Behavior of High Strength Fiber Reinforced Concrete Beams in Shear

Behavior of High Strength Fiber Reinforced
Concrete Beams in Shear

by
Dawood Abdulhai Pandor
B.Eng., Imperial College, University of London
Submitted to the Department of Civil and Environmental
Engineering
in partial fulfillment of the requirements for the degree of
Master of Science in Civil and Environmental Engineering
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 1994
@ Massachusetts Institute of Technology 1994. All rights reserved.
Author ...................... .......... .............

Department of Civil and Environmental Engineering
January, 1994
Certified by.................. .......- . .. .. .. .-..............

Oral Buyukozturk
Professor
Thesis Supervisor
Accepted by ......................•rp"w "r ....

Joseph Sussman
Chairman, Dep on Graduate Students
LCRI I'
Behavior of High Strength Fiber Reinforced Concrete
Beams in Shear
by
Dawood Abdulhai Pandor
Submitted to the Department of Civil and Environmental Engineering

on January, 1994, in partial fulfillment of the
requirements for the degree of
Master of Science in Civil and Environmental Engineering
Abstract
The objective of this thesis is to investigate experimentally the behavior of high
strength fiber reinforced deep concrete beams in shear, in order to arrive at a working
model and equation for predicting the shear strengths of these structural members .
In the experimental program, the volume fraction and aspect ratio of fibers, the
shear span to depth ratio of the specimens, and the maximum aggregate size were
varied in order to determine the effect of these variables on the specimen strength,
on application of four point loading .
In general, it was observed that the addition of fibers led to a significant improve-
ment in the shear strengths of the specimens .
Modification of an existing formula, proposed for the prediction of beam shear
strengths, and the incorporation of a term to account for the contribution from the
fibers to the specimen strength, yielded results which correspond very closely to those
obtained experimentally and in other research programs .
A model for high strength fiber reinforced concrete, based on an existing shear
truss model, was also developed to enable the peak deflections and strengths of these
specimens to be predicted, and this model also yielded very good agreement with the
observed data .
Use of this model therefore represents a useful method for investigating the behav-
ior of high strength fiber reinforced deep concrete beams, and the proposed equation,
after the inclusion of a satisfactory factor of safety, is an especially useful tool for
utilisation in design practise .
Thesis Supervisor: Oral Buyukozturk

Title: Professor
Acknowledgements
I would like to thank the following :
Professor Buyukozturk, for providing the necesssary guidance to enable me to suc-

cessfully complete this thesis
Professor Leung, Thanakorn Pheeraphan, Yiping Geng and Amjad Shahbazker for
providing valuable advice and assistance
W.R. Grace and Co., for providing the materials used in the experimental program .
The American Concrete Institute, for providing partial funding for this research.
My family, for the support and guidance which they have always provided for me .
Contents
1 Introduction 11
1.1 Objectives and scope of the research program . ............ 13
1.1.1 Objectives . . . . . . . . . . . . . . . . . .. . . . . . . . . . . 13
1.1.2 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
1.1.3 Organisation of thesis ....................... 14
2 Review of the shear behavior of concrete beams 15

2.1 Shear behavior of plain concrete ..................... 15
2.2 Shear behavior of normal strength fiber reinforced concrete beams . 17
2.3 Shear behavior of high strength fiber reinforced concrete beams . .. 19
2.3.1 General properties of high strength fiber reinforced concrete (
HSFRC ) . .. .. ... .. .. .. .. ... .. .. .. .. .. 19
2.3.2 Behavior of plain high strength concrete beams in shear . ... 20
2.3.3 Behavior of high strength fiber reinforced concrete beams in shear 23
3 Load-deformation predictions for high-strength fiber reinforced deep

beams 27
3.1 Deep beam shear model ......................... 28
3.1.1 Equilibrium equations ...................... 30
3.1.2 Compatibility equations ..................... 30
3.1.3 Material laws ............................ 31
3.1.4 Simplifying assumptions ................... .. 35
3.1.5 Modifications to existing shear model . ............. 35
3.1.6 Solution procedure ........................
4 Experimental work 44
4.1 Scope . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
4.2 Tests specim ens .............................. 45
4.3 Design of HSFRC beams ......................... 47
4.3.1 Flexural analysis ......................... 47
4.3.2 Shear strength estimation ................... . 50
4.4 Batch design and material selection . .................. 51
4.5 Preparation and casting of specimens . ........... . ..... 53
4.6 Testing procedure ............................. 54
4.6.1 Compression and splitting tensile tests . ............ 54
4.6.2 Beam shear tests ......................... 57
5 Observations and Results 59

5.1 Experimental results ........................... 59
5.1.1 Production ............................ 59
5.1.2 Compression tests ......................... 60
5.1.3 Splitting tensile tests ..................... .. 60
5.1.4 Beam shear tests ......................... 60
6 Discussion of results 73
6.1 Deep beam shear model ......................... 73
6.2 Discussion of proposed equation for shear strength prediction .... . 86
6.2.1 Choice of shear equation ..................... 86
6.2.2 Verification of the applicability of Bazant's formula to HSC
deep beam s ............................ 88
6.2.3 Modification of Bazant's formula . ............... 92
7 Summary, conclusions and recommendations for future work 104

7.1 Summary ............. ...................... 104
7.2 Conclusions ................................ 104
7.2.1 Experimental ........................... 104
7.2.2 Theoretical ............................ 107
7.3 Recommendations for further study . .................. 108
A Computer model 110

List of Figures
2-1 Beam action[5] .................... .......... 16

2-2 Arch action[5] ......... ...................... 16
2-3 Transition layer, containing calcium-hydroxide, around a fiber[7] . . . 18
2-4 Improved post-peak ductility due to fiber addition[17] . ....... 19
2-5 Dependancy of failure mode on shear span ratio[21] . ......... 21
2-6 a)Crack propagation in NC b)Crack propagation in HSC ....... 22
2-7 Effect of Vf on shear capacity[23] . .................. . 24
3-1 Truss model for reinforced concrete element[29] . ........... 28

3-2 Deep beam stress condition[27] ..................... 29
3-3 Stress-strain relationship for steel . .................. .. . 32
3-4 Compressive stress-strain relationship for concrete . .......... 32
3-5 Tensile stress-strain relationship for concrete . ............. 33
3-6 Distribution of compressive stresses for different values of a/d[27] . . 36
3-7 Dowel action idealization: (a)dowel bar, (b) beam on an elastic foun-
dation model, (c)failure condition[33] . ................. 37
3-8 Dowel action at beam-column joint [33] . ................ 37
3-9 Dowel action at beam support ...................... 38
3-10 (a) Bending moment diagram (b)Shear force diagram ......... 40
3-11 Flow chart illustrating model solution procedure . ........... 42
4-1 Sectional drawings ............................ 46

4-2 Stress-strain diagrams for beam cross-section[24] . .......... 48
4-3 Fiber types used and cross sections . .................. 52
4-4 Laboratory cement mixer . . . . . . . . . . . . . . . . . .
4-5 Testing machine .......................
4-6 Testing cylinders capped with hydrostone . . . . . . . . .
4-7 Compressive strength determination . . . . . . . . . . . .
4-8 Tensile splitting test set-up . . . . . . . . . . . . . . . . .
4-9 Beam shear test set-up . . . . . . . . . . . . . . . . . . .
5-1 Split cylinder test for V1 = 0.0 . . . . . . . . . . . . . . .

5-2 (a)Initial crack location in specimens (b)Failed specimen
5-3 Beam tests for V1 = 0.0 (al) ................
5-4 Beam tests for V1 = 0.4 (a2) ................
5-5 Beam tests for V1 = 0.8 (a3) . . . . . . . . . . . . . . . .
5-6 Beam tests for V1 = 1.2 (a4) ................
5-7 Beam tests for a/d = 0.88 (bl) . . . . . . . . . . . . . . .
5-8 Beam tests for a/d = 1.73 (b2) . . . . . . . . . . . . . . .
5-9 Beam tests for hooked end fibers (cl) . . . . . . . . . . .
5-10 Beam tests for straight fibers (c2) . . . . . . . . . . . . .
5-11 Beam tests for maximum d, = 0.25 in (dl) . . . . . . . .
6-1 Action of fibers to (a)improve concrete properties, and (b)perform the

role of stirrups ....................
6-2 al-Model comparison for V1 = 0.0% . . . . . . . .
6-3 a2-Model comparison for V1 = 0.4% . . . . . . . .
6-6 bl-Model comparison for a/d = 0.88 . . . . . . .
6-7 b2-Model comparison for a/d = 1.73 . . . . . . .
6-8 cl-Model comparison for hooked end fibers . . . .
6-9 c2-Model comparison for straight fibers . . . . . .
6-10 Generalised post-peak behavior of specimens . . .
6-11 Beam tests by Ashour[23] . . . . . . . . . . . . .
6-12 Comparison of load-deflection plots for various Vf . .......... 83
6-13 Comparison of load-deflection plots for various V1 . . . . . . . . . .. 84
6-14 Comparison of strength predictions using different formulae[26] . . . 87
6-15 Comparison of Bazant's predictions to Ahmad's results ........ 90
6-16 Comparison of Bazant's to Ahmad's results . .......... . . . 90
6-17 Comparison of Bazant's to Ahmad's results . ............. 91
6-18 Comparison of Bazant's to Ahmad's results . ............. 91
6-19 Normalised experimental results and non-linear optimal curve .... 94
6-20 Predictions for the fiber aspect ratio and effectiveness variation .... 97
6-21 Application of modified equation to tests performed by Shahbazker for
volume fraction variation ......................... 99
Vf = 0.0% . . . . . . . . . .. . .. . . . . .. . . . . . . . .. . .. . 99
Vf = 0.4% . . . .. .. . . ... .. .. .. .. ... .. .. .. ... . 100
Vf = 0.8% . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . .. 100
6-25 Application of modified equation to tests performed by Ashour . . . . 101
List of Tables
II to the specimens tested by Shahbasker

1 .2 Application of eqtns. I and
2.2 Application of Shahbazker's eqtn. to the results obtained by Ashour[23] 26
4.1 Specimen types investigated .......................

4.2 Summary of flexural calculations . . . . . . . . . . . . . . . . . . . .
4.3 Estimated shear capacities . . . . . . . . . . . . . . . . . . . . . . . .
4.4 Mix proportions used ...........................
4.5 OQuantities
,---- ------
used in tv
-,,--
ical castine
---- ,
o ' eration . . . . . . . . . . . . . .
51~ Table of exnerimental results................

5... . .... .. .. .. e u l s.. . . . . . . . . . . . . . . . . . . . ..
6.1 Results from deep beam shear model . . . . . . . . . . .

6.2 Application of model to the results from Shahbazker[23] .
6.3 Application of model to the results from Ashour[23] . . .
6.4 Results from tests by Ahmad[21] and predicted strengths (Method II
discussed in Section 6.2.3) .................
6.5 Results from tests by Elzanaty[22] (fc = 10000 psi) . . .
6.6 Predictions for practical sized specimens using 1987 formula and method
II formula(p = 0.04,fe = 10000 psi,d = 19.6in.) . . . . . . . . . . . . . 93
6.7 Comparison of observed and predicted shear capacities using the mod-
ified equation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
6.8 Shahbazker's test results and modified equation predictions . . . . . . 98
6.9 Ashour's test results and modified equation predictions . . . . . . . . 101
6.10 Comparison of observed and predicted shear capacities using method II 103
Chapter 1
Introduction
Concrete with a 28 day compressive strength of over 7 000 psi (High Strength Con-
crete) has, only over the last 20 years, received serious attention from Structural
Engineers and, indeed, been utilised in engineering applications [1] . Therefore, com-
pared with the research on Normal Strength Concrete(NC) which has occurred over
the last century, as much is not understood about the behavior of High Strength
Concrete(HSC).
One of the first misconceptions to clarify with respect to HSC, is that it is not
an entirely new material [2] . Indeed, HSC can be made by simply reducing the
water/cement ratio of concrete to a suitable level. This may then merit the addition
of superplasticisers in order to improve the workability of the resulting mix.
In order to obtain further increases in strength (over approximately 11 000 psi),
it will then be necessary to add other components, such as silica fume. Further
measures, such as high temperature or pressure curing, and polymer impregnation [3],
and modification of aggregate properties, can be taken to bring about an improvement
in the performance of the material. The basic point, however, is that the essential
components of a concrete mix need not be dramatically altered in order to create a
concrete of higher strength.
A basic understanding for the differences between HSC and NC having been pro-
vided, it is now necessary to examine the advantages and disadvantages to using HSC
in practise.
HSC generally has the advantages, over normal strength concrete, of providing :
1. increased stiffness,
2. reduced axial shortening of members and therefore reduced problems concerning

the maintenance of horizontally supported surfaces,
3. construction time saving, as strength development occurs at an early age,
4. possibility of creating new types of structural supporting systems-an area which

requires further research,
5. economic advantages - the ability to remove formwork rapidly, the increased

availibility of floor area as member sizes can be reduced, and decreased founda-
tion costs as the overall weight of the structure can be reduced.
Other advantages, such as reduced prestressing losses (due to a reduced coefficient

of creep) and a reduction in the quantity of reinforcement required, can also be
incurred. It should be noted that the properties of core samples, removed from
structural members after seven years of service, have indicated that the properties of
HSC remain excellent over time r3].
HSC, however, does have certain drawbacks to its application.
In attempting to reduce element sizes due to the increased strength, attention must
be paid to ensure that the overall member stiffness ( as distinct from the material
stiffness ) is not reduced.
Also, the tensile strength of HSC, while greater than that of NC, does not increase
in proportion to the compressive strength increase. For NC the tensile strength is
usually approximately 10% of the compressive strength, but for HSC, the value may
be as low as 5% [4] .
The most significant drawback with HSC is its ductility, or rather, the lack of
ductility.
HSC demonstrates a very brittle behavior upon reaching the ultimate load, with
very limited post-peak ductility - failure of HSC specimens is usually very dramatic
with little or no warning.
In structural practise it is essential that the ultimate limit state behavior be
such that sufficient warning of impending failure is provided to the occupants of the
structure.
It is also very important for structures to have sufficient ductility to be able to
absorb impact energy, and also to prevent dramatic failure in seismically active areas,
where structures are subjected not only to rapidly varying displacements, but also to
large shear forces. Many of the important metropolitan areas of the world are indeed
located in seismic regions and therefore this consideration must be adressed for HSC.
The use of steel tubes to provide a confining force for concrete has been applied
successfully in practise, most notably in Seattle [1], to improve the ductility of HSC
-columns.
However, a more general method of attaining this beneficial improvement is needed;
a method more easy to implement in practise, and to a wider range of element types.
One possibility is through the use of fiber reinforced HSC, the subject of this
thesis.
1.1 Objectives and scope of the research program
1.1.1 Objectives
The main objectives of this research program are to try to gain a better understanding
for the shear behavior of high strength fiber reinforced concrete beams subjected
to shear loading, and to try to obtain a means of predicting the behavior of such
members.
1.1.2 Scope
In order to satisfy these objectives, it is necessary to divide the research into an
experimental and a theoretical program.
The aims of the experimental program are to investigate the influence on the shear
behavior of HSC beams when the following properties are varied :
1. the volume fraction of fibers ( V1 )
2. the shear span to depth ratio ( a/d )
3. the aggregate size ( d. )
4. the aspect ratio ( Il/d ) of the fibers.
The experimental program allows for an effective investigation of some of the
factors influencing the shear strength of these specimens.
In the theoretical program, the results of these experiments, as well as the results
from previous research studies, will be applied to try to obtain a predictive equation
and model which will enable the shear strength of steel fiber reinforced HSC beams
to be quantified successfully.
1.1.3 Organisation of thesis

