „ s
- = —-—-— (36o)
% /, 2famp[l - sech kxm + (ka - kx„) tanh kxj
These equations identify another nondimensional parameter, fy/nf,, We shall
call it the "strength parameter."

FIG. 8. Load-Deformation Curves

782

J. Struct. Eng. 1990.116:769-790.


Practical values of np approximately range between 0.02 to 0.5, and that
of fy/nf, between 5 and 25. It can be shown that in order for steel yielding
to occur after concrete cracking, np ^ {fy/nf, — l ) - 1 ; otherwise, the crack-
ing of concrete is followed by an unstable failure. From the viewpoint of
studying tension-stiffness effect, we need not consider the pairs of np and
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

fy/nf', associated with the cracking-unstable failure. In a detailed study of


the phenomenon (Maestrini and Gupta 1987), therefore, only the following
np values were considered: 0.1, 0.2, 0.3, 0.4, and 0.5. The fy/nf, values
were: 5, 10, 15, 20, and 25. This constitutes 25 pairs of np, fy/nf, values.
Of these the pairs (0.1, 5), (0.1, 10), and (0.2, 5) are the "unstable" type,
and were discarded. The curves for the remaining 22 pairs are given in
Maestrini and Gupta (1987). For illustration, we shall consider here the fol-
lowing four pairs only: (np,fy/nf,) = (0.2, 10), (0.2, 25), (0.5, 10), and
(0.5, 25).
The load-deformation curves for the four pairs of np, fy/nf, values are
shown in Fig. 8, one plot for each pair. In each plot, the dotted line rep-
resents the lower bound given by the constant bond stress assumption (xm =
0). The upper bound is given by the top continuous line for which the bond
parameter (/) is zero and xm = a. Each plot includes two more curves, one
each f o r / = 2 and 5.
Each curve in Fig. 8 starts at the origin (0), goes linearly to the cracking
point A, and continues almost horizontally to nearby point B. Four curves
diverge beyond this point B. All the curves join the line T/Asfy = 1 at dif-
ferent values of e/ey < 1.
It is observed from Fig. 8 that for the same value of the area parameter
(np), the distance between the adjacent curves diminishes as the strength
parameter {fy/nf,) increases. Similarly, for the same value of fy/nf,, the
adjacent curves become closer for the higher values of np. The increase in
either the area or the strength parameter implies a higher contribution to
stiffness and strength by the steel reinforcement, and a lower contribution
by the concrete component. This explains the closeness in the load-defor-
mation curves for the higher np and fy/nf, values.

SIMPLIFIED RELATIONSHIPS

As we stated earlier, the actual value of the bond parameter (/) is likely
to be highly variable. Functional relationships expressing those variations
are not known. Therefore, it is desirable to develop approximate load-de-
formation relationships that do not include the bond parameter. The proposed
load-deformation curves between the cracking of concrete and the yielding
of steel are shown in Fig. 8 by the dashed lines. The reasoning behind the
proposed relationship is given next.
The dashed line is obtained by joining points B and C. We recall that the
curves for all / values remain identical after cracking at point A until point
B, Beyond point B curves for the lower/values are higher than those for
higher/values. The lower/values are accompanied by higher tension-stiff-
ness contribution. Point C is obtained by intersecting the lowest curve ( / ~
°°, constant bond stress) with the T/AJy = 1 line. It is theorized that at the
initial stages of cracking, the maximum bond stress (fm) is relatively large,
/ relatively small. Therefore, it is appropriate to choose the upper bound
point (B) for the initial stage. At the later stages of cracking, the tension-
783

J. Struct. Eng. 1990.116:769-790.


stiffness effect is likely to diminish. Therefore, we take the lower bound
point (C) at the yield stress end.
Let us denote the abscissa, e/ey by X, and the ordinate T/Asfy by Y. The
coordinates of point A are determined by the cracking strain of concrete
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

nf, nf, 1 + np
XA = YA = (37)
fy ' " fy np
Point B is defined by observation (Maestrini and Gupta 1987)
1 + np nf, 1 + «P
XB = ^ I 1 + Yb = (38)
/y \ lOnp / fy np
Point C is determined by substituting T/Asfy = 1 in Eq. 36, which gives

YA,Y R -

FIG. 9. Simplified Load-Deformation Curve

FIG. 10. Simplified Effective Concrete Tensile Stress-Strain Curve


784

J. Struct. Eng. 1990.116:769-790.


1 nf,
Xc = 1 —, Yc= 1 (39)
2np /,
The simplified load-deformation curve is shown schematically in Fig. 9.
The corresponding effective concrete tensile-stress-strain curve is shown
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

in Fig. 10, in which the abscissa (e/ey) is denoted by X and the ordinate
(a c //,') is denoted by Y
_ nf nf, 1 + np 1 + np
XA — , YA — l; xB = 1 + YB=1~
Jy lOnp io ;

Xc 1 -
J_nf Yc = - (40)
2«P /, '
We recall the issue of change in the bond stress-slip relationship with the
loading that was raised earlier. Elimination of the bond parameter in the
simplified curves allows us to avoid that issue.

