Aci SP-155 - 1995 PDF

Testing of Fiber
--``,`,-`-`,,`,,`,`,,`---
EDITORS
D. J. Stevens, et. al.
Copyright American Concrete Institute
Provided by IHS under license with ACI
SP-155
No reproduction or networking permitted without license from IHS Not for Resale
A C 1 SP-L.55 95 W 0662749 0523086 622 =
Testing of Fiber
EDITORS --``,`,-`-`,,`,,`,`,,`---
D. J. Stevens, et. al.

SP-155
A C 1 SP-I155 95 = 0662749 0523087 5 6 9
DISCUSSION of individual papers in this symposium may be submitted in

accordance with general requirements of the AC1 Publication Policy to AC1
headquarters at the address given below. Closing date for submission of
discussion is January 1, 1996. All discussion approved by the Technical
Activities Committee along with closing remarks by the authors will be
published in the May/June 1996 issue of either AC1 Structural Journal or
AC1 Materials Journal depending on the subject emphasis of the indi-vidual
paper.
The Institute is not responsible for the statements or opinions expressed in

its publications. Institute publications are not able to, nor intended to,
supplant individual training, responsibility, or judgment of the user, or the
supplier, of the information presented.
--``,`,-`-`,,`,,`,`,,`---
The papers in this volume have been reviewed under Institute publication
procedures by individuals expert in the subject areas of the papers.
Copyright O 1995
AMERICAN CONCRETE INSTITUTE
P.O. Box 19150, Redford Station
Detroit, Michigan 48219
All rights reserved including rights of reproduction and use in any form or
by any means, including the making of copies by any photo process, or by any
electronic or mechanical device, printed or written or oral, or recording for
sound or visual reproduction or for use in any knowledge or retrieval system
or device, unless permission in writing is obtained from the copyright
proprietors.
Printed in the United States of America
Editorial production Victoria Lunick
Library of Congress catalog card number 95-76808

A C 1 SP-155 9 5 sl 0 6 6 2 9 4 9 0 5 2 3 0 8 8 4T5 9
PREFACE
Fiber Reinforced Concrete (FRC) is a composite material that possesses

many mechanical, physical, and chemical properties that are distinct from
--``,`,-`-`,,`,,`,`,,`---
unreinforced concrete. Recent advances in testing techniques, instrumenta-

tion, and interpretation of test results for FRC are the subject of the papers
included in this symposium volume. These papers were presented at two
technical sessions on the Testing of Fiber Reinforced Concrete, held during
the 1995 AC1 Spring Convention in Salt Lake City, Utah. These sessions
and this special volume were cosponsored by Committees 544, Fiber Rein-
forced Concrete; and 444, Experimental Analysis of Concrete Structures.
One mechanical property of great practical significance is FRC's super-
ior ability to absorb energy. Many other useful properties such as crack
propagation resistance, ductility, impact resistance, fatigue performance,
freeze-thaw resistance, and durability are directly or indirectly connected to
the composite's energy absorption capacity. Hence, reliable characterization
of energy absorption is essential for evaluation of material performance, as
well as for design. Static flexural tests have been popularly used to deter-
mine toughness values that represent a measure of energy absorption capa-
city, and interest in toughness characterization has resulted in numerous
standards for toughness testing. The first set of papers is dedicated to
discussions of recent developments on the subject.
The second session focuses on testing for several other unique proper-
ties of FRC including: restrained shrinkage testing, durability testing of
FRC with synthetic fibers, impact testing for toughness characterization,
early-age tensile testing, pullout resistance of embedded fibers, and impact
testing on full scale FRC structures.
The editors would like to thank the authors and presenters for their ex-
cellent contributions. All the papers included in this special volume were
peer-reviewed, per AC1 requirements, by researchers within the two spon-
soring committees, as well as by outside experts. We would like to thank the
reviewers, as well as the AC1 staff who have made this special volume a
timely contribution to the state-of-the-art on the testing of Fiber Reinforced
Concrete.
Dave Stevens, Nemy Banthia,

Vellore S . Gopalaratnam, and Peter C. Tatnall
Editors
Copyright American Concrete Institute ...

Provided by IHS under license with ACI 111
A C 1 SP-155 45 m O b b 2 î Y î 0523084 331
AC1 Committee 544

FIBER REINFORCED CONCRETE
James I. Daniel Vellore Gopalaratnam

Chairman Secretary
Shuaib H. Ahmad Carol D. Hays Ernest K. Schrader

M. Arockiasamy George C. Hoff Morris Schupack
P. N. Balaguru Roop L. Jindal Surendra P. Shah
Hiram P. Ball, Jr. C. D. Johnston George Smith
Nemkumar Banthia Mark A. Leppert Philip A. Smith
Gordon B. Batson Clifford MacDonald Parviz Soroushian
M. Ziad Bayasi Pritpal S. Mangat J i m D. Speakman
Marvin E. Criswell Henry N. Marsh, Jr. David J. Stevens
Daniel P. Dorfmueller Nicholas C. Mitchell Narayan Swamy
Marsha Feldstein Henry J. Molloy Peter C. Tatnall
Antonio V. Fernandez Dudley R. Morgan Ben L. Tilsen
Sidney Freedman Antoine E. Naaman George J. Venta
David M. Gale Antonio Nanni Gary L. Vondran
Melvyn A. Galinat Seth L. Pearlman Methi Wecharatana
Antonio J. Guerra Max L. Porter Gilbert R. Williamson
Lloyd E. Hackman V. Ramakrkhnan Spencer T. Wu
C. Geoffrey Hampson Ken B. Rear Robert C. Zellers
M. Nadim Hassoun D. V. Reddy Ronald F. Zoll0
AC1 Committee 444

EXPERIMENTAL ANALYSIS FOR CONCRETE STRUCTURES
--``,`,-`-`,,`,,`,`,,`---
Kirk A. Marchand John R. Hayes, Jr.