This thesis is divided into six main sections :
Chapter 2 provides a brief review and discussion of the literature associated with
this field of study,
Chapter 3 explains the model used in this thesis,
Chapter 4 outlines the experimental program,
Chapter 5 provides the observations and results from this program,
Chapter 6 contains a discussion of the results,
Chapter 7 summarises the overall conclusions from this study.
Chapter 2
Review of the shear behavior of

concrete beams
2.1 Shear behavior of plain concrete
Research over the last forty years has contributed significantly to attainment of a
more complete understanding of the behavior of concrete specimens.
Essentially, failure in concrete beams develops as a result of the interaction be-
tween the applied flexural and shear stresses, cracking occurring when this biaxial
state of stress exceeds the capacity of the concrete.
There are generally two mechanisms for the transfer of shear forces within beams :
a) Beam action (Figure 2-1), and
b) Arch action (Figure 2-2),
the dominance of any one particular type of mechanism being determined by the
a/d ratio of the specimen [5, 6] .
Beam action, dominant at large a/d values when the load cannot be transferred
to the supports by the arch supporting mechanism, is generally considered to result
from the resistance offerred by the uncracked concrete, the longitudinal reinforcement
( dowel action ) and the aggregate interlock mechanism.
Arch action, dominant at low a/d, generally results in the direct transfer of shear
load from the point of application, to the supports. As the support is approached
Compression
zone
1 iE S-r
Figure 2-1: Beam action[5]
P
Line of thrust
.jd d
-L-4
Figure 2-2: Arch action[5]
16
by the load, the depth of the compression zone increases, and thus the mechanism is
facilitated, the horizontal resistance to the opening of the arch being provided by the
longitudinal reinforcement.
2.2 Shear behavior of normal strength fiber rein-

forced concrete beams
In this section, the basic trends which occur when NC beams, reinforced with fibers,
are subjected to shear, are presented.
The application of short, randomly distributed fibers to improve the properties of
building materials is not a new concept, and indeed, its application is recorded in the
scriptures [7].
The major effect of fibers occurs in the post-cracking state, when the fibers bridge
across the cracks that have propogated in the matrix, thereby influencing the ultimate
strength and deformation behavior of the SFRC composite, and its mode of failure.
Fibers generally result in a more uniform redistribution of stresses in the matrix,
and so when these types of specimens are tested, it is generally observed that not only
are the cracks smaller, more numerous, and at a closer spacing, but also spalling of
the concrete is eliminated due to the ability of the fibers to hold parts of the concrete
together in the post-cracking stages [8, 9] .
Fiber addition also results in an increase in the post-cracking stiffness of the
concrete, when the stress-strain behavior begins to deviate from the pseudo-linear
response [10].
In terms of the effect of fiber addition on the ultimate load capacity of fiber
reinforced specimens, some researchers have noted an increase by as much as 30%
[9] , but the majority of reports have shown that fibers have no significant effect on
the shear strength of NC specimens [8, 12, 9] , although the first cracking strength
does improve slightly, indicating an improvement in the tensile strength.
The post-peak ductility is also observed to notably increase on addition of fibers
[8, 9, 12, 11, 13, 14] . This results from the ability of the strain energy to be absorbed
Figure 2-3: Transition layer, containing calcium-hydroxide, around a fiber[7]
during the pullout of the steel fibers from the matrix, resulting from the presence of
a weak interface between fiber and matrix in normal concrete specimens, -due to the
accumulation of a weak calcium-hydroxide layer(see Figure 2-3) at the interface [7].
It has also been observed that fibers also help to improve the contribution from
dowel action to shear resistance, making failures more ductile [11, 15] . This occurs
due to the ability of the fibers to maintain the integrity of the concrete to which the
dowel forces are transferred, thereby preventing the degradation of this shear resisting
mechanism.
As previously mentioned, there are primarily two transfer mechanisms responsible
for shear transfer in beam specimens : beam action and arch action. However, as fibers
do not significantly improve the compressive strength of NC specimens [16, 17] , the
contribution of fibers to arch action is not as significant [12], and the main effectiveness
of fibers is in the beam mechanism ( for larger shear spans ) when the fibers help
bridge across the cracks which propagate in the matrix, and maintain the effectiveness
of the dowel contribution.
10000
8000
Compressive
Stress,
6000
4000
2000
0
0 5000 10000 15000 20000
Axial Strain, millionths
Figure 2-4: Improved post-peak ductility due to fiber addition[17]
2.3 Shear behavior of high strength fiber rein-

forced concrete beams
One of the directions in which the ACI suggested research into HSC be guided, was
towards the incorporation of fibers in the material [18] .
Very limited research efforts have, however, been concentrated on this area of
research, a brief summary of these findings being presented below.
2.3.1 General properties of high strength fiber reinforced

concrete ( HSFRC )
The results of tests carried out on the behavior of HSFRC specimens show that the use
of fibers enables the post-peak behavior to be better investigated [16, 8], as opposed
to plain HSC, where the cracks grow rapidly and unstably(Figure 2-4) .
The inclusion of silica fume in a concrete matrix, to obtain higher strengths, in-
creases the strain corresponding to the peak stress, and the inclusion of steel fibers
provides further increase of this strain. While HSRFC reaches its peak stress at
a compressive strain slightly higher than for lower strength concrete, the ultimate
strain is lower for HSC. A value for ultimate strain of 0.003, specified by ACI com-
mittee 318, could be less conservative for HSC. The presence of steel fibers would,
however, increase this ultimate strain, thereby allowing HSFRC to withstand larger
deformations before failure.
It was also found that if smooth fibers were used and the Vf increased, there was
not a noticeable increase in f., but that if deformed or hooked end fibers were used,
there was a noticeable increase, thus indicating that bond improvement helps bring
about this change [19] .
Change in Vf was also found to be more significant for HSC than for NC, in terms
of increased strength, as fibers tend to yield in HSC as opposed to simple pullout,
due to the improved matrix-fiber interface properties resulting from the addition of
silica fume and the reduction in water content ( silica reacts with the weak calcium-
hydroxide to form a cementitious compound, and as calcium-hydroxide precipitation
is initiated in areas of high water content, the reduced water content causes reduced
precipitation ).
It was observed ( for V1 = 1.25% ) that an increase in f, of approximately 5 % was
obtained [20], and that the modulus of rupture was increased by 67-82 %, compared
with an increase for NC of 50-70 %.
The splitting tensile strengths were also higher than for NC with fibers, and 60 %
higher than for plain HSC.
2.3.2 Behavior of plain high strength concrete beams in

shear
Results from investigations on the behavior of HSC beams in shear have shown that
the failure mode is indeed very brittle [21]
Generally, for a/d <= 2.5, failure occurs by crushing of the arch ribs, while for
P=5.04% p
aid=3
secondar ocal
| Aj:allure atr
lea ds
ilg-,h
AV Iprimary
=5 .04%
aid a local
allure
a•t high load
Figure 2-5: Dependancy of failure mode on shear span ratio[21]
a/d > 2.5, failure was by diagonal tension(Figure 2-5) .

Nilson [22] stated that for high strength beams, aggregate interlock is the first
mechanism to break down, dowel action and shear in the compresion zone providing
the residual shear resisting mechanisms.
If the state of stress meets the failure criterion for the concrete in the compres-
sion zone, an abrupt and sometimes explosive failure occurs. If dowel resistance is
the controlling mechanism, vertical tension in the concrete around the bars causes
splitting cracks along the reinforcement.
Paulay [5] also provides information for NC which is relevant to HSC.
He postulates that at advanced stages of cracking, inclined cracks propagate to-
wards the compression zone and debilitate the cantilever action considerably, resulting
in large rotations at the free end, which may result in exhaustion of the dowel ca-
pacity. When dowel cracks form, a large proportion of the load, for NC, is taken by
aggregate interlock. A reduction in aggregate interlock forces due to some reason,
such as widening of cracks on one side of the cantilever mechanism, causes inbalance
Figure 2-6: a)Crack propagation in NC b)Crack propagation in HSC
and further crack propagation, leading to diagonal tension failure, which is sudden.
Hence, as for HSC the crack surface is smooth due to crack propagation through
the aggregates(Figure 2-6), aggregate interlock forces are minimised and so failure
may follow the aforementioned stages, ignoring the aggregate interlock stage.
It was also observed in previous studies that the reinforcement ratio ( p ) played
a significant role in controlling the type of shear failure.
This is because, not only of the contribution from dowel action to the shear resist-
ing mechanism, but also because an important factor that affects the rate at which
a flexural crack develops into an inclined one is the magnitude of stresses near the
crack tip. The intensity of principle stresses above the flexural crack depends on
the depth of penetration of the crack, and the greater the value of p, the less the
penetration of the flexural crack - The less the penetration of the flexural crack, the
less the principle stress for a given load, and consequently, the greater must be the
shear force to cause the principle stresses that will result in diagonal tension cracking
( This idea could also be extended to an increase in the a/d ratio, which results in an
increased moment at a section, and hence the increased possibility of flexural cracks
with increased crack lengths. Hence, the possibility that these cracks would develop
into diagonal cracks would increase, and consequently, it would be expected that for
increased a/d, the specimen shear capacity would decrease) .
2.3.3 Behavior of high strength fiber reinforced concrete

beams in shear
Review of Ashour
The results of this research indicated the same general relationships between the aid
,atio and the mode of failure previously discussed [21] .
The tests indicated a marked improvement in the post-peak behavior of the spec-
imens, with a significant increase in the ductility. The effect of the fibers on beam
ductility became more pronounced as the a/d ratio increased, and this suipports the
idea [12] that the fibers are more effective in contributing to the beam transfer mech-
anism, than to the arch mechanism.
Most significantly, the results confirmed those noted by Valle [8] , in that it was
observed that the addition of fibers to the specimens caused a noticeable increase in
the shear strength(Figure 2-7).
Two formulae (in MPa) were also proposed for the prediction of the shear capacity
of HSFRC beams :
S= (0.7f +7F)-a +17.2p a (I)

and,
vU = ( 2 .11r + 7F)(pd)0. 33 3 (II)
a
(F = (l/d)Vfdf and d1 is the fiber effectiveness)
When compared with the results obtained by Shahbazker [24] (see Table 2.1 (units
in psi)), it was, however, observed that these give extremely conservative predictions,
for deep beams, and therefore are not satisfactory for practical applications.
see
. $41
-ee
2eei
2310 1/
a
a .d
__e
C 3 43
__
da 0a
__
7a
Defecieo n at seassa in
4N
3 ... 4 e -as
Do, eisc an at asesan (as
Figure 2-7: Effect of V1 on shear capacity[23]
Vf vI(Shabaz.) Eqtn. I Eqtn. II

0 1271.1 721.6 758.0
0.4 1502.2 789.4 835.0
0.8 1608.9 895.6 942.2
Table 2.1: Application of eqtns. I and II to the specimens tested by Shahbazker

Examination of the results of this research, and the methods of analysis adopted by
the authors, indicate that there are two main reasons for the conservative predictions
of these proposed equations :
a) In considering the contribution from the fibers, fiber pull-out was considered
as the dominant mechanism, though verification of this was not provided,
b) Even though flexural failure occurred in many of the specimens, the peak
loads for these specimens was taken as the shear capacity of the specimens, and
these results included in the linear regression analysis used to obtain the final design
equations. Thus, as the results used for the situations where flexural failure occurred
are conservative estimates of the shear capacity of the specimens, it would be expected
that the equations derived would likewise be conservative.
Review of Shahbazker
The results of this research confirm many of the observations made by Ashour [23]
and Valle [8, 40] .
Experiments were also performed in this program incorporating stirrups, and this
was found to result in much greater improvements in ductility, as expected.
The results also seem to indicate that the contribution from fibers and stirrups to
the peak shear stress is not additive, possibly due to a lack of correspondence of peak
strains, the overestimate being of the order of 15 % if the contributions are added
directly.
It should, however, be noted that tests were only performed for a single specimen
in each case ( as indeed were the tests performed by Ashour ) and therefore this
suggests that it would be useful to repeat several of these tests to confirm the actual
quantitative results obtained.
Also, in using Bazant's equation (in psi) [25]:
v= 10V [v + 3000 p/(a/d) ]

1+d/25da
to predict the shear strength of plain HSC, not only is the contribution from dowel ac-
a/d V1 p v,(exper.) eqtn. I ve(Baz.'84) ve(s'baz.)
1 0.5 2.84 1318.1 1394.9 1399.0 1882.0
2 0.5 2.84 698.9 548.1 468.1 951.1
1 1.0 2.84 1847.3 1703.8 1394.0 2100.1
2 1.0 2.84 878.7 662.7 462.9 1169.1
2 1.0 4.58 975.9 765.6 604.8 1432.9
Table 2.2: Application of Shahbazker's eqtn. to the results obtained by Ashour[23]
tion perhaps underestimated, but also the equation has been assumed to not account
for dowel action, which it does.
Examining Table 2.2, when the equation proposed by Shahbazker,
v= 10 r,,[V +3000 p/(a/d)5 + 70F]

1+ d/25da
where F = (1/d)V•V. 5 4 8d is applied to the results of Ashour [23] ( using only those
results in which shear failure was known to have occurred ), it can be seen that the
shear capacities of the sections are dramatically overestimated, suggesting the need
for further modifications to the proposed equation. Modifications to this equation
are discussed in this thesis
Chapter 3
Load-deformation predictions for

high-strength fiber reinforced
deep beams
In attempting to model the behavior of the beams in this program of study, it is

essential to account for their deep nature, as this significantly effects the mode of
shear transfer .
The model adopted in this study is based on the deep beam shear transfer model
proposed by Hsu and Mau [27], as a review of the existing literature confirms the
effectiveness of this model in accurately predicting the shear transfer response of
specimens [8, 28, 29, 30]
The aim in utilising this model was to enable :
* the peak load capacity for HSFRC beams to be predicted,
* the central span deflection behavior to be predicted, and
* to validate the application of the softened truss model to HSFRC beams.

