You are on page 1of 22

F I B E R PULLQUT AND B O N D S L I P .

I:
ANALYTICAL STUDY
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

By Antoine E . Naaman, 1 Member, ASCE, George G. Namur, 2


Jamil M. Alwan, 3 and Husam S. Najm 4

ABSTRACT: This is the first of a two-part paper involving a fundamental study of


bond in fiber-reinforced cement composites. The concept of a relationship between
the bond shear stress and the local slip at the interface between the reinforcement
and the matrix for fiber-reinforced, cement-based composites is relatively new.
Moreover, the difficulties involved in conducting strain measurements on short
discontinuous fibers have made it impossible so far to develop experimental bond
shear stress versus slip curves for fiber-reinforced concrete. This paper involves a
fundamental study of the bond-stress-slip relationship between steel fibers and
concrete. The analysis consists of a primal problem, whereby a complete pullout
curve is predicted from an assumed bond-slip relationship, and a dual problem in
which the bond-slip relationship is obtained from an experimental pullout curve.
The frictional behavior of the interface is also related to and explained by the
shrink-fit and fiber-matrix misfit theory. The theory derived in this paper is vali-
dated in a second paper using experimental pullout load versus slip curves.

INTRODUCTION

Cementitious materials such as mortar and concrete are known for their
weakness in resisting tensile stresses. Fiber reinforcement makes up for this
deficiency. Fiber-reinforced composites resist tensile forces through a com-
posite action, whereby part of the tensile force is resisted by the matrix,
while the balance is taken by the fibers. The transmission of forces between
the fiber and the matrix is achieved through bond defined as the shearing
stress at the interface between the fiber and the surrounding matrix. It is
generally agreed that the fiber contribution to increasing the toughness of
the composite is primarily dictated by the mechanisms of fiber pullout.
Pullout tests on fiber-reinforced cement composites have been conducted
frequently in the past; however, a review of existing literature shows that
no complete analytical study of the mechanics of pullout has been performed
as yet (Edwards and Yannopoulos 1979; Eligehausen et al. 1983; Gao et al.
1988; Gopalaratnam and Cheng 1987; Gopalaratnam and Shah 1987; Gray
1984a, b ; Mandel et al. 1987; Mindess 1987; Naamam et al. 1989; Namur
et al. 1987; Namur and Naaman 1989; Nilson 1972; Pinchin and Tabor 1978;
Wise, et al. 1987; Shah and Jenq 1987; Timoshinko 1941; Wang et al. 1987;
Wei et al. 1986).
'Prof., Dept. of Civ. Engrg., 2340 G. G. Brown Bldg., Univ. of Michigan, Ann
Arbor, MI 48109-2125.
2
Prin. Engr., ABB hnpell Corp., 300 Tristar Int., Suite 400, Lincolnshire, IL 60069.
'Doctoral Student and Res. Asst., Dept. of Civ. Engrg., Univ. of Michigan, Ann
Arbor, MI.
4
Doctoral Student and Res. Asst., Dept. of Civ. Engrg., Univ. of Michigan, Ann
Arbor, MI.
Note. Discussion open until February 1, 1992. Separate discussions should be sub-
mitted for the individual papers in this symposium. 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 Jan-
uary 2, 1990. This paper is part of the Journal of Structural Engineering, Vol.
117, No. 9, September, 1991. ©ASCE, ISSN 0733-9445/91/0009-2769/$1.00 +
$.15 per page. Paper No. 26190.

2769

J. Struct. Eng., 1991, 117(9): 2769-2790


The main objective of this study is to develop a mathematical model that
describes the response of straight smooth fibers embedded in a cementitious
matrix and subjected to a pullout load. In the formulation of the pullout
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

model, a relationship between the bond shear stress and the relative slip at
any point along the fiber-matrix interface is assumed. Such a relationship is
viewed as the constitutive property of the fiber-matrix interface in a similar
manner as the bond-shear-stress-slip relationship for bars in reinforced con-
crete or strands in prestressed concrete (Edwards and Tannopoulos 1979;
Eligehausen et al. 1983; Nilson 1972). While pullout tests of fibers are easy
to conduct, there is, to date, no rigorous experimental method that yields
their bond-slip relationship. Thus, a dual problem was also solved, whereby
the bond-slip relationship can be found from experimental pullout curves.