APPLICATION TO EXPERIMENTAL RESULTS

As a part of an experimental program held at the University of Manitoba,


Canada, seven concrete specimens were tested by Hwang and Rizkalla (1983)
under uniaxial tension. Adequate data were available in their report to apply
the proposed model to their experiments.
The specimen dimensions were 30 in. X 12 in. X 7 in. (76.2 cm X 30.5
cm x 17.8 cm). The main longitudinal reinforcement bars were spaced 3
in. (7.6 cm) center to center and extended 11 in. (27.9 cm) beyond each
concrete end. Only two quantities were varied in the specimens, viz., the
spacing between the transverse bars and the concrete cover. Specimens 2
and 7 did not contain any transverse reinforcement. Relevant data about the
specimens are given in Table 1.
The experimental results showed a marked reduction in the tensile capacity
of concrete at higher deformations, which could not be explained analyti-
cally. It was as if concrete were progressively damaged as deformation is

TABLE 1. Data for the Hwang-Rizkalla (1983) Tests


Specimen Concrete Transverse bar
number cover (in.) / , (1*1) Ec (ksi) spacing (in.) / ; (ksi)
(1) (2) (3) (4) (5) (6)
1 0.75 65.22 3,189.8 8.5 0.4552
2 0.75 65.70 4,380.2 — 0.4236
3 1.50 69.72 3,594.8 2.0 0.4123
4 1.50 66.26 2,128.3 4.0 0.3894
5 1.50 70.20 5,221.5 6.0 0.3937
6 1.50 66.02 3,823.1 10.5 0.3792
7 1.50 68.03 4,031.2 — 0.3893
Note: Reinforcing bar size 10M; p = 0.01476; E, 29,000 ksi; 1 in. = 2.54 cm; 1
ksi = 6.895 MPa.

785

J. Struct. Eng. 1990.116:769-790.


Present Study
1 CD
Experimenial Data
Steel Alone
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

I 40

7 CD

ky /
|P /

0 1 2 3 0 1 2 3
Average Strain x 103 Average Strain x 103

FIG. 11. Comparison of Simplified Load-Deformation Curve with Hwang-Rizkalla


(1983) Test Results

increased. To account for the phenomenon, a variable tensile strength /, is


introduced
/;
/ , = /,' exp [ - C ( e - e C M C * ) ] , (41)
^•crack

in which C is a "damage parameter." A constant value of C equal to 550


786

J. Struct. Eng. 1990.116:769-790.


Proposed Model
Experimental
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

FIG. 12. Comparison of the Simplified Effective Concrete Tensile Stress-Strain


Curve with the Hwang-Hizkalla (1983) Test Results

was used for all the specimens tested by Hwang and Rizkalla (1983). This
topic will be reviewed further in the next section.
To apply this variable tensile strength concept to the load-deformation and
the effective stress-strain curves, we do the following. Initially, it is assumed
that the/, value remains constant, and the equations of the "straight lines"
OA, AB, BC, and CD are developed using Eqs. 37-41. When/, is a function
of e, these equations represent the nonlinear load-deformation and effective
stress-strain curves. Figs. 11 and 12 show the computed load-deformation
and the effective stress-strain curves, respectively, along with the corre-
787

J. Struct. Eng. 1990.116:769-790.


sponding experimental points, for the seven tests performed by Hwang and
Rizkalla (1983).