P}
----
,T.s
Qtc
tC
L11C
'------
= P fI
~ vLIIIi$ rc- Crs

-lc z Pft
T 7
Reinforced
Concrete Concrete Reinforcement
Crr 1•d
V.
I
I
Concrete
-
Figure 3-1: Truss model for reinforced concrete element [29]
3.1 Deep beam shear model

The deep beam shear model has developed from the softened truss approach ( which
relies on the ability to characterise the softening behavior of concrete ), and the
Modified Compression Field Theory [31, 32] .
The principle attraction of this model is that, in addition to not relying on the
assumption of a pre-existing failure plane, it also introduces an effective transverse
compression in the beam web, thus accounting for the arch transfer mechanism.
It is assumed that when the concrete cracks on application of load, the concrete
struts so formed act in conjunction with the longitudinal and transverse reinforce-
ment to form a truss-like mechanism (Figure 3-1) which resists the shear loading (
It should therefore be noted that for small loadings, when the concrete is uncracked,
the predictions of this model will be inappropriate ) .
A key to the variables is provided below :
sym.
t ½ T
d dv
d'
I I
V V
Figure 3-2: Deep beam stress condition[27]
a = normal stress
7 = shear stress
f = stress in steel
p = reinforcement ratio
a = the angle of inclination of the d axis with respect to the x axis (positive counterclockwise'
The subscripts z, y, d, and r correspond to the chosen axis .
Though no transverse reinforcement ( stirrups ) have been provided in this research

program, the truss mechanism is still valid as it is expected that the fibers will be
effective in performing the role of stirrups .
It is assumed that the moment in the concrete section is taken by the couple
formed by the steel reinforcement and an equivalent depth of concrete d' ( where d'
= half the depth needed to provide a compressive force in concrete to balance the
tension in the bottom steel at yield (see Figure 3-2)) .
The shear load is then resisted by the concrete depth d,, where :
d, = d - d'
Essentially the truss mechanism formed must satisfy the laws of compatibility and
equilibrium, and the material laws .
3.1.1 Equilibrium equations

After cracking, the principle stress directions are denoted by d and r (compressive
and tensile respectively - see Figure 3-1) and the angle of inclination of the principle
axis, a .
Hence, using the Mohr's circle stress transformations, and assuming that the steel
reinforcement can take only longitudinal stresses, the following equilibrium equations
are obtained :
O, = ad Cos 2 a + -, sin 2 a + (3.1)
a, = ad sin2 a + , Cos 2 a + pfI (3.2)
r7, = (ad - ar) sin a cos a (3.3)
( p, is the equivalent area of fiber steel across the shear plane and it should be
noted that the fibers are considered to be only effective in the y-axis direction as the
contribution to the x-axis direction is negligible compared to the area of longitudinal
steel )
3.1.2 Compatibility equations

Again using the Mohr's circle stress transformations, and utilising the result from the
Modified Compression Field Theory that the principle stress and strain inclinations
correspond to the inclination of the concrete struts formed, the following equilibrium
results are obtained :
Ex = d cos 2 a + esin2 a (3.4)
ey = ed sin2 a + e, Cos 2 a (3.5)
^Ny = 2(Ed - e) sin a cos a (3.6)
where :
ex, Cy = average normal strains in the x and y directions (tension positive), respectively,
7.y = average shear strain,
Ei,E, = average principle compressive and tensile strains respectively
3.1.3 Material laws

Steel reinforcement
The longitudinal bars, and the steel fibers, are assumed to demonstrate elastic-
perfectly plastic stress-strain characteristics, as shown in Figure 3-3 (the high bond
strength for steel fibers in HSC allows this assumption to be made for the fibers,
even though in reality some fiber debonding will occur before the yield strength is
reached) .
f, = E, , < fy (3.7)
where E, and f, are the Young's Modulus and the yield stress of the steel rein-
forcement respectively .
Concrete
The principle compressive stress-strain curve for HSFRC, which incorporates the
softening effect due to the biaxial state of stress, is shown in Figure 3-4 :
The ascending portion is represented by :
rd = f[ 2(e ) - A(d)2 ] (3.8)
for Ed Ep,
and is the same as for NC, as there is essentially no sigificant change when fibers
are added .
Figure 3-3: Stress-strain relationship for steel
fc.
0
Ed
Figure 3-4: Compressive stress-strain relationship for concrete

Er x10 "3)
Figure 3-5: Tensile stress-strain relationship for concrete
For the descending portion :
L ((3.9)
CO=)2 i
for Ed _> EP
where E, = Eo/A, and lambda is a coefficient to account for the softening phe-
nomenon, where :
A = /0.7- (3.10)
Ed
This is again the same relation as used for NC because, due to the addition of
fibers, it is expected that the post-peak behavior for HSC would be improved only to
the extent that the ductility would be comparable to that for NC .
The assumed principle tensile stress-strain relation was also taken to consist of
two portions : a pre- and a post-cracking behavior (Figure 3-5) :
Before cracking, the ascending portion is given by :
(, = eE, (3.11)
for E,. •c,

where Ec is the Young's Modulus of concrete and ec, is the cracking strain of fiber
concrete,
f, = 7.5 /
Ec = 40000VI + 1.0 x 106
ACI 363[35] notes that for, curing conditions followed by air drying, the use of
7.5Vff is probably fairly close to the full strength range for high strength concrete .
The post-cracking behavior is given by :
r, =
fc. + (ptu) (312)
(3.12)
1+3
= 0050 (3.13)
for E, > fce
This post-cracking behavior occurs because, after cracking, the stress across the
cracks is still quite significant due to the presence of fibers . This stress across the
cracks is denoted by ate, where,
77%V
t771 l 7,r
2r'
r' is the ratio of the area of cross section to the perimeter of the fiber .
The ultimate bond strength of the fibers, r,, has been calculated to be the bond
stress at yielding of the fibers, to account for the higher bond properties of HSC .
3.1.4 Simplifying assumptions
At this stage, there exist 14 unknowns but only 11 equations, and therefore it is
necessary to make several simplifying assumptions to enable a solution to be obtained.
As the depth d. is assumed not to carry any flexure, then it can be assumed that :
Y= 0
Also, Hsu and Mau [29] have shown that :
at = Kr
where K is the ratio of maximum transverse stress to maximum shear stress, and is
given by :
d, h 4 2 a
h a 3 3 h
for 0.5 < a/h < 2

This value takes into account the variation in the distribution of the stresses
within the shear span, when the magnitude of the shear span is varied, as shown in
Figure 3-6 .
Hence, making these two assumptions, for a given value of Ed ( chosen because it
is expected to increase monotonically), the 14 unknowns can be obtained, as outlined
in Section 3.1.6 .
3.1.5 Modifications to existing shear model

Dowel contribution
As the proposed model by Hsu [29] does not account for the dowel contribution from
the longitudinal steel, and does not account for aggregate interlock, it is necessary to
modify the modify the model to attain a higher degree of accuracy .
For HSC, when cracks propagate, they generally pass through the aggregate, as
this offers the path of least resistance . Thus the crack surface is generally quite
a 4 a h/2
'/ \- ,, Distribution
/
of Transverse
Compression
-j,
h/2 , / h/2 .. - Isostatic
/ I -
Compressive
S / Curve
(a) a/h = 0 (b) a/h = 0.25 (c) a/h = 0.5
a=h a = 2h
8 - ,
I
i~-/
(d) a/h = 1 (e) a/h = 2
Figure 3-6: Distribution of compressive stresses for different values of a/d[27]
smooth as there are no protruding aggregate particles .

Thus, aggregate interlock shear transfer is nominal [35], and so will be ignored in
this study.
It is, however, expected that the dowel contribution will be an important shear
transfer process as the area of longitudinal steel is quite high, and so will be incorpo-
rated in this model.
The formulation [33] used to calculate the dowel contribution is based on the
'beam on elastic foundation theory' (Figure 3-7) .
Failure, in this model, is assumed to occur when a plastic hinge is developed at
the section of maximum moment in the bar, by which point a crushed zone of length
c has developed from the crack face under the longitudinal bars .
It should be noted that this approach has been developed for typical beam-column
joints, as shown in Figure 3-8, but is nevertheless valid for this experimental study,
as at failure the shear crack propagates close to the support, which in turn performs
the function of the column ( Figure 3-9 ) .
-,, ý
DU
% BEARINC.
STRESS
T PLASTIC fINwE
:BENOD '4
41t:NT
f~
Figure 3-7: Dowel action idealization: (a)dowel bar, (b) beam on an elastic foundation
model, (c)failure condition[33]
dowel action at
beam-col. joint
Figure 3-8: Dowel action at beam-column joint[33]

dowel action
of bars
support
Figure 3-9: Dowel action at beam support
The theory would not therefore be as appropriate for cracks developed away from
the supports of a beam .
Taking moment equilibrium about the plastic hinge (see Figure 3-7) results in the
following expression for the ultimate dowel contribution :
DU = 0.5 fb (0.37 - db - c)2 + 0.45 f, d2 (1- T /TI)/T
where fb, the bearing strength of the concrete underneath the dowel bar, is given by :
fb = 154 fl/ db
and,
db = dowel bar diameter,
Kf = concrete foundation modulus (1 x 106 psi),
7 = E, /Kfdb,
T = dowel bar axial force,
T, = dowel bar yield axial force
It is then necessary to use emperical relations in order to calculate the dowel load
for a given shear displacement .
The deflection at ultimate load is given by:
S, = 4.26 x 10- 6
D, + 0.00945
and the load D for a given deflection S, is given by :
D= D.• S/ (3.14)
for S < Su, and:

S-S,
D = D, - D, 0.4 - 0.4D0 (3.15)
0.4/db - Su
for S Su
Flexural deflections
By virtual work, the central deflection of the beam is given by :
601.6 P
6 Ecle
where P is the load applied at each of the four load points (Figure 3-10), and Ie, the
effective moment of inertia, is given by :
Me)3 M[
I = (M)3 Ig+ [1 - (
Ma Ma)3] Icr
-P.a -P.a
a)Beding moment diagram b) Shear force diagram
a) Bending moment diagram b) Shear force diagram
Figure 3-10: (a)Bending moment diagram (b)Shear force diagram

Mc, = moment at first cracking,
Ma = applied moment at the section,
Ig = gross moment of inertia of the section,
I, = moment of inertia of the cracked transformed section
It should be noted that the stiffening effect of the fibers has not been included in
the calculation of I, and therefore the predicted deflections are expected to be higher
than the practical values obtained, for larger loads, when the fibers are effective in
bridging the cracks which are developing .
The model used in this study therefore differs from the one used by Valle[8] in
that the greater shear span ratio used has led to the inclusion of a term to account
for the variation in stress within the shear span .
Also, this model includes a contribution from dowel action , as a much greater
area fraction of longitudinal steel has been used . Central deflections have also been
predicted, as opposed to shear strains in the study by Valle .
In addition, for the contribution from the fibers in the material laws, alternative
expressions have been used in some cases in order to try to improve the predictions of
peak strength, which in the case of Valle's work, tended to be overpredicted slightly
. The fibers have also been taken as having a contributing effect to the transverse
steel, and thereby effecting an increase in the shear strength of the beams
3.1.6 Solution procedure
It is necessary to use an iterative procedure in order to solve for the required un-
knowns, and this iterative procedure is outlined below :
1. Choose avalue of ed,
2. Assume a value for e,,
3. Find ad and c-, from the stress-strain relations,
4. Solve for a from eqtn. ,
5. Solve for the updated value of a, from the equilibrium eqtns.,
6. Compare the values of a, calculated from steps 3 and 5; if there is a significant

difference, repeat steps 2-6,
7. Calculate the other desired unknowns - for this case r,, and y, .
A simple flow chart may be used to illustrate the solution procedure, as shown in
Figure 3-11 .
Hence, in this manner, for different values of ed, the shear behavior of the beam
can be traced .
The shear deflection of the beam can then be calculated,
86 = 7,/, d,
eed
Figure 3-11: Flow chart illustrating model solution procedure
42
and consequently, from eqtns. 3.14 and 3.15, the dowel contribution is obtained,
enabling the total load acting on the specimen to be calculated :
Total load = r7,, d b + D
Finally, the overall deflection is simply the sum of the flexural and shear deflec-
tions,
ST = S, + 6f ,
and hence the overall load-deflection behavior is obtained

A computer program to perform these operations was written in the 'C' program-
ming language, and is provided in the appendix .
Chapter 4
Experimental work
4.1 Scope
The variables investigated in this program were :
1. Vf of fibers,
2. 1/d ratio of fibers,
3. a/d ratio of the specimens,
4. size effect of the aggregate,
5. fiber effectiveness .
A total of 22 laboratory size specimens were cast, and the compressive strength
developed was approximately 10 000 psi .
In the remainder of this thesis, the following key will be used to represent the
specimens :
Vf - (a/d) - (lf/d) - d,
where da = maximum aggregate size, and the other terms are as defined previously .
A summary of the specimen types is provided in Table 4.1 .
Specimen Specimen type No. cast
al 0.0-1.46-28.1-0.375 2
a2 0.4-1.46-28.1-0.375 3
a3 0.8-1.46-28.1-0.375 3
a4 1.2-1.46-28.1-0.375 2
bl 0.8-0.93-28.1-0.375 2
b2 0.8-1.73-28.1-0.375 2
cl 0.8-1.46-61.5-0.375 2
c2 0.8-1.46-50.0-0.375 2
dl 0.8-1.46-28.1-0.250 2
d2 0.8-1.46-28.1-0.090 2
Table 4.1: Specimen types investigated
4.2 Tests specimens

Figure 4-1 indicates the dimensions of the beams which have been used in this inves-
tigation .
The formwork used allowed two beams to be cast simultaneously, and was made
from plexiglass (Figure 4-1)
Only tension steel was provided (See Section 4.3), and this consisted of 2 No. 5
and 1 No. 4 bars.
Stirrups were provided outside of the loaded span(Figure 4-1) in order to maintain
the rebars in the correct location with the appropriate nominal cover requirement .
They therefore played no part in the shear transfer process, as it was felt necessary
to first try to understand, and predict, the behavior of specimens without transverse
reinforcement .
7
Bar No. Size Length
1 5 20
2 4 20
/.
' 03
,
30
S.v
Section A-A
stirrup outside shear span

Notes :
/ a) All dimensions in inches
b) Drawings not to scale
c) Cover to main reinforcement
equals 0.44"
I
VZW
\1
Side elevation
Figure 4-1: Sectional drawings