EXISTING MODELS OF FIBER PULLOUT

The primary reason behind the addition of fibers to cementitious matrices


is to delay and contain cracking. While it is generally believed that the in-
clusion of fibers enhances the precracking behavior of a cement composite
by increasing its cracking strength, the effect of fiber addition becomes dra-
matic only after cracking. This is due to the fact that the fibers bridge the
cracked parts of the matrix, thus delaying the sudden global failure of the
composite. In the postcracking stage, therefore, the behavior of fibers is
governed by their interfacial bond stress response as being subjected to pull-
out loads. Thus, the need to know more about the pullout behavior of fibers
has encouraged increased research in this area.
A review of available analytical and experimental studies pertaining to
bond and pullout in fiber-reinforced cement composites was conducted in
Naaman et al. (1989). Here, because of space limitation, only analytical
studies related to the present investigation are briefly reviewed.
Wang et al. (1987) developed a theoretical model to predict the load-crack
separation relationship for synthetic fiber-reinforced concrete. The distinc-
tive feature of the model lies in the recognition that the fiber-matrix bond
strength at any point on the fiber is a function of the slippage accumulation
at that fiber segment. However, the authors did not include the Poisson effect
in their model, claiming that slight fiber misalignment and the presence of
channel surface asperities could offset this effect. This may affect the va-
lidity of their model, especially at the stage where the normal contact pres-
sure on the fiber is critical. Wang et al. concluded that the elastic bond
strength has little effect when low-modulus fibers, such as nylon and poly-
propylene fibers, are used, and frictional bond strength therefore dominates
the pull-out resistance. The fiber was accordingly assumed to be linear-elas-
tic in their model, with sufficient strength to enable complete pullout without
rapture. The contribution of matrix deformation to crack separation was ne-
glected as well.
In addition to the analytical model, Wang et al. (1987) performed pullout
tests on nylon and polypropylene fibers in cementitious matrices. In those
experiments, fiber surface abrasion was observed. This increased the fiber-
matrix bond strength during pullout, resulting in higher pullout forces and
energy absorption compared to the case of constant frictional bond strength.
Shah and Jenq (1987) investigated the interfacial bond properties between
fibers and matrix by conducting an analysis of pullout tests. The bonding
2770

J. Struct. Eng., 1991, 117(9): 2769-2790


between fibers and matrix was assumed to be perfect before the pullout load
is applied. Griffith energy criteria were used to govern the crack propagation
at the interfacial region where debonding begins. A constant frictional shear
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

stress was assumed to exist at the interface in the debonding region due to
surface roughness. Based on this mechanism, Shah and Jenq decomposed
the total pullout load resistance into one offered by the interfacial bond and
another one due to the frictional stresses. The authors also developed a frac-
ture model that predicts the progressive failure of the interfacial bond. How-
ever, their model applies only to the ascending branch of the pullout load
curve. Shah and Jenq reported that the interfacial strength of composite sys-
tems can be characterized by two material parameters: the critical debonding
energy release rate and the interfacial frictional stress. They also observed
that the major contribution of energy absorption due to the addition of fibers
is mainly provided by interfacial frictional forces during fiber pullout. As a
result, they recommended that fiber fracture in a fiber-reinforced composite
system be avoided if high-energy absorption is desired.
Gopalaratnam and Cheng (1987) formulated the pullout problem in one
dimension, with an emphasis on the nonlinear interfacial response. In their
work, they assumed that the fiber and the concentric matrix behave elasti-
cally. The local interfacial bond-slip relationship was also assumed to feature
linear softening. This assumption greatly simplifies the implicit governing
differential equation of the debonding process. However, the fact that this
assumption was not substantiated by any physical evidence may affect the
validity of the model. As part of their work, Gopalaratnam and Cheng in-
vestigated the stability of the debonding process by varying the fundamental
characteristics of the fiber-matrix interface, the fiber embedment length, and
the fiber diameter. Their conclusions were limited to the effect of the soft-
ening of the interface on the pullout strength, composite ductility, energy
absorption, and the stability of the debonding process.
Finally, Namur et al. (1987) and Namur and Naaman (1989) have studied
the problem of pullout as well as the relationship between pullout curves
and bond shear stress versus slip curves. Simplifying assumptions were made
in the analytical development, thus limiting the applications of the model.
Furthermore, they assumed that the frictional stress is constant and inde-
pendent of the slip. However, experimental tests where slip was accurately
measured indicated otherwise. The present model accounts for these defi-
ciencies.

STATEMENT OF PROBLEM

The analysis that follows pertains to a pullout test whereby a tensile force
P is applied to the tip of a fiber embedded over a length I in a cementitious
body (Fig. 1). A monotonic increase in the value of P is accompanied by
a displacement at the tip of the fiber and leads to progressive debonding
along the fiber-matrix interface. Once debonding reaches the embedded end
of the fiber, a dynamic mechanism of pullout is observed, whereas a dis-
placement at the free end is also accompanied by a displacement at the
embedded end.
Initially, it is assumed that the relationship between the bond shear stress
at the interface of the fiber and the matrix as well as the relative slip between
the same components are as shown in Fig. 2. A more accurate relationship
2771

J. Struct. Eng., 1991, 117(9): 2769-2790


Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

FIG. 1. Pullout Test Configuration

(/>
OC max
W 1.
DC
<%£
X

o
SLIP
FIG. 2. Assumed Bond Shear Stress versus Slip Relationship

will be introduced later in this paper. The bond-slip curve is linear-elastic


up to the point where the bond strength Tmax of the interface is reached,
beyond which purely frictional conditions prevail, with a constant frictional
shear stress equal to Tf. It is also assumed that iy cannot exceed Tmas. While
it is understood that a constant bond stress equal to i/ cannot be sustained
for large slips, it is further assumed (in modeling the ascending initial portion
only) that slips causing the bond stress to diminish significantly beyond ly
will not be attained. The writers believe that the curve need not necessarily
have a zero ordinate intercept. Indeed, the chemical adhesion component of
the bond is not slip-induced. However, this study is limited to cases where
the chemical adhesion is assumed negligible. Following the modeling of the
pullout load response at small slips, the case of large slips with frictional
decay is treated separately.
The model described next leads to the axial stress and strain distribution
in the fiber and the matrix, as well as to the shear stress distribution at their
interface.
2772