Chairman Secretary
Leslie A. Clark Mohsen A. Issa Rajan Sen

Fernando E. Fagundo Moussa A. Issa Kwok-Nam Shiu
Fikry K. Garas Theodor Krauthammer David J. Stevens
T.Russell Gentry Robert L. Nigbor Stuart E. Swartz
Harry G. Harris John E. Pearson George V. Teodoru
Cheng-Tzu Thomas Philip C. Perdikaris Richard N. White
Hsu Gajanan M. Sabnis David Z. Yankelevslq
Lorenzo Imperato

Provided by IHS under license with ACI IV
A C 1 SP-155 95 0662949 0 5 2 3 0 9 0 0 5 3
--``,`,-`-`,,`,,`,`,,`---
CONTENTS
PREFACE ........................................... ...

111
DEFLECTION MEASUREMENT CONSIDERATIONS IN

EVALUATING FRC PERFORMANCE USING ASTM C 1018
by C. D. Johnston . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
TOUGHNESS OF FIBER-REINFORCED HIGH-STRENGTH

CONCRETE FROM NOTCHED BEAM TESTS
by D. Jamet, R. Gettu, V. S. Gopalaratnam, and A. Aguado . . . . . . 23
COMPARATIVE TOUGHNESS TESTING OF FIBER

REINFORCED CONCRETE
by L. Chen, S. Mindess, D. R. Morgan, S. P. Shah,
C. D. Johnston, and M. Pigeon . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
MEASURING TOUGHNESS CHARACERISTICS OF SFRC -

A CRITICAL, VIEW OF ASTM C 1018
by D. E. Nemegeer and P. C. Tatnall . . . . . . . . . . . . . . . . . . . . . . . 77
EXPERIMENTAL R-CURVES FOR ASSESSMENT OF

TOUGHENING IN FIBER REINFORCED CEMENTITIOUS
COMPOSITES
by B. Mobasher, C. Y, Li, and A. Anno . . . . . . . . . . . . . . . . . . . . . 93
TEST METHODS FOR DURABILITY OF POLYMERIC

FIBERS IN CONCRETE AND UV LIGHT EXPOSURE
by P. Balaguru and K. Slattum . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
RESTRAINED SHRINKAGE TESTS ON FIBER REINFORCED

CEMENTITIOUS COMPOSITES
by N. Banthia, M. Azzabi, and M. Pigeon . . . . . . . . . . . . . . . . . . . 137
DIRECT TENSILE STRENGTH TESTING AT 6 HOURS OF

FIBER REINFORCED CONCRETE MORTAR FRACTIONS
by P. P. Kraai and G. L. Vondran . . . . . . . . . . . . . . . . . . . . . . . . . 153

No reproduction or networking permitted without license from IHS
V
Not for Resale
A C 1 SP-155 95 0662749 0523093 T 9 T
IMPACT TESTS ON CEMENT-BASED FIBER REINFORCED

COMPOSITES
by N. Banthia, K.Chokri, and J. F. Trottier . . . . . . . . . . . . . . . . . . 171
MEASUREMENT OF THE PULLOUT FORCE AT DIFFERENT

RATES OF LOADING
by A. Pacios and S. P. Shah . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
TESTING APPLIED TO THE EVALUATION O F DAMAGE

TO FRC AND OTHER MATERIAL SYSTEMS CAUSED BY
LARGE MISSILE IMPACT TO BUILDING ENVELOPES
DURING STORM EVENTS
by C. D. Hayes and R. F. Zoll0 . . . . . . . . . . . . . . . . . . . . . . . . . . . 217
FIBER REINFORCED CONCRETE TESTING FOR

--``,`,-`-`,,`,,`,`,,`---
PRACTICAL APPLICATION
by G. Spadea, R. Cava, D. Gallo, and R. N. Swamy . . . . . . . . . . . . 233
SI (Metric) TABLES. .................................. 241
INDEX ............................................ 243

vi
Not for Resale
A C 1 SP-1-55 95 Oh62949 0 5 2 3 0 9 2 9 2 6 I
SP 155-1
DefIectio n Measurement Considerat ions

in Evaluating FRC Performance
Using ASTM C 1018
by C. D. Johnston
Svnopsis: The issue of how the method of determining midspan deflection in

ASTM C1018 toughness tests influences first-crack strength, first-crack deflection,
toughness indices and residual strength factors is addressed by comparing results
obtained using the method now required in the current standard, which is based on net
midspan deflection determined as the nominal midspan deflection minus the average
of the deflections measured at the beam supports, with corresponding same-specimen
results based on nominal midspan deflection only which was not explicitly excluded
in earlier versions of the standard. The problem of dealing with the portion of load-
deflection relationship immediately after first crack when it is unstable is also
discussed.
The range of test specimens for which comparative data are reported includes
a series of third-point-loaded 500x150~150mm beams with three different steel fibers
ranging in length from 18 mm to 63 mm, and a second smaller series of 350x100~100
mm beams that allows for assessment of the effects of beam size and fiber alignment.
Fiber contents vary from 20 to 75 kg/m3 (0.25to 0.94% by volume). Also included
is a series of 350x100~100mm beams with a single type of fibrillated polypropylene
fiber of length 38 to 64 mm in amounts of 0.5 to 0.75% by volume.
The results illustrate the extent to which the Cl018 parameters 15,.Iio, 120,
R5,10,and are effective in distinguishing the performance of the various FRC
mixtures in terms of fiber type, geometry and amount. The index I5 is found to be
least effective and a case is made for greater emphasis on use of residual strength
factors, especially when employing the test to specify and control the quality
of FRC.
Keywords: Beams (supports); cracking (fracturing); deflection; fiber