4.3 Design of HSFRC beams
4.3.1 Flexural analysis

The first stage in the flexural design of the beams, is the calculation of the balanced
steel ratio, Pb :
Effective section size = 3" x 3.75"
As = (7r/4) [2 (5/8)2 + (4/8)2)] = 0.810 in. 2
0.810
3 x 3.75
So,
10000 0.003 E,
Pb = 0.65 x 0.85 x - x = 0.0708
60000 0.003 E, + f,
Hence, as the beams are over-reinforced, only the balanced steel ratio will be used
in the flexural calculations .
It is also advisable to provide this additional steel, because shear stresses will cause
longitudinal stresses to develop in the steel, and so part of the steel area provided
will not be available to resist flexure .
For the principle fibers used in this study, of square cross section,
fiber length, lj = 1"
breadth = 0.045"
depth = 0.011"
Therefore, the effective diameter of the fiber, df, may be calculated :
rx df = 2 x (0.045 + 0.011)
hence,
df(effective) = 0.0357
and,
Ifl/df = 1/0.0357 = 28.1
Figure 4-2 summarises the simplifications made in estimating the flexural capacity
Assumed
Beam Stress Strain
Section Distribution Diagram
b 0o.s85 o.oo0
I 1 C
--- 4 --
h
TI r F{
LT.I
Ixi i 1j jI
..- _.-.. •;
•*
NeutralAxIS
............
A strain in rebars
".......
T2 - tension in librous concrete
Figure 4-2: Stress-strain diagrams for beam cross-section[24]
of the beams .
The flexural capacity of a normal fiber reinforced beam is given by [17, 34] :
a h e a
M, = A,.f, (d- 2) + .b( 2 2
where aft is the tensile strength of the fiber concrete, and is given by :
at = 0.41 x 0.82 x rd (1Ild) V1 Fb,
where Fbe is the fiber efficiency factor, taken as 1.2 for crimped fibers, and rd is the
dynamic bond stress of the fibers .
This equation has also been assumed to be valid for HSC [35], the improved bond
strength being accounted for in the value of rd used (600 psi) .
Now :
((e±+
0.003)
0.003
where ef is the strain in the fiber.
Hence,
(g + 0.003)
0.003
ao is the tensile stress in the fibers during pullout (for normal concrete), and is
obtained by equating the tensile stress in the fibers to the bond stress :
S=
Td Fbe if (bf + wf)
w=
(Note that the fibers are of square cross section, the dimensions being given by w1
and bf .)
This value is independent of the volume fraction of the fibers and may be calcu-
lated at this stage :
600 x 1.2 x 1 x (0.045 + 0.011)

0.045 x 0.011
As the yield strength of the fibers is 60 000 psi, then the above result would tend to
indicate that the fibers will actually yield before pull-out occurs .
Thus af = 60 000 psi.
Now,
_ 60 000
e -- 2.069 x 10- 3
E, 29 x 106
therefore,
0.002069 + 0.003
e= c = 1.69 c
0.003
but a = 0.65 c, so :
e = 2.60 a
The value of a (see Figure 4-2) is determined by equating the compression in the
concrete section to the tension, and the resulting expression may be expressed as :
A, f, + aft b h
0.85 fc b + 2.6 oft b
Specimen Moment capacity(lbin) Total load(kips)
0.0-1.46-28.1-0.375 134374 48.80
0.4-1.46-28.1-0.375 134918 49.06
0.8-1.46-28.1-0.375 135224 49.16
1.2-1.46-28.1-0.375 135529 51.28
0.8-0.93-28.1-0.375 135224 77.27
0.8-1.73-28.1-0.375 135224 41.60
0.8-1.46-61.5-0.375 135952 49.44
0.8-1.46-50.0-0.375 135339 49.21
0.8-1.46-28.1-0.250 135224 49.16
0.8-1.46-28.1-0.090 135224 49.16
Table 4.2: Summary of flexural calculations
The flexural capacity may then be determined using the aforementioned expression
for M,, and consequently the ultimate load capacity obtained .
A summary of the results of the calculations for the different specimen types is
provided in Table 4.2
4.3.2 Shear strength estimation

As the shear capacity of the members is the focus of this study, it is essential that
the shear capacity be less than the flexural capacity of the members .
Even though the formula proposed by Shahbazker [24] needs further modification,
it is nevertheless used to obtain an estimate of the shear capacity of the members.
V= 10" [-/ + 3000 p/(a/d)5 + 70 F]

1+d/25da
and,
VU = Ve + Vd
where :
vd ~ 0.25 v,
A summary of the estimated shear capacities for the different specimens is pre-
Specimen Vc (Ibs) V, (lbs) Total load (Ibs)
0.0-1.46-28.1-0.375 16.12 21.49 42.98
0.4-1.46-28.1-0.375 18.94 25.25 50.50
0.8-1.46-28.1-0.375 20.24 26.99 53.98
1.2-1.46-28.1-0.375 21.26 28.34 56.68
0.8-0.93-28.1-0.375 45.69 60.92 121.84
0.8-1.73-28.1-0.375 16.03 21.37 42.74
0.8-1.46-61.5-0.375 25.13 33.51 67.02
0.8-1.46-50.0-0.375 21.01 28.01 56.02
0.8-1.46-28.1-0.250 18.93 25.24 50.48
0.8-1.46-28.1-0.090 14.67 19.56 39.12
Table 4.3: Estimated shear capacities
sented in Table 4.3 (all loads in kips):

Hence, it is expected that shear failure will occur before flexural failure for the
specimens under investigation as the flexural calculations are very conservative .
4.4 Batch design and material selection

The materials used in the experimental program are outlined below :
* Type I Portland Cement,
* Fine aggregate ( mortar sand ),
* Coarse aggregate ( pea gravel - 3/8"max. size ),
* Condensed silica fume ( Force 10 000 - 50% water by weight ) to enable a

stronger between fiber and matrix and a satisfactorily high strength to be ob-
tained,
* Superplasticizer ( WRDA-19, ASTM C-494, type A & F ), used to increase the

workability of the mix with no loss in strength,
* Steel fibers (ff, = 60000 psi) - 3 types were used (Figure 4-3) :
I/d = 28.1 I/d = 50.0 li/d = 61.5
Figure 4-3: Fiber types used
1. deformed fibers, lid = 28.1,1" x 0.011" x 0.044",
2. hooked end fibers, Ild = 61.5, 1.2" x 0 = 0.0195",
3. straigth fibers, l/d = 50.0, 1" x 0.022" x 0.01" .
The mixing proportions used to prepare the specimens are given in Table 4.4 , the
mix being designed to obtain a strength of 10 000 psi at 28 days .
The quantities used in a typical mixing operation are provided below in Table 4.5
Variable Proportions
sand/cement 2.0
gravel/cement 2.0
silica 10%
superplasticiser 3%
water/(cement + silica) 0.36
Table 4.4: Mix proportions used

Material Quantity (lbs)
cement 13
sand 26
gravel 26
silica fume 2.6
water 3.9
superplasticiser 0.39
fibers 1.75
Table 4.5: Quantities used in typical casting operation
4.5 Preparation and casting of specimens

The rebars were first assembled into a reinforcing cage utilising stirrups (outside the
shear span), and these were all held together with steel wire .
The plexiglass formwork was then coated with bondbreaker, and the reinforcing
cages placed inside the formwork .
The concrete was then prepared following the procedure outlined below :
1. The sand, cement and gravel were placed in a rotating-drum mixer(Figure 4-4)
and mixed for 30 seconds,
2. At this stage, the water, silica fume and superplasticizer (which had all been
previously mixed together) were then gradually added to the mixer,
3. If fibers were used, these were 'sprinkled' into the mixer, ensuring that they did
not ball together as they were added,
4. The mixing was then discontinued 5 minutes after step 2 had been completed.
The concrete was then placed in the formwork and, in addition, 4 test cylinders
were cast.
The concrete was cast according to ACI code requirements [36, 37] and the spec-
imens were then vibrated for 3 minutes .
The specimens were then covered in plastic sheeting, and left in this state for 24
Figure 4-4: Laboratory cement mixer
hours in order to allow setting of the concrete to occur . They were then demoulded
and cured in water for 28 days [35]
4.6 Testing procedure

A 200 kip capacity 'Baldwin Testing Machine' was used for all the tests performed
in this experimental program, and the signals from this machine were recorded using
an IBM AT computer (see Figure 4-5) .
4.6.1 Compression and splitting tensile tests

Of the four cylinders prepared during each casting sequence, two were used for for
compression tests, and two for splitting tensile tests
Prior to testing, the cylinders for the compression tests were capped using hy-
drostone, in order to ensure an even contact surface with the loading platens of the
testing machine (Figure 4-6).
Figure 4-5: Testing machine
Hydrostone to
allow even contact
surface
3 in. diam.
x 6 in. length
Figure 4-6: Testing cylinders capped with hydrostone

Figure 4-7: Compressive strength determination
The compression test program consisted of two main stages :
1. The Young's Modulus of the cylinders were measured by placing two Linear
Voltage Displacement Transducers (LVDT) on diametrically opposite sides of
the cylinders being tested, and then loading the cylinders up to approximately
40% of the expected ultimate capacity, at a displacement rate of 0.016 in/min. .
2. The LVDT's were then removed, and the cylinders reloaded at a displacement
rate of 0.016 in/min. to enable the ultimate compressive strength, f£, to be
determined (Figure 4-7).
In performing the splitting tensile tests, the cylinders were placed flat on their
longitudinal axis (Figure 4-8), and loaded until the occurrence of vertical cracking,
which was readily determined from the load-displacement plot, as loading occurred .
The displacement control rate used was 0.016 in/min.
The results from these splittimg tensile tests were not utilised in this experimental
program, but were only performed in order to provide a measure of the tensile strength
Figure 4-8: Tensile splitting test set-up
and its variation for the specimens
4.6.2 Beam shear tests
The beam shear tests were performed using four point loading . For shear tests on
beams, ACI 318 [36] recommends that the beams be tested soon after removal from
water, to avoid the development of shrinkage cracks and a consequent reduction in
the tensile strength of the specimens . Therefore, all the tests performed on the
specimens (including the cylinders) were performed on the same day of removal from
water .
In order to measure the central deflection of the specimens, an LVDT was used,
located as indicated in Figure 4-9 .
The supports used were designed to allow for rotation at the support points and
thereby maintain a simply supported configuration .
The displacement control rate used was 0.04 in/min., and loading was continued
load
stiff I-beam
LVDT HSC beam
Notes:
3.0 a
a) All dimensions in inches
21.0
b) Drawing not to scale
Figure 4-9: Beam shear test set-up
beyond the ultimate capacity in order for the post-peak behavior to be observed.
Chapter 5
Observations and Results
5.1 Experimental results
5.1.1 Production
Initially, concrete with a strength in the range of 13 000 psi was attempted to be
produced, however, when the mix design used by Shahbazker [24] to achieve such a
strength was utilised, satisfactory workability was not achieved .
The probable reason for this was that the mixer used by Shahbazker was more
effective in terms of the way in which the concrete was mixed, and also because the
concrete could be pressurised, allowing for a more homogeneous mix .
In this program, the mixer available for use was of the rotating drum, gravity
induced type (Figure 4-4), which is not as effective, and therefore it was necessary to
use a higher water/cement ratio in order to achieve a satisfactory workability . This
resulted in a concrete compressive strength of approximately 10 000 psi .
Addition of fibers to the mix did not have an adverse effect on the mix, the
reduction in workability not being very significant .
The cement used, however, provided some mixing problems . This was because,
on using different bags of cement, it was found that there was a noticeable change
in workability . For the cases with diminished workability, mixing was allowed to
continue for longer period of time .
5.1.2 Compression tests
Table 5.1 summarises the results from the compression tests on the cylinders . It
is seen that the values of Young's Modulus are greater than those expected for NC,
confirming the trend of increased stiffness when using HSC .
It was also observed that when the V1 of fibers is increased there is a coresponding
increase, not only in the Young's Modulus, but also in the compressive strength of
the specimens, and this confirms the beneficial effects of the fibers when added to the
mixture.
5.1.3 Splitting tensile tests

The results from the splitting tensile tests (Table 5.1) indicate that, as the V1 in-
creases, so also does the tensile splitting strength, and this is due to the inherent
ability of the steel fibers to carry tensile stress .
It was observed for the case with V1 = 0.0 that the cylinders split completely into
two halves at failure (Figure 5-1)
Figure 5-1 also indicates that the resulting surfaces were also very smooth and
this is because generally in HSC, the cracks propagate directly through the aggregate
due to the improved aggregate/matrix bond .
When fibers were added, the occurrence of failure was not directly observable, as
the specimens remained intact . Failure in those cases was determined by careful
observation of the load/displacement plot as loading was continued .
5.1.4 Beam shear tests

As loading progressed, it was observed that there were no visible flexural cracks
formed at the midspan of the beams .
Diagonal cracking began at approximately 20000 lbs for the specimens with a/d =
1.46 . These cracks began at mid-depth of the beams, as shown in Figure 5-2, and
then propagated towards the supports .
As the cracks approached the supports, they did not continue towards the center
Figure 5-1: Split cylinder test for V1 = 0.0
Specimen fY(psi) ft(psi) Estimated 2YV 2V,(lbs) Ec(ksi)

al 9720.9 406.4 42980 37585.5 3655.97
a2 10157.9 564.0 50500 40993.4 3706.20
a3 10387.4 592.3 53980 42377.8 4368.07
a4 10774.5 725.8 56680 49225.7 3818.03
L bl 9052.6 658.1 121840 74443.6 3858.30
b2 10258.9 612.4 42740 35953.9 4470.64
cl 11695.3 697.6 67020 50691.1 4555.66 1
c2 7859.6 615.0 56020 38237.1 4006.31
dl 9587.1 690.9 50480 40379.1 3890.88
L d2 7154.5 640.5 39120 30980.6 30549.87
Table 5.1: Table of experimental results

Figure 5-2: (a)Initial crack location in specimens (b)Failed specimen
of the supports, but instead towards the closer edges of the supports, as this allowed
for large shear displacements (Figure 3-9) .
It should also be noted that, as observed from the load-deflection diagrams, there
is not a significant decrease in stiffness upon the occurrence of first cracking at ap-
proximately 20000 lbs . This is because the system under study involves deep beam
behavior, with the diagonal cracks initiating close to the neutral axis, and therefore
the cracking has nominal effect on the deflection of the specimens .
If the study had involved slender beam behavior, the occurrence of first cracking
would have probably been more evident from the load-deflection curves . This is
because, for slender beams, shear diagonal cracks generally develop as extensions of
flexural cracks[5] i.e. they develop at the outer tensile fiber (at the bottom of the
specimens in four point loading) . Thus, the reduction in total effective moment of
inertia on first cracking of the specimenns would lead to a reduced overall flexural
rigidity of the specimen, and, therefore, a more noticeable increase in the deflection of
the beam . The fibers present would, however, help to limit the increase in deflection .
As loading continued beyond the peak load, longitudinal cracks developed along
the main reinforcement, indicating that the dowel capacity was being reached .
Also, as the tests progressed into the post-peak range, the relative displacement
between the sides separated by the crack (Figure 3-9) noticeably increased .
The plots for the specimens tested are given in Figure 5-3, Figure 5-4, Figure 5-5,
Figure 5-6, Figure 5-7, Figure 5-8, Figure 5-9, Figure 5-10, and Figure 5-11 .
It should be noted that for the specimens with V1 = 0.0, on reaching the peak
load, failure was brittle (Figure 5-3), with little post-peak ductility, and failure was
accompanied by the emission of a loud noise .
For the specimens with fibers, there was a noticeable increase in the post-peak
ductility (Figure 5-4 to Figure 5-11), however, for V1 = 1.2%, the increase in ductility
was not as significant as was expected (refer to Chapter 6) .
The results also confirm the beneficial effects of adding fibers to the concrete,
because as seen in Table 5.1, as the V1 is increased, there is a corresponding increase
in shear strength of the specimens .
The fibers also helped the specimens to maintain their integrity as the tests pro-
gressed (after reaching the peak load), wheras, in the cases with V1 = 0.0, pieces of
concrete tended to spall off.
For the specimens tested with various shear spans, it is seen that the smaller the
a/d ratio, the greater is the increase in shear strength, as expected .
For the specimens tested with hook ended fibers, it should be noted that after
the ultimate load was reached, continual 'clicking' sounds were heard, a phenomemon
which was not experienced in the other tests, and this was probably the result of the
fibers being pulled out .
It should also be noted that, of the fibers tested, these hooked ended fibers, with
the greatest aspect ratio, caused the greatest increase in shear strength.
In all cases it was noted that for large displacements, the load-displacement plots
eventually levelled out (Figure 5-3 to Figure 5-11), and this load corresponded to the
residual dowel capacity of the specimens .
40000
20000
0.0 0.1 0.2 0.3 0.4
Deflection (in)
Figure 5-3: Beam tests for V1 = 0.0 (al)