J. Struct. Eng., 1991, 117(9): 2769-2790


MATHEMATICAL DERIVATION: SMALL SLIPS

Basic Equations
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

The free-body diagram of an infinitesimal segment of fiber (Fig. 3) leads


to
dF - vifdx = 0 (1)
where F = the local force in the fiber at a distance x from the embedded
end of the fiber; \\i = the perimeter of the fiber; and T = the local shear
stress at the interface between the fiber and the matrix.
Eq. (1) can be rewritten as
dF
^ = T* (2)

The product TI|> = the shear force per unit length, or the shear flow t at
the interface. Thus
t= TI|I (3)
Furthermore, the local tensile force F in the fiber can be related to the
local strain in the fiber ef through the following:
F
= AfEfy (4)
where Af and Ef = the area and the elastic modulus of the fiber, respectively;
and €/ = the local strain in the fiber.
Moreover, in the elastic region, the local shear stress T can be related to
the local slip S through the bond modulus K as follows:
T = KS (5)

Perimeter of fiber = \|/


P

_ ^4^
x

F+dF

FIG. 3. Free-Body Diagram of Infinitesimal Segment of Fiber

2773

J. Struct. Eng., 1991, 117(9): 2769-2790


where K is assumed constant, and the slip S is defined by
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

S = (hf- 8 J = [ef(x) - tjpc)}dx (6)


Jo
where 8y and 8m = the local displacement in the fiber and the matrix, re-
spectively, and e7 and em = the corresponding strains.
To satisfy static equilibrium, the total force P applied at the free tip of
the fiber has to be resisted at any section by the local force in the fiber F,
as well as the local force in the matrix T, or
P =F +T (7)
P = AfEfef + AmEmzm , (8)
where Am and Em = the area and the elastic modulus of the matrix, respec-
tively; and em = the local strain in the matrix.
Differentiating (2) and (6) leads to
d2F IMT
—r = (9)
dx dx
<h= (<®f_d^\
dx \dx dx]
Combining (9) and (10), and recognizing the fact that the first derivative
of the local displacement function is the local strain, it can be proven that
d2F
— = i K ( 6 / - e») (11)
dx
Replacing em in (11) by its value extracted from (8) leads to
fF
— = -KP + KQF{x) (12)
dx-
where
llfK
K = -r— (13)
A F
and

2 = i + -f-r A E
(14)
f f
Eq. (12) is a second-degree differential equation in F of the form
z
dF
-1-\2F=-KP (15)
dx
where
X = VKQ (16)
The solution to this differential equation is of the form

2774

J. Struct. Eng., 1991, 117(9): 2769-2790


F(x) = A'eXx + B'e'** + - (17)
Q
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

The unknown coefficients A' and B' are determined from the following
boundary conditions:
F(0) = 0 . (18)
F(l) = P (19a)
where

A' = •
1- e (•-^u+H <1W
and

B' =
1
-..-IV*
Q
' (19c)

Introducing the coefficients A and 5 equal to A'IP and 5 7 F , respectively,


we get
1 <2o>
A =
('-s^'^H
and

fi = (21)
i
Using these expressions for A and B, the force F(x) in the fiber and the
interfacial shear flow t(x) can be respectively expressed as

F(x) = P\ Ae^ + Be~Xx + - (22)

dF
t(x) = — = P\(AeXx - Be~^), (23)
dx
As can be seen from (23), the interfacial shear flow t is a direct function
of the load P. Furthermore, the shear flow is a direct function of the shear
stress as in (3). Also, the maximum shear stress (or flow) for a given load
will always occur at the point where the fiber penetrates the matrix, or at x
= I.

Critical Force
Given the bond-slip relationship, there will be a critical force P crit that will
induce a shear stress at x = I equal to Tmax, or a shear flow equal to fmax
where

^max TmaxT • (24a)

Setting t(l) = fmax yields the value of Pa

2775

J. Struct. Eng., 1991, 117(9): 2769-2790


(24b)
l-^)(l + e^')+(~)2e-
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

When a force P ,< Pcrit, elastic bond conditions prevail at the interface,
and no debonding occurs, i.e., the fiber remains fully bonded to the sur-
rounding matrix.