reinforced concretes; fibers; polypropylene fibers; strength; tests;
toughness
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 IObb2949 0523093 862
2 Johnston
Colin Johnston is an AC1 Fellow, a past-president of the Alberta Chapter of AC1 and
a professor of civil engineering at The University of Calgary. He was awarded ACI's
Wason Medal for Materials Research in 1976. He was chairman of ASTM 0 . 4 2
on Fiber-Reinforced Concrete from 1980to 1990 and is a member of AC1 Committee
544 on Fiber-Reinforced Concrete.
INTRODUCTION
--``,`,-`-`,,`,,`,`,,`---
The issues of exactly how deflection should be measured and the possible
effects of different methods of determining deflection on the loaddeflection
relationship and the toughness parameters derived from it have been the subject of
much discussion and some controversy since ASTM C1018(1) was first introduced in
1984.
Since 1984 the Apparatus section of the standard has required deflection-
measuring equipmentto "accurately determine the net deflection of the specimen under
load exclusive of any effects due to seating or twisting of the specimen on its
supports, and the Procedure section of the standard has included the imperative
statement in mandatory language "Exercise care to ensure that the measured
deflections are the net values exclusive of any extraneous effects due to seating or
twistingof the specimenon its supportsor deformation of the support system". While
the intent of these statements was clear enough, the specifics of how it might be met
were contained in a non-mandatory note recommending the use of additional
deflection-measuring devices at each beam support, The 1984, 1985, 1989 and 1992
editions of the note also acknowledged that the increased number of deflection-
measuring devices makes the processing of data to obtain average net deflection more
complex and stated that a recommended correction procedure for drawing a tangent
to the initially concave upwards portion of the loaddeflection curve "allows the net
deflection to be obtained reasonably accurately. "
In a 1985state-of-the-artpaper on toughness(2), the writer questioned whether

deflection measurement at the midspan only, termed nominal deflection, was
"reasonably accurate" and showed that it was in fact largely responsible for the wide
variation in midspan deflections at first crack reported in various publications available
at the time. The paper also illustrated the effect on toughness indices for same
specimen loaddeflection relationships obtained using net and nominal midspan
deflection measurement (Fig. i), and acknowledged that nominal deflection
measurement at the midspan was a common and imperfect compromise which is
simpler and more convenient for routine use than net deflection measurement requiring
additional deflection-measuring devices at the supports along with calculations to
average the beam support deflections and subtract the average from the midspan
deflection.
Unfortunately, the notion conveyed by the note in the 1984 and 1985 editions
of ASTM C1018 that nominal deflection measurement was reasonably accurate and
good enough for most testing was widely believed until well after 1985. However,
in the 1989 edition the standard was modified to delete reference to this notion and
A C 1 SP-155 75 0 6 6 2 9 4 9 0523094 î T 9
Testing of FRC 3
replace it with wording stating that "Nominal deflections based only on midspan
measurements may be much larger than corresponding net midspan deflections
obtained by subtracting the average of the deflections measured at the two supports
from the corresponding nominal deflection at the midspan". Also added was the
comment that "Toughness indices based on nominal midspan deflections may be less
than the equivalents calculated using net midspan deflection". This non-mandatory
language stating the desirability of deflection measurement at the supports as well as
the midspan and the possibly significant effect on toughness indices remained
unaltered in the 1992 edition of ASTM C1018.
In 1994 the standard was modified to delete ail reference to testing based on
nominal midspan deflection measurement thus making even more explicit the
requirement to determine net deflection and to compute toughness indices and residual
strength factors solely on that basis. Detailed descriptions with photographs of two
alternative deflection-measuring systems for doing so were added, and a formula for
estimating the first-crack deflection in terms of the size of the test specimen and the
modulus of the concrete was introduced to help users confirm the validity of deflection
measurements.
Like most ASTM Standards, ASTM C1018 has evolved through consensus
and compromise between those who advocated the need for net deflection
measurement despite increased experimental complexity and those who argued for
permitting nominal deflection measurement because of experimental simplicity and
practicality. Naturally, fewer laboratories had the more complex equipment needed
to determine net midspan deflection, and those that did not have it tended to test
specimens using nominal deflection measurement despite failing to comply with the
intent of the standard. For example, in an interlaboratory comparison of data(3),
supposedly obtained according to ASTM C1018-89, only four of six participants
measured net midspan deflection while the two others measured nominal deflection.
Despite much discussion and some controversy over the issue of deflection
measurement, there is little published data comparing results obtained using net
deflection measurement with those obtained using nominal deflection measurement.
This paper makes same-specimen comparisons for a variety of steel and polypropylene
fibers at different fiber contents. The results reflect the evolution of ASTM C1018
from 1984 when only toughness indices Is and Il0 and the ratio Ilo/Is were reported,
to 1989 when the residual strength factor R S , became
~ ~ a requirement and the index
120 was highlighted instead of 130 as a first option along with the residual strength
factor R10,20 This followed introduction of the concept of residual strength factor
in 1986(4).
Since the validity and accuracy of deflection measurements influences the

values of toughness indices (Fig. 1) along with the values of residual strength factors
derived from them, the issue of how effective the various ASTM C1018 parameters
are in distinguishing the performance of FRC's in terms of fiber type, geometry and
content is also addressed in the paper. This issue is also a subject of much discussion
and some controversy since the standard has been used in specifications to assess FRC
performance using parameters ranging from the lowest permissible end-point
deflection criterion and toughness index, Is, to higher end-point deflection criteria and
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-I155 95 IO b 6 2 9 4 9 0523095 635
4 Johnston
corresponding residual strength factors, such as R10,5@ The merits of low-end point
versus higher end-point deflections and the significance of toughness indices versus
residual strength factors can also depend on the response of the testing system. The
discussion also deals with the effect of the testing system response rate on the load-
deflection function immediately following first crack when there is sometimes a rapid
and unstable decrease in load and increase in deflection, particularly for low fiber
contents.
EXPERIMENTALPROGRAM
In the first set of tests reported, the fiber distributor who sponsored the tests
in 1987 wished to compare the performance of several types of steel fiber available
in Canada in the context primarily of the relatively low fiber contents and low to
medium strength concrete matrices associated with industrial floor slab applications.
In choosingthe specimen size it was recognized that stiff fibers of length 50 to 63 mm
would be subject to significant preferential fiber alignment if evaluated using
300x100~100mm beams and that the ASTM Cl018 requirementfor thicksectionsthat
the minimum specimen dimension be at least 3 times the fiber length would be
severely violated. In order to obtain results more representativeof thick sections and
meet more closely the specimen size/fiber length minimum of 3, the heavier less
convenient 450x150~150mm beam size was employed in the majority of the tests,
although additional 300x100~100mm beams were tested in two cases to get an idea
of the effect of specimen size and associated preferential fiber alignment on test
results.
In the second set of tests also in 1987 another fiber distributor who wished to
evaluate the performance of fibrillated polypropylene fibers of length 38 to 64 mm
chose the more economical 350x100~100mm specimen size recognizing that flexible
polypropylenefibers would probably be less subject to the effects of preferential fiber
alignment.
Both sponsors selected specific fiber types, lengths and amounts to be