40000
20000
0.0 0.1 0.2 0.3 0.4
Deflection (in)
50000
40000
30000
20000
10000
0
0.0 0.1 0.2 0.3 0.4
Deflection (in)
Figure 5-4: Beam tests for Vf = 0.4 (a2)

0UUUU
40000
30000
20000
10000
0
0.00 0.05 0.10 0.15 0.20 0.25
Deflection (in)
50000
40000
30000
20000
10000
0
0.00 0.05 0.10 0.15 0.20 0.25
Deflection (in)
Figure 5-5: Beam tests for V1 = 0.8(a3)

50000
40000
30000
20000
10000
0
0.0 0.1 0.2 0.3
Deflection (in)
50000
40000
30000
20000
10000
0
0.00 0.05 0.10 0.15 0.20 0.25
Deflection (in)
Figure 5-6: Beam tests for V1 = 1.2 (a4)

80000
60000
40000
20000
0
0.0 0.1 0.2 0.3
Deflection (in)
80000
60000
40000
20000
0
0.0 0.1 0.2 0.3
Deflection (in)
Figure 5-7: Beam tests for a/d = 0.88 (bl)

40000
20000
0.0 0.1 0.2
Deflection (in)
40000
20000
0
0.00 0.05 0.10 0.15 0.20 0.25
Deflection (in)
Figure 5-8: Beam tests for a/d = 1.73 (b2)

OVUUU
50000
40000
30000
20000
10000
0
0.0 0.1 0.2 0.3
Deflection (in)
OUUUU
40000
20000
0
0.0 0.1 0.2 0.3
Deflection (in)
Figure 5-9: Beam tests for hooked end fibers (cl)

4UUUU
20000
0
0.00 0.05 0.10 0.15 0.20 0.25
Deflection (in)
40000
20000
-0.05 0.00 0.05 0.10 0.15
Deflection (in)
Figure 5-10: Beam tests for straight fibers (c2)

40000
20000
0 I
0.00 0.05 0.10 0.15 0.20 0.25
Deflection (in)
50000
40000
30000
20000
10000
0
0.00 0.05 0.10 0.15 0.20 0.25
Deflection (in)
Figure 5-11: Beam tests for maximum d. = 0.25 in (dl)

Chapter 6
Discussion of results
6.1 Deep beam shear model

In order to account for the noticeable increase in the shear capacity for the HSFRC
specimens in the deep beam shear model, it was first necessary to make two reasonsble
assumptions regarding the contribution from the steel fibers :
1. The fibers help to improve the inherent strength of the concrete in resisting shear
loading, by applying a confining effect to crack propagation, thereby improving
the stress-strain properties of the concrete (Figure 6-1 (a)),
2. The fibers perform the role of transverse steel reinforcement (stirrups) and in
the process act to support some of the shear load acting on the concrete (see
Figure 6-1 (b)).
As previously discussed in Chapter 3, the effect of assumption 1 has already been

accounted for in the stress-strain relations for the HSFRC .
In order to account for assumption 2, values for py (the equivalent fiber contri-
bution to the transverse reinforcement) were chosen which best fit the observations
from the experimental program .
A table of the results from the application of this model, as well as the values of
p, used, is presented below (Table 6.1):

Figure 6-1: Action of fibers to (a)improve concrete properties, and (b)perform the
role of stirrups
Specimen type Observed (lbs) Predicted (lbs' A.(in2)

(al) 0.0-1.46-28.1-0.375 37585.5 37740 0.000
(a2) 0.4-1.46-28.1-0.375 40993.4 40503 0.027
(a3) 0.8-1.46-28.1-0.375 42377.8 42769 0.053
(a4) 1.2-1.46-28.1-0.375 49225.7 44934 0.081
(bl) 0.8-0.93-28.1-0.375 I 74443.6 50519 0.053
(b2) 0.8-1.73-28.1-0.375 35953.9 38510 0.053
(cl) 0.8-1.46-61.5-0.375 50691.1 48968 0.150
(c2) 0.8-1.46-50.0-0.375 38237.1 436860 0.027
Table 6.1: Results from deep beam shear model

40000
-- Theoretical
V 20000
- Experimental
0.0 0.1 0.2 0.3 0.4
Deflection (in)
Figure 6-2: Model comparison for V1 = 0.0%
Comparison of the experimental and theoretical plots are shown in Figure 6-2,
Figure 6-3, Figure 6-4, Figure 6-5, Figure 6-6, Figure 6-7, Figure 6-8, and Figure 6-9 .
It should be noted from the plots, that as the peak load capacity is appproached,
the theoretical prediction indicates a decreasing stiffening behavior of the specimens,
wheras, with the experimental results, this decrease in stiffness is not as dramatic .
This arises because, as previously discussed in Chapter 3, in the model study, the
ability of the fibers to preserve the concrete flexural rigidity, after cracking is initiated,
is not accounted for in terms of the value of I. used .
Ashour[38] , proposed a modification to the formula for I. to account for this stiff-
ening effect, and so one of the directions for future study could be the implementation
of this stiffening behavior in the model .
The post-peak slopes, indicated in the plots, have been obtained using a simple
empirical analysis (see Chapter 3) .
This was obtained by analysing the results from this experimental program, as
well as those due to Shahbazker[24] .
50000 50000
40000 40000
30000 30000
---. Theoretical
Experimental
20000 20000
10000 10000
0 0
0.0 0.1 0.2 0.3 0.4 0.0 0.1 0.2 0.3 0.4
Deflection (in) Deflection (in)
50000 ,uuuu~r
40000
40000
30000
30000
---. Theoretical
20000 - Experimental
20000
10000
10000
0
0
0.0 0.1 0.2
0.0 0.1 0.2
Deflection (in)
Deflection (in)

50000 50000
40000 40000
30000 30000
.... Theoretical
- Experimental
20000 20000
10000 10000
0 0
0.0 0.1 0.2 0.0 0.1 0.2
Deflection (In) Deflection (In)
Figure 6-5: Model comparison for Vf = 1.2%
^^^^'
80000 u0000uu
60000 60000
40000 --- Experimental

- 40000
... Theoretical
20000 20000
0 0
0.0 0.1 0.2 0.3 0.0 0.1 0.2
Figure 6-6: Model comparison for a/d = 0.88

40000 40000
20000 20000 - Experimental

a .... Theoretical
0 0
|
0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25
Figure 6-7: Model comparison for a/d = 1.73
60000 60000
40000 40000
- Experimental
---. Theoretical
20000 20000
0 0
0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3
Deflection (in) Deflection (ins)
Figure 6-8: Model comparison for hooked end fibers

40000 40000
Experimental
a 20000 20000
0
-- Theoretical
0
0
0.00 0.05 0.10 0.15 0.20 0.25 0.00 0.05 0.10 0.15 0.20 0.25
Figure 6-9: Model comparison for straight fibers
noticeable drop in capacity
PA gradual decrease in capacity

D
0 P
-j B , plateau.
Deflection
Figure 6-10: Generalised post-peak behavior of specimens

As seen in Figure 6-10, the fiber contribution to the improved post-peak behavior
was obtained by subtracting the (approximate) residual dowel capacity from the load
at which the post-peak ductility comes into effect . The gradient of the post-peak
portion for the various volume fractions was then obtained, and a simple linear for-
mula was used to fit these results for the particular steel fiber type used in this study,
of the form ( Figure 6-10 ) :
PB = PA - 183333.3 x dB x 0.008/Vf
This modification to the post-peak behavior of the model was necessary as the
model predictions, when this modification was not included, did not correspond to
the observed behavior for the post peak portion .
It should, however, be noted that the main purpose of this study was not to study
the post-peak behavior of the specimens, but to predict the ultimate capacities of the
specimens .
The drop in capacity at peak load, followed by a gradual decrease in capacity,
probably arises because the combination of arch, fiber and dowel action are active
up to the maximum load, however, on exceeding the corresponding strain, the arch
mechanism is no longer active and so the load can only then be supported by the
combined dowel and fiber supporting action .
Hence, as the capacity of these two mechanisms is not as great as when the arch
action is also present, in a displacement controlled system, it is expected that a drop
in capacity will be exhibited, until the residual load can be supported, as observed .
The gradual gradient then results from a combination of further fiber yield and
rupture, and pullout of fibers (which will occur for some fibers before the yield
strength is reached, even though the bond strength has been improved), and dowel
action .
This sudden drop in capacity at peak load was also evident in the tests performed
by Shahbazker[24] and Ashour[23].
480
358 200
z 300 .59 ./
2150
2 56 .
. ',0o
,8 ": "-. v 110 .0
z
lee •oe~·
.,: , .
ISoO -4.58% *0 5I /
/,
,,
... . O...o,
tO000
5B :/0
'
So
5e
l eI 26 38 46 5 6SB 70 e
9e I6 28 30 48 58 6B 76 9
Deflection
on n,( s
mtdipn,C(i ) DIflection at
aat mldspan,(tm )
Figure 6-11: Beam tests by Ashour[23]
Indeed, Ashour[23] noted this fact and remarked that "HSFRC beams with low
fiber content failed in shear and ezhibited sudden failure at the ultimate stages "
Beam tests performed by Ashour[23], and illustrative of this behavior, are shown in
Figure 6-11 .
Thus the behavior observed in this research program is consistent with the results
from past studies .
Valle[8], however, observed noticeable increases in the ductility of push-off speci-
mens at V1 = 1.0% .
This may be explained by the fact that the experimental variable in Valle's push-
off tests was the shear strain .
In the beam tests, however, central span deflection results from a combination of
shear and flexural strains . Thus, on exceeding the peak load, at a load level below
that of the ultimate capacity, while the shear deflection might be increasing, there is
a decrease in flexural deflection and, thus, the overall deflection is less than if shear
strains were the only contributing factor to the beam deflection (which is essentially
81
the situation with Valle's tests)
Comparing the results for the beam tests with different volume fractions of fibers
(Figure 6-12 and Figure 6-13), it is seen that as the volume fraction of fibers increases,
there is, indeed, a trend of increase in the ductility of the specimens, in terms of the
area below the curves .
Most notably, addition of fibers raises the load at which the plateau occurs, thus
indicating that, for this portion of the post-peak behavior, both fibers and dowel
action are active .
The plateau therefore results from a combination of dowel steel yielding and fiber
yielding/pullout, and contributes greatly to an increase in ductility, with reference to
the specimen without any fibers .
Also note that the dowel steel ratio is p = 7.08%, which is a very high prcentage.
Thus, if we assume that, in a practical specimen, the steel ratio might only be half
of this value, then the ultimate dowel capacity would be reduced from approximately
16000 lbs at V1 = 0.0% in Figure 6-12, to approximately 8000 lbs .
Therefore, for the specimen with V1 = 0.4%, the increase in plateau load, and
hence ductility, would be 50%, as opposed to the current value of approximately 30%
(Figure 6-12), thus showing the significant improvement in ductility on the addition
of fibers .
50000
40000
30000
- 0.0
--- 0.4
- - 0.8
1.2
20000
10000
0
0.3
Deflection (in)
Figure 6-12: Comparison of load-deflection plots for various V1

50000
40000
30000
-0.0
--- 0.4
- - 0.8
- - 1.2
20000
10000
0
0.1 0.2 0.3
Deflection (in)
Figure 6-13: Comparison of load-deflection plots for various Vf

Specimen Observed (lbs) Predicted (lbs) As (in 2)
0.0-0 38200 40764 0.000
0.0-1 45200 43922 0.000
0.0-2 48168 47584 0.000
0.4-0 45200 44065 0.029
0.4-1 50072 52207 0.029
0.4-2 52356 54073 0.029
0.8-0 48400 47863 0.065
0.8-1 54832 55438 0.065
0.8-2 57138 57155 0.065
Table 6.2: Application of model to the results from Shahbazker[23]
It should also be noted that when comparing the plots for the theoretical and
observed results, the peak loads do not match .
This is because each of the plots only makes a comparison to a single test specimen,
and Table 6.1 shows that the results from the model actually offer a satisfactory fit
to the average of the two test data obtained for each case .
The predicted peak load deflections also do not correspond in all cases to the
observed deflections, and this primarily arises because of the settling in deflection
which is incurred in the experimental tests .
A further discussion of the value of py used will be provided in the following
section .
Table 6.2 (first column indicates the V1 as well as the number of stirrups) indicates
that the model predictions correlate quite well to the data from other tests, although,
Table 6.3 shows that the results for the tests performed by Ashour[38] tend to be
overestimated, the exact reason for this not being certain . This suggests the need
for a much greater database of results in order to further test the applicability of this
model .
a/d V1 p Observed (lbs) Predicted (ibs)
1 0.5 2.84 110984 158413
2 0.5 2.84 58847 103106
1 1.0 2.84 155427 162300
2 1.0 2.84 73987 103556
Table 6.3: Application of model to the results from Ashour[23]
6.2 Discussion of proposed equation for shear strength

prediction
-While the use of a computer model gives useful insight into the behavior of the deep
beam specimens, nevertheless, the development of a design equation would provide a
more versatile means of strength prediction to the practising engineer .
Therefore, specific emphasis has been placed in this research program on modifying
a practical formula which can readily be used for strength predictions
6.2.1 Choice of shear equation
In examining the available procedures for shear strength estimation, the formula
which appears to offer the best results was proposed by Bazant[26] :
vC = 6.5 pl/3( + 3000 p/(a/d)5) (1+