Elastic Displacement for P < Pctit


The displacement of the free end of the fiber (at x = /) can be evaluated
using

) - cMkfc (25)
Jo

Carrying out the integral in (25), A is found to be


A P(Q - 2) / l - e" x '\
A= — (26)
kAmEm \l+e-uJ
As can be seen from (26), there is a linear relationship between the applied
force P and the displacement of the fiber free end. Moreover, in a pullout
curve, the slope of the ascending linear portion of the curve is given by

(2?)
(fi-2)!-,-

Debonding Zone
When the applied force P exceeds Pctit, a region identified as the zone of
debonding will develop and grow as the applied force P increases. In other
words, two interfacial zones will adjacently coexist, one that is bonded
and one that is debonded because shear stresses have exceeded Tmax. Fig. 4
shows typical bond stress distributions for different values of the applied
load P. The forces resisted by these two individual zones will be identified
as bonded force Pb and debonded force Pd, respectively. To satisfy static
equilibrium, it can be inferred that for any load P a Pcrit and less than
the peak load
P = P„ + Pi (28)
Based on the assumed bond-shear-stress-slip relationship shown in Fig.
2, the interfacial shear stress prevailing in the zone of debonding is constant
and equal to if. This means that the normal force distribution in the fiber is
linear, decreasing at the rate of tf per unit length, where
fr = T/«|» (29)
In this study, the length of the zone of debonding will be noted by u. The
length of the bonded zone is thus (I — u). Over this bonded length, the same
shear-stress distribution prevails as in the case where P s PCIit, except that
the force is P' = (P - tfu), and the length is (/ - u), as shown in Fig. 5.
2776

J. Struct. Eng., 1991, 117(9): 2769-2790


Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

maxy
x
f

I
i
tr
r-u . u

r-u u
FIG. 4. Typical Bond Stress Distribution for Cases: (a) p = Pcril; (b) P = P^>
Pcrit; and (c) P = P2 > p ,

Therefore, by using the proper distribution of the interfacial bond stress, one
can state that
1 - e -2\(I-u)
PH = - (30)
X 2
-MZ-u) M
+ 1 - - [1 + e^ '-"']

and
P i = ?>« (3D
Then, P, which is equal to the sum of Pb [given in (30)] and Pd [as per
(31)], can be found as
1 - e-2HI-U)
P = tfu + (32)

Q \ Ql
2777

J. Struct. Eng., 1991, 117(9): 2769-2790


Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

T(x)

x =0
p>p
crif

F(x)

x =0 X=f-D X =C

FIG. 5. Bond Stresses and Normal Force Distributions under Partially Debonded
Conditions

To find the displacement at the free end, (25) is still valid, except that
the integral will be carried out separately over each of the two zones. Thus
1 -e~ e-2
P(Q - l)« - ^ - (2 2) + (P - tfu) - ttul
1 + e~
A=- (33)
AmEm
It should be observed that the given mathematical solution to predict the
ascending branch of the pullout load versus end slip curve applies for a
pullout test as well as for a pull-through test having the same embedded
length.
To obtain the presented closed-form solution [(30)-(33)] for the ascend-
ing branch of the pullout load versus slip curve, it was assumed that the
bond shear stress versus slip curve remains constant after the peak stress
(Fig. 2). This assumption, as discussed in the following, is only a good
approximation. However, it can be shown that when the peak pullout
load is attained, the corresponding maximum end slip, Aa on the pullout
load versus end slip curve is relatively small, and thus the error that may
be introduced is negligible. On the other hand, if it is assumed that the
shear stress decays continuously following the peak stress, only a numer-
ical solution would be possible. This was not found necessary in this
study.

2778

J. Struct. Eng., 1991, 117(9): 2769-2790


DYNAMIC MECHANISM OF PULLOUT

The value of the length of the debonded zone u goes from zero at the
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

onset of debonding to / at complete debonding (purely frictional bond shear


stresses at the interface). For each value of u, one value for the pullout load
and its corresponding end slip exist.
Once u has reached /, a dynamic mechanism of pullout develops. The
relationship between the pullout load and the slip can now be derived. If
the rigid-body displacement of the fiber is referred to as v (Fig. 6), then the
pullout force is
P = tfQ-v) (34)
The force F within the fiber at a distance x from the point at which the
fiber penetrates the matrix is
F(x) = tf{l -v-x) (35)
The dynamic pullout slip A is now equal to the total rigid-body movement
of the fiber v added to the elastic elongation of the fiber relative to the
matrix. However, the elastic elongation is relatively small as compared to
the rigid-body movement of the fiber. This suggests an almost perfectly lin-
ear relationship between the applied pullout force P and the dynamic pullout
slip A of the form
P = ?/(/- A) (36)
Pullout tests conducted in Naaman et al. (1989) led to a descending branch
that is not linear but descends more steeply than expected (Fig. 7), sug-
gesting a decay in the frictional bond. In other words, the assumption that
the bond shear stress remains constant after the bond strength Tmax is reached
is not representative at large slips.

(a)
1

-0. (b)
C-y

FIG. 6. Pullout Test under: (a) Pre-Peak Conditions; and, (b) Dynamic Pullout
Conditions
2779

J. Struct. Eng., 1991, 117(9): 2769-2790


Smooth Fibers
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

max
Full Debonding

Frictlonal Phase

A. A A 0
cnt max End Slip
FIG. 7. Typical Pullout Load versus Slip Relationship of Smooth Fibers

An alternate way to look at the descending branch would be to assume


that the bond shear stress versus slip curve is as shown in Fig. 8. Therefore,
it is assumed that at some point after the bond strength Tmax is reached, the
shear stress starts decreasing with the slip. In what follows, it will be as-
sumed that the end slip for which the friction starts decaying and the dy-
namic mechanism of pullout begins is A„. Under this assumption, the equa-
tions given previously for both the fully bonded and partial-bonding cases
are still applicable. In this stage, a state of complete failure of the elastic
bond between the fiber and the matrix exists. Thus, the only phenomenon
that describes the resistance of the fiber against pullout is friction.