evaluated in concretes of specified strength 25 MPa or 30 MPa as shown in Tables 1
and 2, and both wished to have the tests conducted with deflection measured strictly
in accordance with intent of C1018-85 "that the measured deflections are the net
values exclusive of any extraneous effects due to seating or twisting of the specimen
on its supports or deformation of the support system". This was accomplished using
the 3-transducer arrangement (Fig. 2), now included in the 1994 edition of ASTM
C1018, in which the deflection reproduced on the x-y plotter is the net deflection
obtained by subtracting the voltage representing average of the deflections at each
beam support from the voltage representing the midspan deflection. However, since
the writer was also interested in comparing net deflection with nominal midspan
deflection and in the associated comparison of toughness parameters, an additional x-y
plotter was added to allow same-specimen plots of load versus net deflection and load
versus nominal deflection. Sets of four specimens were tested for each fiber-matrix
combination, and the results in Tables 1 and 2 are the mean values for each set
calculated in accordance with C1018-89 for toughness indices 15, Il0 and, I and
residual strength factors R5,io and R10,20. Values based on net deflection
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 IO662949 0523096 571 I
Testing of FRC 5
measurement are in bold print while those based on nominal deflection are in normal
print.
DISCUSSION OF RESULTS
The discussion is confined to toughness indices (I) and residual strength

factors (R). First-crack parameters are included in Table 1 and 2. Generally, first-
crack deflections for nominal deflection measurement are 2.8 to 3.2 times larger than
corresponding values for net deflection measurement, while first-crack strengths are
essentially equal for both methods of deflection measurement.
Effect of Deflection-Measurine Techniaue
The effect on I and R values of calculating them on the basis of net deflection
versus calculating them on the basis of nominal deflection varies with fiber type,
geometry and amount as illustrated by the specimen examples in Fig. 3 and 4.
Starting with 450~150x150mm beams, the three relationships for steel fibers
in Fig. 3 illustrate the main possibilities.
The case of gradual stable strainsoftening after first crack (Fig. 3a) is perhaps
the easiest to explain because of its approximately constant slope. In this case, values
of toughness indices denoted by I' are derived from the relationship obtained using
nominal midspan deflection (broken line in ali figures), and they are less than the
indices denoted by I determined in the proper way from the relationship obtained
using net midspan deflection (solid line in all figures). Clearly, this results from the
fact that the average vertical ordinate for the total area representing the numerator in
any index decreases as the end-point deflection moves to the right, making this
ordinate for any end point on the broken line less than the ordinate for the
corresponding end point on the solid line, while the horizontal abscissae for the
numerator representing total areas and the denominator representing first-crack areas
remain in the same proportion for both broken and solid lines. Thus, Is', Il,' and
Izo' are less respectively than 15: Il, and 120. Likewise, residual strength factors,
which are in fact the average ordinate between consecutive end-points divided by the
first-crack ordinate, decrease as the end points move to the right, so and
R'lo,20are less respectively than R5,10and R,,,,,
Considering the special case of elastic-plastic behaviour where both broken
and solid lines are horizontal from first crack, it should readily be understood that I'
and R' values will be the same as I and R values because the vertical ordinate remains
the same for all parameters. Extending the analysis to the case of gradual stable
strain-hardening where the ordinate actually increases as the end-point deflection
moves to the right (Fig. 3b), it is expected that I' and R ' values will exceed I and R
values. This situation is relatively uncommon, but is known to happen for certain
type-amount combinations of hooked-end fibers as in Fig. 3b.
The third case is unstable strain-softening immediately after first crack

followed by stable nearly plastic behaviour thereafter (Fig. 3c). In these cases the
average vertical ordinate for total area is again less for the broken line relationship
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 IObb29Y9 0523097 408
--``,`,-`-`,,`,,`,`,,`---
6 Johnston
than for the solid line one, so I' values are less than I values. However, when the
portions of the relationships between consecutive end points used to calculate R' and
R values are horizontal (plastic) R' and R values may be nearly equal, for example
~ ~ Rg,10in Fig. 3c.
R ' S , and
Similar situations arise with the 300x100~100mm beams. For example, with
steel fibers (Fig. 4a) the extensive plastic portion of both relationships following first
crack accounts first for R10 2o almost equal to R S , and ~ ~ second for almost
equal to Rs Polypropyiene fibers (Fig. 4b) can also give rise to the situation
where I' vdues are much less I values while R' and R values are not substantially
different.
Effect of Unstable Strain-Softening Immediatelv After First Crack
Some fiber type-amount combinations, particularly those using low fiber

contents, are associated with a rapid decrease in load and increase in deflection
immediately after first crack which occurs so quickly that the response rate of the load
and deflection-recording system may not be fast enough to reflect what is really
happening. The relationships in Fig. 3c and 4b are examples typically obtained under
open loop control conditions.
Part of Fig. 3c enlarged to highlight the unstable region immediately following

fist crack is shown as Fig. 5a. The unstable region in question is AY in general,
although a portion of it, AX, appears stable initially from the clearly defined track of
the pen on the plotter, while the portion XY is poorly defined with only a faint linear
pen track. In Fig. 5b the transition corresponding to the change in slope (specimen
stiffness) at X is not discernible and the whole of AY is likely unstable. In the
limiting case of concrete without fibers, the beam breaks suddenly at A and has no
residual strength thereafter, so the load drops instantaneously to zero before deflection
can increase, AZB in Fig. 5, even though the plotter records a line somewhere
between AXY and AZB. The same behavior applies at very low fiber contents. With
sufficient fibers there is a deflection increase as the load drops from first crack to the
residual value that can be sustained stably over a period of time, as depicted by the
portion of the loaddeflection relationship to the right of Y. Dealing with uncertainty
regarding the position of AY which could be anywhere between AXY and AZY is the
problem. Obviously, the same issue arises with regard to A'Y' (Fig. 5) for nominal
deflection measurement.
Toughness indices calculated in the normal way using AXY will obviously
decrease if recalculated using AZY. In Fig. 5a the effect will be greatest for 15,
derived from the area AXYCDB, and relatively less for Il0, derived from AXYEFB
in which the portion CEFD is unaffected by the position of AY, and of course less
still for Izo. Perhaps the most important point is that Rs,lo, which is based only on
the area CEFD, is unaffected by any uncertainty in the position of AY because C is
the right of Y, and likewise R10,20(CC63 fibres in Table 3). This applies also to
Fig. 4a. However, in Fig. 5b the effect of uncertainty regarding the position of AY
extends to R5,io because the area CEFD is affected since C is left of Y.
Nevertheless, which is based on the area to the right of EF is unaffected @E18