SUMMARY AND CONCLUSIONS


Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

The objective of the present study was to develop a model to account for
the tension-stiffness effect in reinforced concrete structures. The structure
considered here is a reinforced concrete bar under uniaxial tension. Some
of the methodologies that have been used previously are referenced in the
introduction. Three common effective concrete stress-strain curves are shown
in Fig. 1. These curves and similar other curves are currently used in various
finite element analysis computer programs. The curves in Fig. 1 are inde-
pendent of the parameters of the problem. Often the same curves are used
in two- and three-dimensional problems. In the present paper, we have in-
vestigated the parametric functionality of the tension-stiffness effect in uni-
axially loaded reinforced concrete bars only. The two-dimensional plane stress
problem is studied as an extension of the present study by Maestrini and
Gupta (1987) and Gupta and Maestrini (1989a, 1989b).
It is first shown that for the assumed model the tension-stiffness effect is
a function of the area parameter (np) and the bond parameter ( / =
Af',/k~ipfm). Stress distribution in the steel bar obtained from the present
model is in close agreement with that obtained experimentally by Houde and
Mirza (1972) and Mirza and Houde (1979). Since the inputs to the model
were determined by matching three points on the experimental curve, this
comparison shows that the model gives a realistic shape of the stress-distri-
bution curves, a necessary, but not sufficient requirement for any model.
Two of the constitutive properties, the slip modulus (A) and the maximum
bond stress (/„,), that define the bond parameter (/) are based on an idealized
bond stress-slip relationship and are highly variable. Therefore, the effect of
the variation in the value of / on the reinforced concrete bar force-defor-
mation relationship was studied.
In view of the fact that the tensile strength of a reinforced concrete bar is
limited by the yield capacity of the steel bar, Asfy, the latter was used to
nondimensionalize the applied tensile force. Similarly, the average strain was
nondimensionalized using the steel yield strain, e r In the absence of con-
crete, the yield point is defined by (T/Asfy,e/ey) = (1,1). The force-defor-
mation relationship in conjunction with the new nondimensionalizing quan-
tities identified another nondimensionalized parameter, the strength parameter

The force-deformation and the effective concrete tensile stress-strain curves


were drawn for several pairs of the area and strength parameters with a vari-
able bond parameter (/). The upper bound curves are obtained when/ = 0,
which corresonds to the linear bond stress-slip relationship. The constant
bond stress case gives / = °° and gives the lower bound curves. All the
curves for different/-values coincide at the early stages of the post-cracking
behavior, indicating a linear bond stress-slip relationship. Then the curves
diverge, become roughly parallel, subsequently reaching the yield level, T/
AJy = 1, at different locations. At the latter stage it is proposed to use the
lower bound point ( / = °°). For a given pair of np and/,/n/,' a simplified
curve is obtained by joining the first divergent point to the last lower bound
point. The simplified curves so obtained are independent of the bond pa-
788

J. Struct. Eng. 1990.116:769-790.


rameter. This technique also allows us to avoid the issue of possible change
in the bond stress-slip relationship with the loading.
The simplified force-deformation and the effective concrete tensile stress-
strain curves were used to predict the behavior of the reinforced concrete
bars tested by Hwang and Rizkalla (1983). The experimental results showed
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

a marked reduction in the tensile capacity of concrete at higher deformations,


which could not be explained by the proposed relationships. A tentative so-
lution was to use a variable concrete tensile strength that diminished expo-
nentially as a function of the average tensile strain multiplied by a "damage
parameter," C. A constant value of C equal to 550 gave curves that were
in reasonable agreement with the experimental points.
Further investigations are needed to study the character of the damage
parameter C. It may well be possible to arrive at a function of variables
related to the properties of the materials, geometry of the structural member,
and loading level that would describe the parameter.
We believe that the proposed model is a step in the right direction. A
more comprehensive model may include a bond stress-slip relationship that
accurately accounts for the effect of different variables, and a better descrip-
tion of the spatial deformations and stresses along and across the bar—es-
pecially near the cracks, which may include concepts of fracture mechanics.

POST SCRIPT

The present paper on uniaxial behavior of reinforced concrete and the two
other papers dealing with biaxial behavior (Gupta and Maestrini 1989a, 1989b)
are all based on Maestrini and Gupta (1987). The later date of publication
of the present paper represents the extra time it took to rewrite the paper in
response to the first set of review comments, not because of additional new
research by us. The comments resulted in many improvements in the paper.
We thank the reviewers for their effort.

APPENDIX. REFERENCES

Bazant, Z. P., and Oh. B. H. (1984). "Deformation of progressively cracking rein-