Normal Contact Pressure


The frictional stress is equal to the normal contact pressure on the steel
fiber multiplied by a coefficient of friction.
Trf = W» (37)

where pN = the normal contact pressure between the fiber and the matrix.
From examining the free-body diagram of the embedded fiber in Fig. 6,
one can write from equilibrium that
dF = Trfi}itfcc , (38)
where dF = the differential of the local force in the fiber at a distance x.
If we divide (38) by the cross-sectional area of the fiber, we get
2td
dff= — dx (39)
r
f

2780

J. Struct. Eng., 1991, 117(9): 2769-2790


€0
CO T
UJ *
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

DC max
V)
Exponential frictional decay
<

cnt SLIP

Exponential frictiona! decay


Assumed initial friction

y
SLIP A100
FIG. 8. Alternative Bond Shear Stress versus Slip Relationship with Frictional
Decay: (a) Total Curve; and (h) Enlarged Scale

Substituting the right-hand side of (37) into (39) leads to


2 W N
At A (40)
dff = dx
But from the shrink-fit theory (Timoshinko 1941), the interfacial contact
pressure pN in a shrink-fit configuration with no load on the fiber and with
rf « rm is given by

PN '• (41)
(1 + vj | (1 - vf)

2781

J. Struct. Eng., 1991, 117(9): 2769-2790


fiber
Misfit
mom -_[^ r a d j u s jl_r|^hole radius Ji
after shririkage
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

Fiber diameter
or hole before
shrinkage

Matrix
Shrinkage i i
Hole after
shrinkage
(in absence of the fiber)

FIG. 9. Fiber-Matrix Misfit

where e.mr = the radial shrinkage strain in the matrix in addition to the strain
due to any externally applied confining load present.
The radial shrinkage strain emr in the matrix can be expressed as
8
emr = - (42)
r
f

where 8 = the matrix-fiber misfit (Fig. 9). The misfit was defined by Pinchin
and Tabor (1978) as "the difference between the fiber radius and the hole
radius in the absence of the fiber." Numerical computation values obtained
from (41) seem to lead to a very reasonable prediction of bond strength.

Poisson's Effect
When the fiber is loaded longitudinally with a stress^-, it will be subjected
to a Poisson's contraction e/r such that

«* = f"/ («)
This strain eA will reduce the interfacial contact pressure caused by the
original matrix strain; thus, let us define
«eff = €mr - *fr (44)

or

8 ff
rf Ef
Substituting the right-hand side of (45) into (41) and solving the resulting
differential equation, the average stress in the fiber can be expressed as
2782

J. Struct. Eng., 1991, 117(9): 2769-2790


ft = I 1 - exp (46)
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

where x = the embedded length of the fiber.


Multiplying (46) by the cross-sectional area of the fiber yields an equation
for the pullout load as a function of the embedded length x
-2vf[ix SEfitrf
P = I 1 - exp (47)
(1 + vj , (1 - vf
E r
\ ff
Em

Effect of Decay in Misfit


It is worth noting that 8, the fiber-matrix misfit, deteriorates and decreases
as the fiber is pulled out, due to a combined action of abrasion and com-
paction of cement and sand particles surrounding the fiber. This is partic-
ularly true for steel fibers, which are harder than the matrix.
In this study, it will be assumed that the decreasing trend of the misfit is
exponential, as suggested by experimental pullout curves. The following
expression for the fiber-matrix misfit 8 is proposed in this study:
C-(A-A> _ *,-</)"
8 = 8„ —rr^TT^r
-((-A+A^l
(48)
1-fc'
in which A„ = end slip of the fiber at the onset of full debonding (u = I);
A = the end slip; 8„ = the initial fiber-matrix misfit at the onset of dynamic
mechanism; and ij = a constant to give the analytical descending branch the
same asymptotic value as the experimental one.
The initial fiber-matrix misfit 8„ is calculated by equating the pullout load
that corresponds to an embedded length x = I in (47), i.e., P is replaced by
tf /, and thus
JflVf_
8.= 1 - exp ) — —[\ (49)
EfrrA (1 J , (1 - vf)
\Efrf

A0 can be obtained from (33), with P = tfl and u I, or


2
_ (fi - 2)tfl
(50)

In pullout tests, the embedded length of the fiber is equal to the difference
between the original embedded length and the end slip, i.e., x = I — A +
A0. On the other hand, in pull-through tests, the embedded length of the
fiber is constant and equal to the original embedded length, i.e., x = I. Fig.
10 shows a typical plot of the variation of the misfit as related to the pull-
out end slip of the fiber.