A C 1 SP-155 95 I0662'749 0523098 3 4 q I
Testing of FRC 7
fibres in Table 3). This applies also to Fig. 3a, 3b and 4b, but Fig. 5b represents the
worst case observed for the data set in terms of C being furthest to the left of Y.
Since the position of the unstable portion of the load-deflection relationship

AXY is probably influenced by both the response rate of the data-recording system
and the stiffness of the testing frame and that of the specimen, an approach to
performance assessment that eliminates this uncertainty is desirable. Thus, it is
imperative that the widely practised tendency to highlight 15, which is most severely
affected, and not consider parameters which are less affected, such as 12,. or usually
not affected at all, such as must change. Reaching the conclusion in the 1991
interlaboratory study (3) that "ASTM C1018 toughness indices are observed to be
relatively insensitive to fiber type, volume fraction and specimen size", while
highlighting 1s and ignoring R values in the published analysis, despite the fact that
reporting of R5,10was mandatory in the 1989 edition of the standard and R10,20was
identified as optional, presented an incomplete impression of the effectiveness of
ASTM Cl018 for distinguishing performance in terms of fiber parameters.
Effectiveness of Cl018 Parameters in Distinguishing FRC Performance
While the five single-specimen examples in Fig. 3 and 4 permit differences

in performance to be distinguished in terms of the appearance of the load-deflection
relationships and the numerical values of parameters such as 12, (range 10.2 to 17.3)
and R10,20.(range 38 to 87), the effects of variables like fiber geometry, amount and
specimen size are best distinguished using the mean values from Tables 1 and 2
plotted graphically as in Fig. 6 for toughness indices and in Fig. 7 for residual
strength factors. To make the indices graphically comparable, the scales for Is, I,,,
and 12, are chosen to correspond to the values of 5, 10 and 20 corresponding to
--``,`,-`-`,,`,,`,`,,`---
elastic-plastic or yield-like behavior which is the reference level against which actual
performance is usually compared (Appendix XI of ASTM ClOl8). Accordingly, the
1s scale is twice as large as the Il0 scale and four times as large as the I,, scale.
Naturally, R5,10 and R,,,,, are plotted to the same scale as they both have the same
range of O to 1 0 .
Effect of Fiber Geometry and Amount--Three varieties of steel fibers

designated CC63, EE18 and HE50 are compared in terms of fiber content for
450x150~150mm beams (Fig. 6 and 7). A fourth, CW60, is included at one fiber
content. Both the solid line trends (based on net deflection) and the broken line trends
(based on nominal deflection) are definitive in illustrating the importance of fiber
content, but the values in the latter case are lower except for some HE50 concretes
consistent with the reasoning given earlier in discussing Fig. 3 and 4. Subsequent
discussion is limited to the solid line trends which represent proper accurate
measurement of deflection. However, they are subject to the uncertainty associated
with the unstable AXY (Fig. 5) portion of the loaddeflection relationship, especially
for low fiber contents, the effect of which is to decrease some I values, particularly
15, below the values plotted.
Since the graphs for toughness indices (Fig. 6) are scaled vertically to make
the 15, Il,, and 12, values graphically comparable, the steeper the slope the better the
distinction of performance in terms of fiber content. Clearly, Is is least effective and

A C 1 SP-155 95 = Obb29Y9 0523099 2 8 0
8 Johnston
--``,`,-`-`,,`,,`,`,,`---
120 is the most effective with Il0 almost as effective as 120. In terms of residual
strength (Fig. 7), both R5.10 and R10,20 appear effective in making the distinction by
fiber content. However, is in fact better because the uncertainty associated
with the unstable AXY portion (Fig. 5) of the ioaddeflection relationship which
influences some R5, values at low fiber contents is without exception eliminated for
R10,20*
The importance of fiber geometry for steel fibers reflects the influence of
aspect ratio and improvements to pullout resistance by use of crimping (CC and CW)
and hooked ends (HE) or enlarged ends F E ) . For example, at 0.5% by volume or
40 kg/m3 of fibers values are about 88 for HE 50, 82 for CW 60,62 for CC
63 and 42 for EE 18 over a range of aspect ratio of 100 for HE 50 fibers to 38 for
EE 18 fibers.
Concrete matrix strength within the limited 25 to 30 MPa range examined has
little influence on I or R values.
Effect ofFiber five and Amount--Steel and polypropylenefibers are compared

in terms of fiber content for 300x100~100mm beams (Fig. 8 and 9 left). Differences
between the solid line trends (based on net deflection) and the broken line trends
(based on nominal deflection) are similar to those in Fig. 6 and 7, and are consistent
with the reasoning given earlier in discussing Fig. 3 and 4. Only the solid line trends
are discussed further.
Once again, the manner in which the graphs in Fig. 8 are scaled means that
slope is an indicator of the effectiveness of each toughness index in distinguishing
performance. Clearly, 120 and Il0 are again more effective for this purpose than 15
(Fig. 8 left), just as in Fig. 6, For residual strength, Rs,lo and are both
effective (Fig. 9 left), and illustrate the expected influence of fiber content and in the
case of the polypropylene fibers the effect of length or aspect ratio. For example, at
0.5% by volume of fibers the R10,20 values are about 80 for the CC 63 steel fiber,
37 to 39 the FP 38 and FP 64 polypropylene, and, by interpolation on Fig. 7, for the
EE 18 fiber about 42 plus a small amount attributable to specimen size effect (see next
section).
Effect ofSmcimen Size--From the limited data available, a significant increase

in both I and R values is associated with reducing beam size from 450x150~150mm
to 300x100~100mm for the long stiff CC 63 steel fibers when preferential fiber
alignment by the mold surfaces is significant (Fig. 8 and 9, right). This is to be
expected since the ratio of specimen cross-section to fiber length is 1.6 compared with
the minimum of 3 required for thick sections in ASTM C1018. For the shorter EE
18 fiber, where the corresponding ratio is 5.3, the fiber alignment effect is probably
minimal as suggested by the comparison at 75 kg/m3. This is consistent with the
results of an earlier study (5) to examine the effect of preferential fiber alignment on
test results using fibers of length 76 mm and 25 mm in molded and sawn specimens
of 100 mm cross-section.