+1+ d/25da
p = longitudinal steel ratio,

da = maximum aggregate size,
pU = stirrup reinforcement ratio,

1/po = 400 (1 + tanh(2 (a/d - 2.8))) ,
This formula is a modified version of the formula proposed by the same researcher
in 1984[25], and is more powerful because it takes into account :
1. not just the size effect of the specimen, but also the effect of the maximum
aggregate size,
2. the effect of stirrup action in improving the concrete shear capacity, due to the
application of a confining effect on the concrete, and also by offering support to
the rebars and thereby improving the dowel contribution .
It should perhaps at this stage be noted that the specimen size effect arises prin-
cipally because of a release of strain energy from the beam into the cracking zone,
as it extends . Thus, as the size of the structure increases, the greater is the energy
release into the this zone .
The area of the cracking zone is assumed to be proportional to the maximum
aggregate size in this treatment .
The obvious advantage of this formulation is therefore the consideration of the size
effect, for even though the formula proposed by Zsuttty and used in practise offers
satisfactory predictions, it nevertheless does not account for the size effect
Therefore, in Zsutty's formulation, as well as in the ACI formulation, as the size of
the structure increases, the margin of safety between the predicted and actual failure
load decreases, i.e. the factor of safety is not uniform .
Comparison of Bazant's formula to others available for shear strength estimation,
confirms its applicability, as shown in Figure 6-14 .
Bazant's formula is also attractive for practical applications because it has been
derived largely from theoretical considerations (though some emperical analysis has
been utilised), and therefore actual material behavior has been utilised in effecting
its derivation.
To implement the equation in design practise, Bazant suggests using a coefficient
of 4.5 instead of 6.5, as this leads to a lower bound estimate of the beam shear
capacity .
J
S(b) ZS/rM
V,.I Mu.0WOOPdiII1'latidD
:15 ,t
Vc-60UtP&O"' for&14
11 S
, ,
a2 2S 'wD X 328pp
Y 360 p6
No. ddaial 7
SS.D.65.3 pm
r 0.956
S 0.181
S3
rA (d) PROPOSED
FORMULA.BUT WITTHOU
CONSIDERING
56 2 ThE STIRRUPSEFFECT
S.D.- . p ,to
X - 348 t
Y * 360*
r 0.955,
2.5 v * 0.156 .
No. of Dsta 7
2i
dI "Is
to V.
NAmu Ul(M#re Ar srur wdhSainrr CalUedad. l (i. in

InUA
Figure 6-11: Comparison of strength predictions using different formulae[26]
6.2.2 Verification of the applicability of Bazant's formula

to HSC deep beams
As mentioned in the previous section, the correlation between Bazant's formula and
experimentally observed data is very good, however, the data-base used for the com-
parison was generally based on normal strength concrete, with f' < 7 000 psi .
Therefore, before modifying this formula to account for the beneficial contribution
from fibers, it was necessary to validate its applicability to HSC deep beams .
Now, as there is not a large database of results for HSC, the procedure adopted
was to use the existing formula to see whether it offerred satisfactory predictions i.e.
whether the constants of the equation were adequate .
This was achieved by utilising the deep beam results from tests performed by
Ahmad[21] , in which the behavior of HSC beams, with f, = 10 000 psi, were investi-
gated
As seen from Figure 6-12, Figure 6-13, Figure 6-14, Figure 6-15, and from Ta-
Specimen a/d p Observed (psi) Bazant (psi) Method II
A4 2.3 0.0393 525 485 497
A5 2.0 0.0393 938 572 587
A6
A10
1.0
2.3
0.0393
0.0177
2250
452
f 1951
317
2001
326
All 2.0 0.0177 305 362 371
A12 1.0 0.0177 1221 1068 1096
2.3
B4
B5
0.0504
2.0 0.0504
811
605 'F 572
680
587
697
B10
Bl1
B12
2.3 0.0225
2.0 0.0225
0.0225
S1.0 1
352
672
1172
'F
I.
370
424
1287
379
435
1320
C4 2.3 0.0664 - 552 674 690
C5 2.0 0.0664 1531 811 832
C10 2.3 0.0326 315 444 455
Cll 2.0 0.0326 590 518 532
C12 1.0 0.0326 1353 f 1695 1738
Table 6.4: Results from tests by Ahmad[21] and predicted strengths (Method II
discussed in Section 6.2.3)
ble 6.4, comparison of the observed and predicted results indicate that the formula
offers a satisfactory means of prediction for the shear strengths of HSC deep beams .
The results from Elzanaty[22] with ff = 10000 psi also indicate the capacity of
this equation to predict the shear failure loads for deep beams as shown in Table 6.5.
In addressing the issue of why, as the equation implies, the only factor which really
determines the ultimate strength of the specimens, when changing from NC to HSC,
is f,, it must be realised that the principal method of shear transfer is through the
arch mechanism, the effectiveness of which will ultimately depend on the strength of
the concrete, f .
6.2.3 Modification of Bazant's formula

As Bazant's formula has been shown to be valid for application to HC deep beam
shear behavior, it is therefore necessary to modify this formula to enable the increase
a/dl p Observed (lbs) Predicted-Bazant (ibs) I Predicted-Bazant(coef.=4.5) (lbs)
2.0 0.012 35447 I 41524 28748
2.0 0.025 49281 60382 41803
Table 6.5: Results from tests by Elzanaty[22] (fe = 10000 psi)
Ahmad-Khaloo - Specimens A4-A6

2500
2000
1500
- Observed
-o Predicted (Bazant '87)
1000
1.5 2.0
Figure 6-12: Comparison of Bazant's predictions to Ahmad's results

Ahmad-Khaloo - Specimens A10-A12
11al flL
flL
14UI
1200
1000
800 - Observed
-0- Predicted (Bazant '87)
600
400
1.0 1.5 2.0
Figure 6-13: Comparison of Bazant's to Ahmad's results
Ahmad-Khaloo - Specimens B10-B12

1400
1200
1000
-- Observed
800
-0- Predicted (Bazant '87)
600
400
200
1.5 2.0

Ahmad-Khaloo - Specimens C10-C12
2000
1500
- Observed
L 1000
-o Predicted (Bazant '87)
I-
500
0
1.0 1.5 2.0 2.5
a/d
in strength, through the addition of fibers, to be satisfactorily predicted .

In applying Bazant's formula, it is necessary to note that the d. term in V11 + d/25da
is actually included to effect the nondimensionalisation of the ratio d/da .
Bazant[26] argues that every physically correct formula must be reduceable to a
nondimensional form . As the maximum aggregate size, d4, is the only nondimen-
sionalising constant which is easily defined, it was therefore used for this purpose in
the derivation.
The question then arises as to the effect of the maximum aggregate size variation
on beams of equivalent depth, since 1 + d/25da is meant to only account for the
actual specimen size variation.
To account for this effect, Bazant[26] added the modifying term (1 + 0.2/d)
based on a study of the spread of cracks in the contact area wit aggregates, to account
for the actual effect of the aggregate size on the specimen behavior .
Now, accepting the modifying term 1 + d/25d, as being representative of the
specimen size effect, the use of the term (1 + 0.2/d) would not seem applicable to
HSC as the aggregate size is not expected to have any significant effect on the material
behavior , as cracks generally propagate directly through the aggregate (Figure 2-6).
For the case of NC, the use of (1 + VO.2 /da) is acceptable as the fracture process
zone ( concrete is a brittle material characterised by brittle fracture - the tip of a crack
is blunted by a zone of micro-cracking that lies ahead of the crack tip ) is essentially
determined by the material inhomogenieties, but this is not the case for HSC, which
is a much more homogeneous material than NC .
If this term is then excluded for HSC, there is also a need to eliminate the in-
troduction of da in the expression 1 + d/25da, as not only would HSC be expected
to have a much smaller fracture process zone than NC (and of more constant size
as the material is more homogeneous), but also, as previously discussed, the actual
aggregate size variation should not have an effect on the strength of the specimens.
Thus, in terms of utilising an expression for predicting the shear response of HSC
specimens, there are two options available :
1. Utilise Bazant's 1987 formula, noting that though the theory behind its deriva-
tion is not necessarily applicable to HSC, the term (1 + 0.2/da) offsets the
da term in 1 + d/25da, and thereby makes the expression provide more con-
stant results for maximum aggregate size variation at any particular depth -
method I .
In practise, the maximum aggregate size will only vary within a very narrow
range of typically 0.75-1.0 in., and therefore this is a satisfactory assumption .
Table 6.6 indicates how small an effect the aggregate size variation would have
on typical practical specimens of HSC using the 1987 formula ( and the formula
from method II - see later ).
2. Modify the 1987 formula to allow for a constant nondimensionalising term to

be used which is not reflective of the maximum aggregate size - method II.
This is because even if the fracture process zone ahead of the crack tip is treated
as a line, a certain characteristic length must be introduced for this zone, if the
idea of nondimensionality is adhered to .
a/d da = 0.5 in da= 0.75 in da = 1.0 in Method II
1.5 423070 440360 449990 43576
2.0 274390 285590 291850 28261
Table 6.6: Predictions for practical sized specimens using 1987 formula and method
II formula(p = 0.04,f = 10000 psi,d = 19.6in.)
Method I - use of Bazant's 1987 formula
As previously mentioned, the contribution from the fibers can be assumed to be

two-fold :
1. they cause an improvement in the inherent behavior of the concrete, and
2. they perform the role of stirrups .
Assumption 1 is generally expected (and verified by the values of f£ and ft etc.)

on adding fibers to concrete, and the validity of assumption 2 is indicated in Figure 6-
19.
From the figure it is seen that there is a definite, almost linear, trend in shear
strength as the volume fraction of fibers is increased .
While the fibers do contribute to the inherent strength increase of the concrete,
this strength increase is not so dramatic as to create such a noticeable increase in the
shear strength of the beams .
Therefore some other factor must account for this strength improvement, and this
factor is taken to be the effective contribution from the fibers to the truss supporting
action i.e. the fibers perform the role of transverse steel .
As v, = v, + v, then the proposed modification to the equation is of the form :
vU = 6.5 p/ 3a(Vf + 3000 /p/(a/d)5 ) + f

1+ 25d, (1+(p,+pf)/po)
xO1 Non-linear optim. for exp. results(*)
3.
4.(
4.4
4.2
"•.6
0 0.2 0.4 0.6 0.8 1 1.2
Vol. Fraction (Vf)
Figure 6-16: Normalised experimental results and non-linear optimal curve
v, = ultimate shear stress capacity,

vc = contribution from the concrete,
v, = contribution from the stirrups,
P1 = equivalent transverse steel contribution from the fibers
Figure 6-16 also shows the results of a non-linear optimization analysis, performed
using the 'Matlab' matrix analysis package, to enable the unknown paramater pf to
be identified for the particular situation under study .
Analysis, treating V1 as the only variable, resulted in an expression for the un-
known variable of the form :
P=C V
1
where the coefficient in the expression is actually a function of the fiber and
specimen properties .
Therefore, the reason for the large number of variables in the experimental pro-
gram is identified, for these variations, as well as the results from the studies by
Shahbazber[24] and Ashour[23] , enable the variables which influence the value of p1
to be determined .
Examination of the results due to Shahbazker indicates that there is a trend of
increased fiber contribution as the strength of the concrete increases .
This is taken to result from the fact that, as the strength of the concrete increases,
it is expected that the fiber-matrix bond will correspondingly increase (within the
strength range used of f: < 15 000 psi), thus accounting for the term (1lo0 .
Study of the results for the other variables affecting the behavior of the fibers has
-led to an expression of the form :
p,= b.d Vf
- (-dV F)y. 1a
000
s = a fora>=d,
s = d fora<d
Valid for Vf 5 1.2% and f, < 15 000 psi
The fiber effectiveness is given by the term F, and is equal to 1.0 for crimped/hooked
end fibers, and 0.5 for smooth-straight fibers .
The term ()0o.2 expresses the influence of the aspect ratio of the fiber on the
overall strength increase of the specimens . The small value of the index indicates
that the pullout mechanism is not as major a contributing factor to the fiber behavior
as in NC .
The term b.s is essentially related to the overall size of the specimen . It is used
to account for the fact that as the size of the specimen increases, so the quantity of
fiber 'transverse' reinforcement on the shear plane will correspondingly increase .
The square root reflects the fact that all the fibers are not effective at any one cross
section i.e. it includes the contribution from an orientation factor, and as well, as
Specimen Observed (lbs) Predicted-6.5 (lbs) Predicted-4.5 (lbs) Af (in2
al 37585.5 35850 24819 0.000
a2 40993.4 39927 28820 0.032
a3 42377.8 43998 32843 0.065
a4 49225.7 48308 37085 0.099
bl 74443.6 98362 69871 0.048
b2 35953.9 35282 27001 0.070
cl 50691.1 46443 35111 0.080
c2 38237.1 38802 28026 0.031
dl 40379.1 44234 32913 0.061
d2 30980.6 41124 30449 0.054
Table 6.7: Comparison of observed and predicted shear capacities using the modified
equation
the beam size increases, the fibers are less likely to be evenly distributed and bonded
( as the size of the specimen increases, specimen quality generally decreases ), and
therefore it helps to account for this size effect . In addition, as all the fibers along
the failure plane do not yield at the peak load, it therefore also reflects the equvalent
area of failure plane on which yielding occurs.
Table 6.7 (see also Table 4.1) provides the results obtained using this equation (as
well as the results obtained using a design coefficient of 4.5 instead of 6.5) .
Comparison of the observed results for the aspect ratio and fiber effectiveness
with the predicted results shows that the behavioral trend is correctly captured by
this modified version of Bazant's equation (Figure 6-20) .
It should, however, be noted that as the aggregate size was decreased, the strength
tended to drop, and not increase, as predicted by the equation . This might result
from the fact that, on decreasing the size of the aggregate, the workability of the mix
also decreased, consequently affecting the overall shear strength of the specimens . It
should also be noted that the observed strength of specimen d0.09 (see Table 6.7) is
far below the predicted value, and this is because the decrease in workability when
using such small aggregate (of size equivalent to that of sand) was so distinct as to
cause a dramatic reduction in strength .
Aspect ratio and Fiber effectiveness variation
55UUV -
50000 -
Observed
a 45000 - -o-Observed
-a-
-o- Predicted
Predicted (Pandor)
(Pandor)
F
40000 -
35000 I I I I
20 30 40 50 60 70
aspect ratio
Figure 6-17: Predictions for the fiber aspect ratio and effectiveness variation
The result for specimen d0.25, with only a small drop in strength, might, however,
indicate that, as previously discussed, the aggregate size effect is not necessarily
significant for HSC, as the cracks pass directly through the aggregate particles . This
would therefore indicate that the inclusion of an aggregate size effect term is not
necessary in the modified equation, however, as also previously mentioned, further
tests would need to be performed in order to confirm this .
Comparison of the predictions to the results obtained by Shahbazker (Figure 6-
18, Figure 6-19, Figure 6-20, and Figure 6-21) indicate that the modified equation
again allows for satisfactory correlation to the observed trends .These results are also
provided in Table 6.8 .
With two stirrups, however, it is seen that the equation tends to overpredict
slightly the observed strength of the specimens .
This probably results from the fact that in practise, Shahbazker[24] observed that
in these cases the stirrups had not yielded at failure, wheras, in applying the modified
equation, it is assumed that the stirrups have yielded .
Shahbazker - No stirrups
i - Observed
- Predicted (Pandor)
0.0 0.2 0.4A 0.6 0.8
Vol. fraction (%)
Figure 6-18: Application of modified equation to tests performed by Shahbazker for

volume fraction variation
Shahbazker - Vf=0.0
55000
50000
- Experimental
45000
-- Predicted (Pandor)
40000
35000
0 1 2
No. of stirrups

V1 = 0.0%
Shahbazker - Vf=0.4%
55000
50000
- Observed
SPredicted tPandor)
45000
A0000
No. of stirrups

Vf = 0.4%
Shahbazker - Vf=0.8%
60000
55000
-o-Observed
*- Predicted (Pandor)
50000
45000
0 1 2
No. of stirrups