Frictional Shear
It is important to mention that the described model makes it possible to
find an equivalent value for the frictional shear bond for any given pull-out
load, or end slip A, since

2783

J. Struct. Eng., 1991, 117(9): 2769-2790


Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

0.0 0.2 0.4 0.6 0.8 1.0


Pull-Out End Slip (In)
FIG. 10. Deterioration Function of Misfit, H2SF Series (1 in. = 25 mm)

P= ^A)* (51)
where x = the embedded length of the fiber; and

•tfA) = — = (52)
\}JJC «|» (I - A + A„)

The value of P obtained from (47) can be divided by the corresponding


embedded length, thus giving an equivalent value for the frictional shear
bond for a given slip (see Fig. 11). Thus
-2vf\x,(l - A + A„)
T,(A) 1 — exp
"(1 + vJ (1 - vf)~
\Efrf
Cf J'
-(A-AJ" m 11-
-CO
50[e &~ Jfi/irr,
( A+A (53)
*[1 - ge- '- »n(Z - A + A0)vf
is valid for A > A„. A typical plot of Trf versus x, where x = I — A + A„
is shown in Fig. 11.
The initial frictional stress ly is equal to Trf(A0), thus
—2vf\il b0EfTrrf
rf = I 1 - exp (54)
(1 + vm) | (1 - vf) tylvf
\Efrf

Therefore, the decaying frictional stress id can be expressed as


ke-df
-(A-A)l
T,,(A) = lf •
1 - g e -<'- A+ V
2784

J. Struct. Eng., 1991, 117(9): 2769-2790


w 300
SAMPLE: H2SF
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

V) Equivalent Bond Stress


200 Due to Fractional Shear
if
to
o 100
W Elastic Bond Stress

o
0.00 0.02 0.04 0.06
Pu!I-Out End Slip (in)
FIG. 11. Equivalent Frictional Bond H2SF Series (1 in. = 25 mm; 1,000 psi = 7
MPa)

-2vf\L{l - A + A0)
1 — exp
'd + vj (1 ~ v/)"
\Efrf
(55)
—2vf\iJ
1 — exp
(1 + v j (1 - vf)
\Efrf
*-<m *f J'

PREDICTION OF PULLOUT CURVE: PRIMAL PROBLEM

For a given bond-shear-stress-slip relationship of the type shown in Fig.


8, a complete pullout curve can be predicted. The predicted curve (Fig. 7)
is divided into three distinct regions and can be obtained using the following
procedure:

1. Precritical region: The critical point (Pent Ant) obtained from (24) and (27)
is enough to describe this linear zone, which extends from the origin to the
critical point.
2. Partial debonding region: For each value of the length of the debonding
zone u, one point on the curve can be found by computing the corresponding
pullout force and end slip [(30-(33)]. As many points as needed within this range
can be found, provided that the value of u is taken between zero and the fiber
length /. A typical example showing pullout load versus debonded length is shown
in Fig. 12.
3. The pullout region: For each value of the end slip A (A„ ^ A < / ) , a value
for the pullout load leading to one point on the pullout curve can be obtained

2785

J. Struct. Eng., 1991, 117(9): 2769-2790


JU"
T m = 400 psi; Fiber Oiam. = 0.019 in
Bond Modulus = 5E(6)psl/ln x
f = 400 psi
V) Embedded Length = 1.0 In
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

x
20- f = 300 psi

re x
© f = 200 psi

3 10-
x
o 1 = SO psi
I
"3
a.
o-
0.0 0.2 0.4 0.6 0.8 1.0
Normalized Debonded Length (u/L)
FIG. 12. Pullout Load versus Debonded Length (1 in. = 25 mm; 1,000 psi = 7
MPa; 1 lb = 4.45 N)

from (47), where x = I — A — A0When


. (A — A„) reaches the value of /, the
force would be equal to zero.

PREDICTION OF BOND-SLIP CURVE: DUAL PROBLEM

Given an experimental pullout load versus slip curve for a given fiber, the
bond shear stress versus slip curve can be theoretically obtained, assuming
that it is of the type described in Fig. 8(a) and 8(&), i. e., the linear-elastic
ascending branch, followed by a purely frictional region, then a deteriorating
frictional zone of the kind described in (55). The whole curve can thus be
described by five parameters: the bond modulus K; the bond strength Tmax;
the constant frictional bond stress y the value of the end slip A0 at which
the bond is assumed to deteriorate; and the decaying frictional parameters £
and T|, describing the deteriorating frictional zone.
The bond modulus K is determined from the slope of the linear asscending
portion of the pullout curve, which can be determined graphically. Further-
more, the value of Q [(14)] can be evaluated from the physical and me-
chanical properties of the fiber and the matrix. Judgment and common sense
are to be used in evaluating the area of the matrix Am. Indeed, only a fraction
of the matrix cross-sectional area is effective if the proportion of the spec-
imen's cross-sectional area to that of the fiber is relatively large. However,
for low-volume fractions of fibers such as used in conventional fiber con-
crete, it can be shown that the solution to the dual problem is not very
sensitive to the value of Am. Once Q and the slope F/A are known, X. can
be solved for in (27) by iteration or by some numerical procedure. Having
found X, and using (16), K can now be solved as follows:

(56)
Q
Finally, the value of the bond modulus K can be derived from (13)
2786

J. Struct. Eng., 1991, 117(9): 2769-2790


1
(57)

The next step is to evaluate the bond strength of the interface Tmax as well
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

as the frictional bond y. The peak load Pp and the corresponding end slip
Ap are to be studied first. The peak load will generally occur under partial
debonding conditions. The value of u corresponding to that case is the value
of u that would maximize P. Thus, for u = up, we have

£) =0
du/u=u
p

To make the differentiation easier, a change of variables is in order. Intro-


ducing the variable X, defined as
x = e-X('-») (58a)
thus
dP\
= 0. (58fc)
dX/x=x
P

where X„ = <rM'"~V.
To get the values of tf and tmax, a system of three nonlinear equations in
three unknowns must be solved, the three unknowns being tf, fmax, and Xp.
The system of three equations in three unknowns is thus
-2
4 U - - ^ - 5
l
f 'max
(59)
XX,
1 - - \XP + -XP,-
Ql Q Q

fln(Xp) + tfl + ^ — i ~xi (60)


A. X 2
x+ +x
a ' V~-e)" '>
In (Xp) In (Xp) In (Xp)
PP(Q ~ 1) +I '/ +/ ( 2 - 2) - tfl +1

AmEm
In (Xp) (Q-2)/i-Xf
+1
1 +X
(61)
AmEm
Once (59)-(61) have been solved simultaneously for tf, tmax, and Xp, the
values of ly and Tmax can be found from (24) and (29). As a check for the
solution, the value of up corresponding to the value found for Xp must be
between zero and I.
The experimental pullout curves developed in the course of this investi-
gation suggested a steep initial decay in the post-peak pull-out behavior. A

2787

J. Struct. Eng., 1991, 117(9): 2769-2790


parametric study of (55) indicated that the rate of initial decay can be rep-
resented by a factor t\. A value of T] = 0.2 was found to best fit the ex-
perimental curves obtained in Naaman et al. (1989).
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

To evaluate the factor £, one point on the experimental pullout curve is


needed, say (JPX,AX). Then, from (47), and using the values of Px and Ax, 8
can be solved as follows:

8 = ^ - (l - exp f ^ —— I\ (62)

\ N E
Then, substituting in (48), £ can be solved as follows:

2
e -'°' _A e -('-A,-A/ 2

2
e-(A,-A/'_ A
8„
where 8 is given in (62) and 80 is given in (49).
With the five basic parameters K, Tmax, y, A0, and £ known, the whole
bond-shear-stress-slip relationship can be constructed.

CONCLUDING REMARKS

The analysis developed in this investigation allows for a realistic modeling


of the entire pullout process of a smooth fiber from a cementitious matrix.
It is also applicable to fiber-reinforced ceramic matrices in which similar
conditions prevail, and to the problem of pullout of a smooth reinforcing
bar or wire embedded in concrete. While the primal problem assumed an
idealized bond-stress-slip relationship of the interface, a modified relation-
ship with decaying response could also be accommodated, especially in the
descending branch. The solution of the dual problem led to the prediction
of the bond shear stress versus slip curve assuming the pullout load versus
slip curve is given or obtained from experiments. The bond shear stress ver-
sus slip curve can be considered a constitutive property of the interface. The
pull-out model proposed applies primarily to smooth fibers or bars, or to
fibers having a uniformly rough surface. It does not apply to the case of
hooked fibers or deformed bars for which the mechanical contribution of the
hook or lugs to the pull-out load and slip must be modeled separately. Mod-
eling this mechanical component of bond is the subject of a future research.
The validation of the model developed in this part of the study is dem-
onstrated in a second paper, using experimental pullout curves of steel fibers
embedded in cementitious matrices.

ACKNOWLEDGMENTS

This study was supported by a grant F49620-87-C-0063 from the Air Force
Office of Scientific Research to the University of Michigan, Ann Arbor,
Michigan, with Spencer T. Wu as program director. This support is grate-
fully acknowledged. Any opinions, findings, and conclusions expressed in
2788

J. Struct. Eng., 1991, 117(9): 2769-2790


this paper are those of the writers and do not necessarily reflect the views
of the sponsor.
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