A C 1 SP-155 9 5 H 0662949 0 5 2 3 1 0 0 ô 2 2
Testing of FRC 9
Precision of Results
Within-batch coefficients of variation for each set of four specimens are given
in Tables 4 and 5 for the two different specimen sizes.
While there is no conclusive link between the individual values in terms of

fiber type, geometry or amount, the highest values are associated mainly with low
fiber contents of CC63 steel fibers. These are the largest in individual size. At low
fiber contents this means the lowest number of fibers per unit volume of concrete,
which may contribute to more marked nonuniformity in the fiber distribution and a
more variable FRC than for the other fibers.
The most meaningful numbers are the mean values which indicate the level
of precision that should be expected on average when testing multiple sets of
specimens. In this regard, there are no major differences for the two specimen sizes,
--``,`,-`-`,,`,,`,`,,`---
which if combined total 24 sets, and it is clear that the parameters which it has been
argued should be highlighted more in future testing, that is 120, RS,JO,and R10,20,
can be evaluated with reasonable precision. The highest mean within-batch coefficient
of variation is about 13%for and 19 of 24 values are less than 18%. It must
be recognized than the higher the end-point deflection the greater the variability of the
results, but in view of the arguments previously discussed it is pointless to rely on Is
determinations simply because of better precision.
The mean values reported in Tables 4 and 5 compare closely with values for
another recent data set (6) analyzed in the same way for 300x100~100mm beams
tested at various ages and machine stroke rates.
CONCLUSIONS
1. Effectively utilizing ASTM Cl018 to evaluate the performance of FRC and

distinguish the importance of the material type, geometry and amount of the
fibers depends on recognition of the following limitations:
(i) Toughness indices and residual strength factors derived using nominal
deflection measurement are usually less, often considerably less, than
values derived using net deflection measurement as required by the
standard. However, there are exceptions where nominal deflection
measurement may produce values equal to or slightly more than
values obtained according to the standard.
(i¡) Some fiber-matrix combinations, particularly with low fiber contents,

exhibit rapid load decrease with deflection increase after first crack
that is unstable and may not be detected accurately by the load and
deflection-recording system. The uncertainty associated with this
unstable portion of the relationship affects the toughness index I5 most
severely, and its effect lessens with increasing end-point deflection,
and is usually quite small for 120 and higher deflection indices.
Evaluating performance in terms of residual strength factors can

A C 1 SP-155 95 W Obb29i49 0523101 769
10 Johnston
eliminate the effect of this uncertainty entirely when using R10,20,

and the effect on Rg,10 is often quite minimal. It exists when C or
E are to the left of Y in Fig. 5 and is eliminated when C and E are
to the right of Y in Fig. 5.
(iii) The use of the preferred 300x100~100mm beam with long rigid
fibers like steel tends to produce toughness indices and residual
strength factors greater than for otherwise comparable 450x150~150
mm beams.
2. The data set discussed in the paper which is based on 22 fiber-matrix

combinations and 88 tests demonstrates that some ASTM C1018 parameters
are more effective than others for distinguishing the effects of fiber type,
geometry and amount on the performance. The index I5 is certainly least
effective. Both Il0 and 120 (which is a required test parameter in the 1994
standard) are much more effective.
The index I20 is least influenced by uncertainty regarding the position of the
portion of the loaddeflection immediately following first crack in cases where
rapid unstable behaviour occurs. The fact that the effect of this uncertainty
can be minimized or eliminated entirely by calculating residual strength
factors, coupled with the fact that these factors are more easily understood
than toughness indices and have more potential for direct application in
strength-based design (7), is an argument for less emphasis on I values and
more emphasis on R values, particularly R10,20, in using the test method to
specify and control the quality of FRC. Certainly, it is imperative that the
widely practised tendency to highlight I5 and ignore R5,10 and must
change.
--``,`,-`-`,,`,,`,`,,`---
3. The conclusion that parameters such as I?o. and R10,20, and to a slightly lesser
degree Il0 and Rg,lO> are effective in distinguishingperformance differences
by fiber type, geometry and amount is not unique to these two series of
specimens or to the particular equipment used. Similar results were reported
(8) in the same format as Fig. 6, 7, 8 and 9 for a series of 108 tests on 18
steel fiber-rnatrix combinations performance on different equipment using
750x150~100mm beams and supervised by the writer in Stockholm in 1989.
However, both investigations employed the 3-transducer deflection-measuring
system (Fig. 2) rather than the rectangular jig arrangement identified as an
alternative in the 1994 edition of ASTM ClOl8.
ACKNOWLEDGEMENT
Financial support provided by Domecrete Canada Ltd., Elsro Construction

Products, Forta Corporation, and the Canadian Government I M P Program is
gratefully acknowledged, along with the efforts of Geoff Hampson, Brian Richardson
and Bob Zellers in planning the program and providing fibers.