forced concrete beams." J. Amer. Concr. Inst., 81(3), 268-278.
Edwards, A. D., and Picard, A. (1972). "Theory of cracking in concrete members."
J. Struct. Div., ASCE, 98(12), 2687-2700.
Floegl, H., and Mang, H. A. (1982). "Tension stiffening concept based on bond
slip." J. Struct. Div., ASCE, 108(12), 2681-2701.
Gilbert, R. I., and Warner, R. F. (1978). "Tension-stiffening in reinforced concrete
slabs." J. Struct. Div., ASCE, 104(12), 1885-1900.
Goto, Y. (1971). "Cracks formed in concrete around deformed tension bars." J.
Amer. Concr. Inst., 68(4), 244-251.
Gupta, A. K., and Maestrini, S. R. (1989a). "Post-cracking behavior of membrane
reinforced concrete elements including tension-stiffening." J. Struc. Engrg., 115(4),
957-976.
Gupta, A. K., and Maestrini, S. R. (1989b). "Unified approach to modeling post-
cracking membrane behavior of reinforced concrete," J. Struct. Engrg., 115(4),
977-993.
Hageman, L. J., Murakami, H., and Hegemier, G. A. (1986). "On simulating steel-
concrete interaction in reinforced concrete. Part II: Validation studies." Mech. of
Matls., 5(2), 187-197.
Hegemier, G. A., Murakami, H., andHageman, L. J. (1985). "On tension-stiffening
789

J. Struct. Eng. 1990.116:769-790.


in reinforced concrete." Mech. of Math., 4(2), 161-179.
Houde, J., and Mirza, M. S. (1972). "A study of bond stress-slip relationships in
reinforced concrete." Struct. Cotter. Series, No. 72-8, Structs. Lab., McGill Univ.,
Montreal, Canada.
Hwang, L. S., and Rizkalla, S. H. (1983). "Behavior of reinforced concrete in ten-
Downloaded from ascelibrary.org by MARRIOTT LIB-UNIV OF UT on 11/23/14. Copyright ASCE. For personal use only; all rights reserved.

sion at post-cracking range." Engrg. Rept., Dept. of Civ. Engrg., Univ. of Man-
itoba, Winnipeg, Canada.
Jiang, D. H., Shah, S. P., and Andonian, A. T. (1984). "Study of the transfer of
tensile forces by bond." J. Amer. Cotter. Inst., 81(3), 251-259.
Lin, C. S., and Scordelis, A. C. (1975). "Nonlinear anslysis of RC shells of general
forms." J. Struct. Div., ASCE, 111(3), 523-538.
Lutz, L. A., and Gergely, P. (1967). "Mechanics of bond and slip of deformed bars
in concrete." J. Amer. Cotter. Inst., 64(11), 711-721.
Maestrini, S. R., and Gupta, A. K. (1987). "Membrane behavior of reinforced con-
crete shell elements including tension-stiffening." Reinforced Concr. Shell Res.
Rept., Dept. of Civ. Engrg., North Carolina State Univ., Raleigh, N.C.
Mirza, S. M., and Houde, J. (1979). "Study of bond stress-slip relationships in
reinforced concrete." J. Amer. Concr. Inst., 76(1), 19-46.
Murakami, H., and Hegemier, G. A. (1986). "On simulating steel-concrete inter-
action in reinforced concrete, part I: Theoretical development." Mech. of Matls.,
5(2), 71-185.
Nilson, A. H. (1971). "Bond stress-slip relations in reinforced concrete." Report No.
345, Dept. of Struct. Engrg., Cornell Univ., Ithaca, N.Y.
Rizkalla, S. H., and Hwang, L. S. (1984). "Crack prediction for members in uniaxial
tension." / . Amer. Concr. Inst., 81(6), 572-579.
Scanlon, A., and Murray, D. W. (1974). "Time dependent reinforced concrete slab
deflections." J. Struct. Div., ASCE, 100(9), 1911-1924.
Somayaji, S. (1979). "Composite response, bond stress-slip relationships and crack-
ing in ferrocement and reinforced concrete." Thesis presented to the University of
Illinois, at Chicago, 111., in partial fulfillment of the requirements for the degree
of Doctor of Philosophy.
Somayaji, S., and Shah, S. P. (1981). "Bond stress versus slip relationships and
cracking response of tension members." / . Amer. Concr. Inst., 78(3), 217-225.
Tassios, T. P., Yannopoulos, P. J. (1981). "Analytical studies on reinforced concrete
members under cyclic loading based on bond stress-slip relationships." J. Amer.
Cone. Inst., 78(3), 206-216.
Tepfers, R. (1979). "Cracking of concrete cover along anchored deformed reinforc-
ing bars." Concr. Res., 31(106), 3-12.
Tepfers, R. (1980). "Bond stress along lapped reinforcing bars." Concr. Res., 32(112),
135-142.
Yankelevsky, D. Z. (1985). "Bond action between concrete and a deformed bar—
A new model." J. Amer. Concr. Inst., 82(2), 154-161.

790

J. Struct. Eng. 1990.116:769-790.

You might also like