Vf = 0.8%
100
Specimen type 1 Observed (lbs) Method I Method II
HB-0.0-0 38200 36827 36862
HB-0.0-1 45200 43522 43537
HB-0.0-2 48168 50848 50843
HB-0.4-0 45200 41319 41306
HB-0.4-1 50072 48013 47980
HB-0.4-2 52356 54465 54413
HB-0.8-0 48400 47119 47050
HB-0.8-1 54832 53063 52979
HB-0.8-2 57138 59753 59649
Table 6.8: Shahbazker's test results and modified equation predictions
a/d V1 p Observed (lbs) Method I Method II

1 0.5 0.0284 110984 143020 147880
2 0.5 0.0284 58847 60413 62092
1 1.0 0.0284 155427 154020 158070
2 1.0 0.0284 73987 75405 77307
2 1.0 0.0458 82171 89017 91680
Table 6.9: Ashour's test results and modified equation predictions
Applying the modified equation to the test specimens used by Ashour[23] provides
the results shown in Table 6.9 .
It is again seen that the correlation between experimental and observed values is
excellent, and this is shown graphically in Figure 6-25 .
The proposed equation therefore satisfies the available test data for HSFRC spec-
imens .
Method II - Replacement of da term
In this method, it will be attempted to replace the d, term in the 1987 equation
(Bazant[26]) with a constant which reflects the fracture process zone length for HSC
beams.
Bazant[39] arrived at a value for a term do, which is related to the length of the
fracture process zone ( and the specimen geometry ) at f] = 12400 psi . While it
101
Ashour-Wafa (Specimens h and i)
a" - Observed
- Predicted (Pandor)
1.0 1.5 2.0
a/d
Figure 6-22: Application of modified equation to tests performed by Ashour
would be expected that this value would decrease as the strength of the concrete
increases, and therefore the concrete becomes more brittle, it is not expected that
this decrease would become too dramatic, and therefore this term will be used as the
constant in the size effect expression .
It is also noted that the magnitude of do (0.52 in) is of the same order of magnitude
as the maximum aggregate size .
This implies that the other constants will not be dramatically affected by the
replacement of the d. term by do in order to arrive at a reasonable correlation between
the experimental readings and the predicted values .
As the trend of strength increase on inclusion of fibers will be the same for either
method I or method II, then the contribution from the fibers to the strength of the
specimens will be the same using either method, and so pf does not change .
Therefore the modified equation by method II is of the form :
= 1+ 10.8 1/"(1 i+ 3000 p/(a/d) )+ (pv + pf) f,

25do (1+(pv+pj)/po)
Comparison of the predictions of this formula to the observed data ( Table 6.10,
Table 6.9, Table 6.4 and Table 6.8 ) indicate that the correlation between the results
102
Specimen Observed (lbs) Predicted-Method II (lbs)
a0.0 37585.5 35885
a0.4 40993.4 39950
a0.8 42377.8 44009
al.2 49225.7 48307
b0.93 74443.6 98406
bl.73 35953.9 35289
c61.5 50691.1 46449
c50.0 38237.1 38825
d0.25 40379.1 43346
d0.09 30980.6 41139
Table 6.10: Comparison of observed and predicted shear capacities using method II
is very satisfactory and the predictions are indeed extremely close to those using
method I (typical graphs to illustrate the closeness of fit for method II were not
plotted for this reason .)
Returning to the model used, and examining the value for p, used in trying to fit
the observed data to the predictions, it is seen, by comparing Table 6.1 to Table 6.7,
that the expression for py could be of the form :
.s f) 0.2 v f
p,= 0.82 -b. 4 (Id F)o. F,1 000
Thus, while the values of p, and pf do not exactly correspond, they are nev-
ertheless very close in magnitude, and this further tends to support the use of the
expression for pf in the modification to Bazant's equation .
The use of a coefficient of 4.5 instead of 6.5 (Table 6.7) in method I allows for
a higher degree of safety when applied to actual design practise . Indeed the pre-
dicted results were all on the conservative side when compared to the experimental
observations, even for the cases with two stirrups .
The application of the same safety margin to method II would again yield consis-
tent results
103
Chapter 7
Summary, conclusions and

recommendations for future work
7.1 Summary
In the experimental program of this research study, the behavior of HSFRC deep
beams was studied when the experimental variables of a/d, V1, 1/d, and da were
varied .
Overall 22 specimens were tested, with 4 concrete cylinders being tested under
splitting tensile loading and direct compression for every 2 beams cast .
In the theoretical program, an existing shear truss model was modified in order to
incorporate the effect of HSC, as opposed to NC, and also to include the contribution
from the fibers to the specimen behavior .
An existing shear strength equation for NC was also modified to enable the shear
strength of HSFRC deep beam specimens to be predicted .
7.2 Conclusions
7.2.1 Experimental
1. Some workability problems were incurred in trying to achieve a 28 day compres-
104
sive strength of 12000 psi, which necessitated the use of lower strength concrete
of f, = 10000 psi . This was thought to result mainly from the effectiveness of
the mixer used in this study, which did not offer the advantage of a pressurised
container, and relied on gravitational action for mixing .
At the water/cement used for this lower strength mix, satisfactory workability
was, however, attained .
2. Compared to plain NC, the post-peak softening behavior of plain HSC is dra-
matically reduced .
This was clearly observed in the compression tests, and in the actual beam shear
tests, where nominal post-peak 'ductility' was observed . This primarily results
from the less distributed cracking in HSC compared to NC, and to the ability of
the cracks to pass directly through the aggregates, which are therefore unable
to perform a crack-arresting role .
3. As noted above, the cracks in HSC were observed to pass directly through
the aggregates when propagating, instead of around them . This therefore
indicates an improved fracture toughness of the aggregate-mortar interface when
compared to the fracture toughness of the aggregate.
4. A trend of increased compressive strength (approximately 200- 300psi for every

0.4% of fibers added) was observed on adding fibers to the HSC .
The splitting tensile strength also showed this trend of improved strength on
addition of fibers, and a comparison of the f.t and f' values indicates that the
ratio f.t/lf for the HSC is not as high as would be expected for NC . This
indicates that though an improved tensile strength is observed for HSC, the
increase in tensile strength does not occur at the same rate as the increase in
the compressive strength .
5. The beams generally failed by the development of a crack at mid-depth, which

eventually propagated to the supports, resulting in a loss in the supporting arch
transfer mechanism .
105
Cracking began at approximately 20000 lbs for all the beams tested . As fiber
action occurs mainly in the post-cracking region when the fibers bridge across
cracks that have propagated within the matrix, this almost constant value of
first cracking load is expected, as the fibers do not therefore have a significant
influence on this load .
6. Addition of fibers to HSC causes a substantial increase in shear strength (for

Vf = 1.2%, shear strength increase = 30%), for the range of fiber volume frac-
tion investigated in this testing program (It should be noted that Shahbazker[24]
obtained only nominal improvements in shear strengths on adding fibers to NC) .
This suggests an improved bonding of the fibers to the HSC matrix, and that
the fibers are performing the role of pseudo-stirrups .
The experiments also show that the hooked ended fibers, with the largest aspect
ratio, are most effective in increasing the shear strengths of the beams
The aggregate size effect tests performed were inconclusive .
7. The post-peak behavior of the beams may be divided into three categories :
* an initial drop in capacity as the arch mechanism terminates,
* a gradual decreased capacity, during which period the fiber and dowel
action are contributing to the load carrying capacity, and
* a final plateau dominated by dowel yielding and fiber yielding/pullout
8. Addition of fibers caused an improvement to the post-peak ductility of the

beams, as more fibers were available to contribute to the load carrying capacity .
This fact is of significance to construction in seismic regions, where the energy
absorption capacity of a structure is very important .
9. The addition of fibers caused the specimens to maintain total integrity for the
compression and splitting tensile tests . The same observation was made for the
shear tests, and in addition , with fibers, the ultimate load was not characterised
by a loud noise, as the capacity diminished .
106
7.2.2 Theoretical
From the discussion presented in Chapter 6, the following conclusions are drawn :
1. The modified shear truss model applied to HSFRC may be used successfully to
investigate the behavior of these specimens, in terms of the prediction of the
peak loads, and the corresponding central point deflections .
2. The equation proposed by Bazant[26], may be successfully modified to account

for the behavior of HSFRC specimens .
As the aggregate size effect has not been established for HSC, two formulae
have been provided for this purpose .
The first formula includes the aggregate size effect, but the presence of the
aggregate variable in two sections of this equation, essentially leads to the ef-
fects balancing each other i.e. maximum aggregate size variation for the same
specimen depth does not have a dramatic effect on the specimen strength .
The second formula does not include the effect of maximum aggregate size
variation, and this is achieved by the replacement of the maximum aggregate
size term by a constant related to the length of the fracture process zone .
The fracture process zone, in this treatment, is quantified as being constant in

size because HSC is a much more homogeneous material than NC, and therefore,
as the fracture zone is generally related to the inhomogenieties in the material,.
then it is not expected that the size of the fracture process zone will vary much
for HSC specimens.
In both the model and the proposed equations, the fiber behavior has been
treated as leading to an inherent increase in the strength of the concrete, through
a crack arresting mechanism, and also, as performing the role of stirrups across
the shear plane, and the experimental program allowed for the equivalent area
of transverse steel provided by the fibers to be quantified .
107
3. The experimental variations in a/d, l/d, F, d, and V1 were performed in order
to validate the model and the proposed equations . It is seen that the model
and equations offer very satisfactory predictions for these various cases .
It should, however, be noted that the model and equations do not provide good
estimates for a/d < 1, and therefore should not be applied to these situations .
4. Application of the model and the proposed equations to the results from other
studies on HSFRC deep beam behavior indicate that the model, and especially
the proposed formulae, predict results which correlate very closely to the ob-
served data .
They therefore provide the basis for the use as tools in practical design appli-
cations .
7.3 Recommendations for further study

Several areas may be identified which need further study in order for a more complete
understanding of the behavior of HSFRC beams to be attained :
* Aggregate sizes for larger specimen should be varied in order to verify the effect
of aggregate size on specimen behavior
* The V1 of fibers should be increased above 1.2% in order to determine the limit
of applicability of the proposed formulae and to investigate any further change
in the post-peak behavior of the specimens .
* The post-peak behavior of deep beam specimens needs to be modelled more

accurately .
* A wider database of specimen test results is needed, as current data is very

limited .
* Practically sized specimens need to be tested to verify the applicability of the

proposed relationships, especially with respect to the specimen size effect .
108
* The behavior of test specimens with f, > 14000 psi must also be investigated
as current tests have been limited to lower strengths .
* Research must be extended into the area of HSFRC slender beams ( aid > 2.5 ),
as these are the members principally used in structural practise .
109
Appendix A
Computer model
C program model for the behavior of HSFRC .
110
#include<stdio.h>
#include<math.h>
main()
FILE * fpo;
int m,n,check,k
double fc=9052.6,b=3.0,h=4.5,dv=2.52,stu,sqlamda,lamda,al,a2 ;
double ex,ey,er,ed,load,txy,fr,Ig,d=3.75,yt,Mcr,Ma,Mu,Icr,kd,Ie,delta,Ect;
double Ec,fcr,v,e0=0.003,sd,sr,Be,ecr,fx,fy,prod
double Es = 29e6,sx,sy,rh,rhy,S,nbars=3,T,Vf ;
double gxy,sum,K,a=3.5,ans 1,ans2,load 1,deltal,disp ;
double db=0.583,fb,c,Kf= le6,sqgdow,gdow,Du,totaldow,Su,D;
/* s=sigmaa=alpharh=area_steelgxy=shear_straine=epsilon Be=par._beta
db=ave._bar_diam.c=crushed_zoneDu=ult._dow. Kf=conc.Jfndtn._mod.
Su=peak_dow._defl. D=actual_dow._load*/
/* note that used a positive eO above - modify below, as tension +ve */
/* note dv used */
Vf = 0.8;
111
-2-
stu = 0.41 *0.82*(Vf/100)* 1*600/(2*4.42e-3);
Ec = 40000*sqrt(fc) + 1000000;
fcr = 7.5*sqrt(fc);
ecr = fcr/Ec;
K = dv/h*(h/a*(4/3-2/3*a/h)) ;
fy = 60000 ;
fr = 7.5*sqrt(fc);
Ig = b*h*h*h/12;
yt = d/2.0;
Mcr = fr*Ig/yt;
Mu = 26000*5.5; /* mom. capacity of section - check!! */
Ect = 0.4*(Ec*(100-Vf) + Es*Vf*0.41*0.82)/100.0 ;
sqgdow = sqrt(Es/(Kf*db));
gdow = sqrt(sqgdow) ;
fb = 154*sqrt(fc)/cbrt(db);
fb = 18434.5;
load = 0.0 ;
rh = 0.797/(b*d) ;
rhy = (0.053)/(b*d)*Vf/0.8 ; /* note well !!!!!!!!!!!!!!*/
fpo = fopen("moda2.mat","w");
if(fpo == NULL) /* check for opening error */
112
-3-
{
printf("File opening error\n");
exit(- 1);
for(ed=-0.0002;ed>-0.0046;ed = ed - 0.0001)
ans2 = 1000000.0;
for(er=-0.00001 ; fabs(ans2) > 0.05 ; er = er + 0.00001)
printf("c\n");
sqlamda = 0.7-er/ed;
lamda = sqrt(sqlamda);
/* compute stress using stress-strainconditions -
remember to use appropriatepartof stress-straincurve */
if(-ed < (eO/lamda) && ed < 0.0000)
/* sd */ sd = -fc*(2*ed/-e0-lamda*ed*ed/(eO*e0));
/* tension +ve, sofc -ve */
if(-ed > (eO/lamda) && ed < 0.0000) /* post peak */
113
-4-
sd = -fc/lamda*( 1-(ed/-eO-1/lamda)*(ed/-e0-1/
lamda)/((2-1/lamda)*(2-1/lamda)));
if(ed < ecr && ed > 0.0000 )
sd = ed*Ec ;
if(ed > ecr && ed > 0.0000 ) /* post peak */
{
Be = sqrt((ed-ecr)/0.005) ;
sd = (fcr+Be*stu)/(1+Be);
if(er < ecr && er > 0.0000 )
/* sr *V sr = er*Ec ;
if(er > ecr && er > 0.0000 ) /* post peak */
Be = sqrt((er-ecr)/0.005);
sr =0.2*(fcr+Be*stu)/( 1+Be);
if(-er < (e0/lamda) && er < 0.0000)
sr = -fc*(2*er/-e0-lamda*er*er/(e0*e0)) ;
114
-5-
if(-er > (eO/lamda) && er < 0.0000) /* post peak */
sr = -fc/lamda*(l1-(er/-eO- I/lamda)*(er/-e0- 1/
lamda)/((2-1/lamda)*(2-1/lamda)));
al=acos(sqrt((sr+rh*Es*er)/(sr-sd+rh*Es*(er-ed)))) ;
ex = ed*cos(al)*cos(al)+er*sin(al.)*sin(al);
ey = ed*sin(al)*sin(al)+er*cos(al)*cos(al);
if(fabs(ex) > fy/Es)
al = acos(sqrt((sr+rh*fy)/(sr-sd)));
ans l=(sd*(K*sin(al)*cos(al)-sin(al)*sin(al))-rhy*Es*ey)/
(K*sin(al)*cos(al)+ cos(al)*cos(al)) ;
if( fabs(ey) > fy/Es )
ans l=(sd*(K*sin(al)*cos(al )-sin(al)*sin(al))-rhy*fy)/
(K*sin(al)*cos(al)+ cos(al)*cos(a1));
ans2 = (sr - ansl)/sr ;
printf("fb %If er %If ed %lf\n",fb,er,ed);
115
-6-
if( fabs(ey) < fy/Es )
prod = Es*ey ;
else
prod = fy ;
sy=sd*sin(al1)*sin(al1)+sr*cos(al1)*cos(a 1)+rhy*prod;
gxy = 2*(ed-er)*sin(al)*cos(al);
txy = sy/K ;
load = -txy*b*dv;
/* dowel calcs. */
c = 0.05*fy*db*sin(al)/fc ;
T = Ma/(dv*rh*b*d) /* Ma/Mu*60000*/;
/* tension in each bar*/
Du = 2*nbars*(0.5*fb*(0.37*gdow*db-c)*(0.37*gdow*db-c)
+ 0.45*fy*db*db*(1-T*T/(fy*fy))/gdow);
/* Du is 2 *loadfor one section */
Su = 4.26/le6*Du/2.0 + 0.00945 ;
S = -gxy*dv ;
116
-7-
if(S <Su)
D = Du*sqrt(S/Su);
else
{
D = Du-Du*(S-Su)/(0.4/db-Su);
if( D < 0.4*Du )
D= 0.7*Du;
totaldow = D;
load = 2*load + totaldow ;
/* deflection */ Ma = a*load/2.0 ;
kd = fy*rh*b*d/(b*fc) ;
Icr = Mu*kd/fc ;
if( Ma < Mcr)
Ie = Ig ;
else
Ie = (Mcr/Ma)*(Mcr/Ma)*(Mcr/Ma)*Ig +
(1 - (Mcr/Ma)*(Mcr/Ma)*(Mcr/Ma))*Icr ;
117
-8-
if(loadl > load)
Ie = Icr ; /*for large deflections */
load I = load; /* update this check */
delta = 601.59*(load/2.0)/(6*Ect*Ie) + - gxy*dv;
if(deltal <delta)
fprintf(fpo,"%lf %lf %If %lf %lf %lf\n",
Du,Su,D,load,delta,gxy);
if(deltal > delta IIed == -0.0046 )
for(disp = 0.02; load > D; disp = disp + 0.02)
{
load = -183333.3*disp*0.8/Vf + loadl ;
delta = deltal + disp;
Ma = a*load/2.0 ;
T = Ma/(dv*rh*b*d) /* Ma/Mu*60000*/;
Du = 2*nbars*(0.5*fb*(0.37*gdow*db-c)*(0.37*gdow*db-c)
+ 0.45*fy*db*db*(1-T*T/(fy*fy))/gdow) ;
118
-9-
D = 0.7*Du ;
fprintf(fpo,"%If %If %lf %lf %If %lf\n",
Du,Su,D,load,delta,gxy);
ed = -0.0051 ; /* to terminate loop */
continue;
deltal = delta;
printf("total load %If gxy %lf %lf %lf\n",load,gxy,fb,c);
fclose(fpo);
119
Bibliography
[1] Webb, J., High-Strength Concrete: Economics, Design and Ductility, Concrete
International, Jan. 1993
[2] Neville, A. and Aitchin, P.A., High-PerformanceConcrete Demystified, Concrete