A C 1 SP-155 95 0 6 6 2 9 4 9 0 5 2 3 3 0 2 bT5
Testing ofFRC 11
REFERENCES
1. ASTM C1018, Standard Test Method for "Flexural Toughness and First-
Crack Strength of Fiber-Reinforced Concrete (Using Beam with Third-Point
Loading)", ASTM Standards Vol. 04.02, Concrete and Aggregates, 1992.
2. Johnston, C.D., "Toughness of Steel Fiber-Reinforced Concrete", Proceedings
of a U.S.-Sweden Joint Seminar, Swedish Cement and Concrete Institute, Ed.
S.P. Shah and A. Skarendahl, Elsevier Applied Science, London, 1986, pp.
333-360.
3. Gopalaratnam, V.S., et. al., "Fracture Toughness of Fiber Reinforced
Concrete", AC1 Materials Journal, Vol. 88, July-August 1991, pp. 339-353
and Discussion by C.D. Johnston, May-June 1992, pp. 304-309.
4. Johnston, C.D. and Gray, R.J., "Flexural Toughness and First-Crack Strength
of Fiber-Reinforced Concrete Using ASTM Standard Cl018", Proceedings of
Third International RILEM Symposium of Developments in Fibre Reinforced
Cement and Concrete, Ed. R.N. Swamy, University of Sheffield, 1986, Paper
5.1.
5. Johnston, C.D., "Effects on Flexural Performance of Sawing Plain Concrete
and of Sawing and other Methods of Altering the Degree of Fiber Alignment
in Fiber-Reinforced Concrete", ASTM Cement, Concrete and Aggregates,
CCAGDP, Vol. 11, No. 1, Summer 1989, pp. 23-29.
6. Johnston, C.D., "Effects of Testing Rate and Age on ASTM Cl018
Toughness Parameters and Their Precision for Steel Fiber-Reinforced
Concrete", ASTM Cement, Concrete and Aggregates, CCAGDP, Vol. 15,
No. 1, Summer 1993, pp. 50-58.
7. Johnston, C.D., "Fiber-Reinforced Concrete" Chapter 5 1 of Significance of
Tests and Properties of Concrete", ASTM STP 169C, 1994, pp. 547-561.
8. Johnston, C.D. and Skarendahl, A., "Comparative Flexural Performance
Evaluation of Steel Fiber-Reinforced Concrete According to ASTM C1018
Shows Importance of Fiber Parameters", RILEM Materials and Structures,
Vol. 25, No. 148, May 1992, pp. 191-200.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0 6 6 2 9 4 9 0523303 5 3 1
12 Johnston
TABLE 1 - TEST RESULTS FOR 450 x 150 x 150 mm BEAMS (VALUES BASED ON NET
DEFLECTION SHOWN BOLD)
= - -
Matrix Fiber Amount Vol First-Cmck Toughness Indices Residual
Codeb kg/m3 x - -- -
Stren Factors
25
- --
---- -
O
---- Oft 6, '5 '10 -
120
-
Rs.ia p10.20
O 4.45 0.052
0.168
25 CC63 25 0.32 4.53 0.061 4.11 6.27 10.41 43.2 41.4
O. 179 3.64 5.67 8.38 40.6 27.1
25 CC63 40 0.51 4.68 0.060 4.38 7.67 13.90 65.8 62.3
0.181 3.95 6.91 11.16 59.2 42.5
25 CC63 55 0.70 4.88 0.060 4.66 8.63 15.95 79.4 73.2
0.193 4.29 7.80 14.09 70.2 62.9
25 EE18 25 0.32 4.44 0.060 4.40 6.7ï 10.08 47.4 33.1
0.174 3.61 5.13 7.01 30.4 18.8
25 EE18 50 0.64 4.80 0.064 4.54 7.66 12.45 62.4 47.9
0.186 3.71 5.87 8.64 43.2 27.7
25 EE18 62.5 0.80 5.15 0.060 4.60 8.12 14.54 70.4 64.2
0.189 3.97 6.94 11.16 59.4 42.2
25 @E18 75 0.96 5.11 0.059 4.64 8.39 15.39 75.0 70.0
0.193 4.10 7.34 12.36 64.8 50.2
30 EE18 62.5 0.80 5.33 0.062 4.64 8.24 14.27 72.0 60.3
0.193 3.95 6.37 9.07 48.4 27.0
30 EE18 75 0.96 5.62 0.065 4.64 8.33 15.07 73.8 67.4
0.216 4.07 7.03 10.79 59.2 37.6
--``,`,-`-`,,`,,`,`,,`---
30 CC63 40 0.51 5.01 0.052 4.68 8.30 14.76 72.4 64.6

0.184 3.94 7.09 11.52 63.0 44.3
30 CC63 55 0.70 5.16 0.059 4.67 8.85 16.97 83.6 81.2
O. 189 4.41 8.32 14.36 78.2 60.4
25 HE50 20 0.25 4.68 0.054 4.44 7.17 11.27 54.6 41.0
0.175 3.56 5.81 10.38 45.0 45.7
25 HE50 33 0.42 4.47 0.054 4.49 7.73 14.70 64.8 69.7
0.165 3.95 7.66 13.95 74.2 62.9
30 HE50 40 0.51 4.86 0.053 4.71 8.71 17.50 80.0 87.9
0.175 4.38 8.90 18.13 90.4 92.3
30 cw60 40 0.51 5.17 0.053 4.63 8.54 16.78 78.2 82.4
0.197 4.31 8.48 15.88 83.4 74.0
30 O O 5.17 0.058
= - 0.186 - -
= - - =
a - Specified Compressive Strength in MPa
- CC is Crimped Crescent-Shaped Steel, CW is Crimped Wire,
HE is Hooked-End Wire, E@is Enlarged-End Slit Sheet Metal.
Numbers are Nominal Length in mm.

A C 1 SP-155 95 O b 6 2 9 4 9 0 5 2 3 3 0 4 478
Testing of FRC 13
TABLE 2 - TEST RESULTS FOR 300 x 100 x 100 mm BEAMS (VALUES BASED ON NET
DEFLECTION SHOWN IN BOLD)
Matrix
Typea
-
I
Fiber
Code
b
1
Amount
kglm3
x s t - C r a c k
-6fc
Toughness Indices
--
---
-
I20 Rs,io
Residual
Strength
- Factors
R10.20
25 CC63 25 0.32 4.81 0.039 4.45 7.40 12.80 59.0 54.0