APPENDIX I. REFERENCES

Edwards, A. D., and Yannopoulos, P. J. (1979). "Local bond-stress to slip rela-


tionships for hot rolled deformed bars and mild steel plain bars." ACI J. 76(3),
405-420.
Eligehausen, R., Popov, E. P., and Bertero, V. V. (1983). "Local bond stress-slip
relationships of deformed bars under generalized excitations." Report No. XJCBI
BERC-83123, Earthquake Engrg. Res. Ctr. Univ. of California, Berkeley, Calif.
Gao, Y. C , Mai, Y. W., and Cotterell, B. (1988). "Fracture of fiber-reinforced
materials." / . Appl. Math. Phys., 39, 550-572.
Gopalaratnam, V. S., and Cheng, J. (1987). "On the modeling of inelastic interfaces
in fibrous composites." Symp. Proc, Materials Research Society, 114, 225—231.
Gopalaratnam, V. S., and Shah, S. P. (1987). "Tensile failure of steel fiber-rein-
forced mortar." J. Engrg. Mech., ASCE, 113(5), 635-652.
Gray, R. J. (1984). "Analysis of the effect of embedded fibre length on fibre de-
bonding and pull-out from an elastic matrix; Part 1, Review of theories." J. Mater.
Sci., 19(3), 861-870.
Gray, R. J. (1984b) "Analysis of the effect of embedded fibre length on fibre de-
bonding and pull-out from an elastic matrix; Part 2, Application to a steel fibre-
cementitious matrix composite system." J. Mater. Sci., 19(5), 1680-1691.
Mandel, J. A., Wei, S., and Said, S. (1987). "Studies of the properties of the fiber-
matrix interface in steel fiber reinforced mortar." ACI Mater. J., Mar.-Apr., 101—
109.
Mindess, S. (1987). "Bonding in cementitious composites: How important is it?"
Symp. on Cement Based Composites, Materials Research Society, 114, 3-10.
Naaman, A. E., Nammur, G., Najm, H., and Alwan, J. (1989). "Bond mechanisms
in fiber reinforced cement-based composites." Report UMCE 89-9, Dept. of Civ.
Engrg., Univ. of Michigan, Ann Arbor, Mich., Aug.
Namur, G. G., and Naaman, A. E. (1989). "A bond stress model for fiber reinforced
concrete based on bond stress slip relationship." ACI Mater. J., 86(1), 45-57.
Namur, G. G., Naaman, A. E., and Clark, S. K. (1987). "Analytical prediction of
pullout behavior of steel fibers in cementitious matrices." Symp. Proc., Materials
Research Society, 114, 217-224.
Nilson, A. H. (1972). "Internal measurement of bond slip." ACI J., 69(7), 439-
441.
Pinchin, D. J., and Tabor, D. (1978). "Interfacial contact pressure and frictional
stress transfer in steel fiber." Proc, RILEM Symp. on Testing and Test Methods
of Fibre Cement Composites, R. N. Swamy, ed., The Construction Press, 337-
344.
Shah, S.P., and Jenq, Y. S. (1987). "Fracture mechanics of interfaces." Symp. Proc,
Materials Research Society, 114, 205-216.
Timoshinko, S. (1941). Strength of materials. MacMillan, London, U.K.
Wang, Y., Li, V. C , and Backer, S. (1987). "Analysis of synthetic fiber pullout
from a cement matrix." Symp. Proc, Materials Research Society, 114, 159—165.
Wei, S., Mandel, J., and Said, S. (1986). "Study of the interface strength in steel
fiber-reinforced cement-based composites." ACI Mater. J., 83(4); 597—605.
Wise, S., Jones, K., Herzfeld, C , and Double, D. (1987). "Chopped steel fiber
reinforced chemically bonded ceramic composites." Symp. Proc, Materials Re-
search Society, 114, 197-203.

APPENDIX II. NOTATION

The following symbols are used in this paper:

Af = area of fiber;

2789

J. Struct. Eng., 1991, 117(9): 2769-2790


Am = area of matrix;
d = diameter of reinforcement;
Ef = modulus of elasticity of fiber;
Downloaded from ascelibrary.org by "Indian Institute of Technology, Hyderabad" on 09/14/23. Copyright ASCE. For personal use only; all rights reserved.

^m = modulus of elasticity of matrix;


F = tensile force in fiber;
Fb = tensile force in fiber in bonded zone;
Fd = tensile force in fiber in debonded zone;
ft = tensile stress in fiber;
Jm = tensile stress in matrix;
l = fiber length;
n = modular ratio;
P = pullout force;
P„ = bonded force;
"crit = critical pull-out load;
Pa = debonded force;
PP = peak pullout force;
S = local slip between fiber/reinforcement and matrix;
t = interfacial shear flow;
h = shear flow in bonded region;
tf = interfacial frictional shear flow;
'max = maximum allowable interfacial shear flow;
h = maximum interfacial shear flow;
vf = fiber volume fraction;
A = end slip;
A0 = pullout end slip at full debonding;
\ = pullout end slip at peak load;
8 = fiber-matrix misfit;
8/ = local displacement of fiber;
8m = local displacement of matrix;
8D = fiber-matrix misfit at onset of dynamic mechanism;
e
/ = local strain of fiber;
em = local strain of matrix;
•n = factor reflective of steepness of descending branch of pullout curve;
K = bond modulus;
V
f = Poisson's ratio for fiber;
Vm = Poisson's ratio for matrix;
T = shear stress at interface between fiber and matrix;
Trf = bond shear stress at interface between fiber and matrix in decay-
ing frictional zone;
T
/ = maximum frictional bond shear stress at interface between fiber
and matrix;
^rnax = bond strength of interface between fiber and matrix; and
«!» = perimeter of reinforcement.

2790

J. Struct. Eng., 1991, 117(9): 2769-2790

You might also like