International, Jan. 1993 .
[3] Green, E.C., Behavior of Fiber Reinforced Concrete, M.S. Thesis, Department
of Civil Engineering, Massachusetts Institute of Technology, 1989 .
[4] Sarker, S.L., Performance of High Strength Field Concrete at 7 years, Concrete
International, Jan. 1993 .
[5] Park, R. and Paulay, T., Reinforced Concrete Structures, John Wiley & Sons
Inc., 1975 .
[6] Kani, G.N.J., Basic Facts Concerning Shear Failure, ACI Journal, Vol. 63, No.
6, June 1966, pp. 675-691 .
[7] Mindess, S. and Bentur, A., Fibre Reinforced Cementitious Composites, Elsevier
Applied Sciences, London, 1990 .
[8] Valle, M.O., Shear Transfer in Fiber Reinforced Concrete, M.S. Thesis, Depart-
ment of Civil Engineering, Massachusetts Institute of Technology, 1991 .
[9] Swamy, R.N. and Bahia, H.M., The Effectiveness of Steel Fibers as Shear Rein-
forcement, Concrete International : Design and Construction, Vol. 7, 1985, pp.
35-40 .
120
[10] Swamy, R.N., Jones, R. and Chiam, T., Shear Transfer in Steel Fiber Reinforcde
Concrete, Fiber Reinforced Concrete - Properties and Applications, ACI SP 105-
29, Detroit, 1987, pp. 565-592 .
[11] Narayanan, R. and Darwish, I.Y.S., Use of Steel Fibers as Shear Reinforcement,
ACI Structural Journal, Vol. 84, No. 3, March 1983, pp. 216-227 .
[12] Ward, R., Li, C. and Hamza, A.M., Steel and Synthetic Fibers as Shear Rein-
forcement, ACI Materials Journal, Vol. 5, Sept.-Oct. 1992, pp. 499-509 .
[13] Sharma, A.K., Shear Strength of Steel Fiber Reinforced Concrete Beams, ACI
Journal, Vol. 83, 1986, pp. 624-628 .
[14] Batson, G.B., Use of Steel Fibers for Shear Reinforcement and Ductility, Steel
Fiber Concrete U.S.-Sweden joint seminar, Stockholm, 1985, pp. 377-383 .
[15] Swamy, R.N. and Bahia, H.M., Influence of Fiber Reinforcement on the Dowel
Resistance to Shear, ACI Journal, Proceedings, Vol. 76, No. 2, Feb. 1979, pp.
327-355 .
[16] Ezeldin, A.S. and Balaguru, P.N., Bond Behavior of Normal and High Strength
Fiber Reinforced Concrete, ACI Materials Journal, Vol. 86, No. 5, pp. 515-524 .
[17] ACI Committee 544, Design Considerationsfor Steel Fiber Reinforced Concrete,
ACI 544.5R, ACI Structural Journal, 1988, pp. 563-580 .
[18] ACI Committee 363, Research Needs for High Strength Concrete, ACI Materials
Journal, Vol. 84, No. 6, Nov.-Dec. 1987 .
[19] Naaman, A.E. and Homrich, J.R., Properties of High-Strength Fiber reinforced
Concrete, High Strength Concrete, ACI SP-87, 1985, pp. 110-146 .
[20] Wafa, F.F. and Ashour, S., Mechanical Propertiesof High Strength Fiber Rein-
forced Concrete, ACI Materials Journal, Vol. 5, 1992, pp. 449-456 .
[21] Ahmad, S.H., Khaloo, A.R. and Poveda, A., Shear Capacity of Reinforced High-
Strength Concrete Beams, ACI JOurnal, Vol. 83, 1986, pp. 297-305 .
121
[22] Elzanaty, E.H., Nilson, A.H. and Slate, F.O., Shear Capacity of Reinforced Con-
crete Beams using High Strength Concrete, ACI Journal, Vol. 86, No. 3, March-
April 1986 .
[23] Ashour, S., Hasanain, G.S. and Wafa, F.F., Shear Behavior of High Strength
FiberReinforced Concrete Beams, ACI Structural Journal, Vol. 89, No. 2, March-
April 1992, pp. 176-184 .
[24] Shahbazker, A., Shear Behavior of Fiber Reinforced High Strength Concrete
Beams, M.S. Thesis, Department of Civil and Environmental Engineering, Mas-
sachusetts Institute of Technology, 1993 .
[25] Bazant, Z. and Kim, J., Size effect in Shear Failure of Longitudinally Reinforced
Beams, ACI Journal, Title no. 81-38, Sept.-Oct. 1984, pp. 456-468 .
[26] Bazant, Z. and Sun, H.H., Size Effect in Diagonal Shear Failure: Influence of
Aggregate Size and Stirrups,ACI Materials Journal, Title no. 84-M27, July-Aug.
1987, pp. 259-272 .
[27] Hsu, T.T.C., and Mau, S.T., Shear Strength Prediction for Deep Beams with
Web Reinforcement, ACI Structural Journal, Title no. 84-S53, Nov.-Dec. 1987,
pp. 513-523 .
[28] Tan, K.H. and Mansur, M.A., Shear Transfer in Reinforced Fiber Concrete,
ASCE Journal of Materials in Civ. Eng., Vol. 2, No. 4, Nov. 1990, pp. 202-214 .
[29] Hsu, T.T.C, Mau, S.T. and Chen, C., Theory of Shear Transfer Strength of
Reinforced Concrete, ACI Structural Journal, Title no. 84-S16, March-April 1987,
pp. 149-160 .
[30] Mansur, M.A. and Ong, K.C.G., Behavior of Reinforced Fiber Concrete Deep
Beams in Shear, ACI Structural Journal, Title no. 88-S13, Jan.-Feb. 1991, pp.
98-105 .
122
[31] Vecchio, F.J. and Collins, M.P., Modified Compression Field Theory for Rein-
forced Concrete in Pure Shear, ACI Journal, Proceedings Vol. 85, No. 3, March-
April 1986, pp. 219-231.
[32] Vecchio, F.J. and Collins, M.P., Predicting the Response of Reinforced Concrete
Beams Sublected to Shear, Using the Modified Com,pression Field Theory, ACI
Structural Journal, Vol. 85, No. 3, May-June 1988, pp. 258-268 .
[33] Soroushian, P., Obaseki, K., Rojas, M.C. and Sim, J., Analysis of Dowel Bars
Acting Against Concrete Core, ACI Journal, Title no. 83-59, July-August 1986,
pp. 642-649.
[34] Henager, C.H. and Doherty, T.J., Analysis of Reinforced Fibrous Concrete
Beams, ASCE, Journal of the Structural Division, Proceedings, Vol. 102, No.
ST1, Jan. 1976 .
[35] ACI Committee 363, State-of-the-Art Report on High-Strength Concrete, Title

no. 81-34, ACI 363R-84, July-August 1984, pp. 364-410 .
[36] ACI, Building Code Requirementsfor Reinforced Concrete and Commentary, ACI
318-89, pp. 156-165 .
[37] ACI Committee 544, Guide for Specifying, Proportioning,Mizing, Placing and
Finishing Steel Fiber Reinforced Concrete, ACI Materials Journal, Vol. 90, No.
1, 1992, pp. 94-102 .
[38] Ashour, S.A. and Wafa, F.F., Flezural Behavior of High-Strength Fiber Rein-
forced Concrete Beams, ACI Structural Journal, Vol. 90, Title no. 90-s29, May-
June 1993, pp. 279-287 .
[39] Bazant, Z.P., Gettu, R. and Kerr, M.E., Fracture Properties and Brittleness of
Hish-Strength Concrete, ACI Materials Journal, Vol. 88, Title no. 87-m66, Nov.-
Dec. 1990, pp. 608-618 .
123
[40] Valle, M. and Buyukozturk, O., Behavior of Fiber reinforced High Strength Con-
crete under Direct Shear, ACI Materials Journal, Vol. 90, Title no. 90-m13,
March-April 1993, pp. 122-133 .
124

Behavior of High Strength Fiber Reinforced Concrete Beams in Shear

Uploaded by

Copyright:

Behavior of High Strength Fiber Reinforced Concrete Beams in Shear

Uploaded by

Document Information

Original Description:

Original Title

Copyright

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Behavior of High Strength Fiber Reinforced Concrete Beams in Shear

Uploaded by

Copyright:

Behavior of High Strength Fiber Reinforced

Concrete Beams in Shear

@ Massachusetts Institute of Technology 1994. All rights reserved.

Author ...................... .......... .............

Certified by.................. .......- . .. .. .. .-..............

Accepted by ......................•rp"w "r ....

Submitted to the Department of Civil and Environmental Engineering

Thesis Supervisor: Oral Buyukozturk

Professor Buyukozturk, for providing the necesssary guidance to enable me to suc-

2 Review of the shear behavior of concrete beams 15

3 Load-deformation predictions for high-strength fiber reinforced deep

5 Observations and Results 59

7 Summary, conclusions and recommendations for future work 104

A Computer model 110

2-1 Beam action[5] .................... .......... 16

3-1 Truss model for reinforced concrete element[29] . ........... 28

4-1 Sectional drawings ............................ 46

5-1 Split cylinder test for V1 = 0.0 . . . . . . . . . . . . . . .

6-1 Action of fibers to (a)improve concrete properties, and (b)perform the

II to the specimens tested by Shahbasker

2.2 Application of Shahbazker's eqtn. to the results obtained by Ashour[23] 26

4.1 Specimen types investigated .......................

51~ Table of exnerimental results................

6.1 Results from deep beam shear model . . . . . . . . . . .

2. reduced axial shortening of members and therefore reduced problems concerning

3. construction time saving, as strength development occurs at an early age,

4. possibility of creating new types of structural supporting systems-an area which

5. economic advantages - the ability to remove formwork rapidly, the increased

Other advantages, such as reduced prestressing losses (due to a reduced coefficient

1.1 Objectives and scope of the research program

1.1.3 Organisation of thesis

Review of the shear behavior of

2.1 Shear behavior of plain concrete

Figure 2-1: Beam action[5]

Figure 2-2: Arch action[5]

2.2 Shear behavior of normal strength fiber rein-

Axial Strain, millionths

Figure 2-4: Improved post-peak ductility due to fiber addition[17]

2.3 Shear behavior of high strength fiber rein-

2.3.1 General properties of high strength fiber reinforced

2.3.2 Behavior of plain high strength concrete beams in

Figure 2-5: Dependancy of failure mode on shear span ratio[21]

a/d > 2.5, failure was by diagonal tension(Figure 2-5) .

2.3.3 Behavior of high strength fiber reinforced concrete

S= (0.7f +7F)-a +17.2p a (I)

Figure 2-7: Effect of V1 on shear capacity[23]

Vf vI(Shabaz.) Eqtn. I Eqtn. II

Table 2.1: Application of eqtns. I and II to the specimens tested by Shahbazker

v= 10V [v + 3000 p/(a/d) ]

Table 2.2: Application of Shahbazker's eqtn. to the results obtained by Ashour[23]

v= 10 r,,[V +3000 p/(a/d)5 + 70F]

Load-deformation predictions for

In attempting to model the behavior of the beams in this program of study, it is

* the peak load capacity for HSFRC beams to be predicted,

* the central span deflection behavior to be predicted, and

* to validate the application of the softened truss model to HSFRC beams.

~ vLIIIi$ rc- Crs

3.1 Deep beam shear model

Figure 3-2: Deep beam stress condition[27]

The subscripts z, y, d, and r correspond to the chosen axis .

Though no transverse reinforcement ( stirrups ) have been provided in this research

3.1.1 Equilibrium equations