0.119 3.66 6.28 10.44 52.4 41.6
25 CC63 40 0.51 4.96 0.040 4.68 8.68 16.69 80.0 80.1
0.114 4.36 8.33 15.51 79.4 71.8
30 EE18 75 0.96 5.86 0.044 4.76 8.61 15.41 77.0 68.0
0.134 4.12 6.97 10.05 57.0 30.8
30 Fp38 4.55 0.50 5.58 0.044 4.12 6.22 9.92 42.0 37.0
0.123 3.45 5.35 9.21 38.0 38.6
30 FP38 5.91 0.65 5.86 0.047 4.12 6.29 10.19 43.4 39.0
0.133 3.43 5.44 9.30 40.2 38.6
30 FP38 6.83 0.75 5.81 0.050 4.19 6.84 11.76 53.0 49.2
0.135 3.51 6.02 10.84 50.2 48.2
30 FP57 5.91 0.65 5.53 0.045 4.30 6.95 11.66 53.0 47.1
0.126 3.45 5.90 11.21 49.0 53.1
30 FP64 5.91 0.65 5.69 0.046 4.34 6.91 11.51 51.4 46.0
0.127 3.51 5.90 10.97 47.8 50.7
30 FP64 4.55 0.50 5.93 0.049 4.08 6.24 10.10 43.2 38.6
0.130 3.41 5.39 9.29 39.6 39.0
30 O O 5.71 0.047
0.124 I-
--
a - Specified Compressive Strength in MPa
- CC is Crimped Crescent-Shaped Steel,
EE is Enlarged-End Slit Sheet Metal,
FP is Fibrillated Polypropylene. Numbers are Nominal Length in mm
TABLE 3 - COMPARATIVE I AND U VALUES CALCULATED FROM AREAS BOUNDED BY AXY AND
Alï (Fig. 5)
Fiber Amount Boundar Is I10 I20 R5,10 R10,20

Code kg/m3 Y
CC63 25 AXYa 4.11 6.27 10.41 43.2 41.4

AZYa 3.39 5.54 9.69 43.0 41.5
EE18 25 AXYb 4.40 6.77 10.08 41.4 33.1
AZïb 2.37 4.09 1.39 34.4 33.0
--``,`,-`-`,,`,,`,`,,`---

= - ~ ~~~
MatliX
Type*
Fiber
Codeb
Amount
kg/m3
Vol
96
--
Toughness Indices
~1 m
I
h
:e
t-
Residual
I
Factors
-- - - - -
- 6fc Ufc 15 I I10 I 120 I %,i0
-
R10,20
25 CC63 25 0.32 2.3 3.6 3.3 10.1 13.6 25.0 19.0

25 CC63 40 0.51 4.4 4.6 7.9 15.1 21.6 24.9 29.7
25 CC63 55 0.70 3.1 13.1 1.1 7.1 16.0 14.6 26.6
25 EEl8 25 0.32 8.1 10.2 1.9 1.8 3.3 3.6 11.3
25 EE18 50 0.64 2.6 3.6 2.2 4.3 3.6 7.5 3.5
25 EE18 62.5 0.80 4.3 2.6 1.0 1.6 2.2 2.4 3.0
25 EE18 75 0.96 4.2 4.0 1.2 3.7 6.5 6.8 9.9
30 EE18 62.5 0.80 3.0 1.2 1.1 4.3 8.5 8.9 14.4
30 EEl8 75 0.96 1.3 12.7 1.4 3.6 7.6 7.2 12.6
30 CC63 40 0.51 3.8 8.2 0.9 1.9 6.2 4.2 12.3
30 CC63 55 0.70 3.5 4.4 2.1 6.8 10.4 12.1 14.5
25 HE50 20 0.25 5.4 12.5 0.9 2.4 6.5 5.0 17.6
25 HE50 33 0.42 8.0 9.9 3.4 5.4 8.2 9.4 13.1
30 HE50 40 0.51 7.1 10.3 1.1 2.1 5.4 4.7 9.0
-- -30 CW60 40
Mean
0.51
Values
-
3.5 8.0
7.3
1.8
2.1
3.3
4.9
4.4
8.3
6.0
9.5
6.2
13.5
4.7
=
* - Specified CompressiveStrength in MPa
- CC is Crimped Crescent-Shaped Steel, CW is Crimped Wire.
HE is Hooked-End Wire, EE is Enlarged-End Slit Sheet Metal.
Numbers are Nominal Length in mm.
TABLE 5 - WITHIN-BATCH COEFFICIENTS OF VARIATION FOR SETS OF FOUR 300 x 100 x

100 mm BEAMS
Matrix Fiber Amount Vol First-Crack Toughness Indices Residual

Typ’ Codeb kglm3 %
-- - Strenj Factors
Ufc *fC I
I
15 I
I
I10
-
120
--
& R10.20
25 CC63 25 0.32 1.5 14.5 18.1 23.5
25 CC63 40 0.51 1.7 7.2 10.6 8.7
30 EE18 75 0.96 3.5 13.0 5.9 7.8 7.5 11.0
30 FP38 4.55 0.50 10.7 18.6 5.4 11.5 6.3
30 FP38 5.91 0.65 6.1 10.3 15.0 17.6
30 FF’38 6.83 0.75 3.9 3.0 5.2 5.4
30 FP57 5.91 0.65 4.1 7.8 5.3 17.4
30 FP64 5.91 0.65 2.6 9.8 14.3 13.9
30 FP64 4.55 0.50 5.3 5.4
- 8.9 7.2
Mean Values 4.4

-=
7.9
10.7 12.3
FF’ is Fibrillated Polypropylene. Numbers are Nominal Length in mm.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 m 0662949 0523306 240 m
Testing of FRC 15
20
15
z
Y
I
o IO
a
3
5
I
O
DEFLECTION -mm
Fig. 1-Load-deflection relationships and associated toughness indices I, and I,, compared on
the basis of nominal versus net deflection in 1985 (2)
--``,`,-`-`,,`,,`,`,,`---
Fig. 2-Flexural testing equipment with 350 x 100 x

Copyright American Concrete Institute 100 mm specimen on a 300 mrn span showing the 3-
transducer arrangement Not
No reproduction or networking permitted without license from IHS to fordetermine
Resale net deflection
A C 1 SP-355 75 IObb2747 0523307 387
16 Johnston
DEFLECTION
Fig. LSingle-specimen load-deflection relationships for 450 x 150 x 150 mm beams showing
end-point deflections and derived I and R values for nominal deflection (broken) and net deflec-
tion (solid) with 75 kg/m3, 0.94 percent volume of EE 18 steel fibers (top), 40 kglm3, 0.5 percent
volume of HE50 steel fibers (center), and 25 kg/m3, 0.31 percent volume of CC63 steel fibers
--``,`,-`-`,,`,,`,`,,`---

(bottom)
A C 1 SP-1.55 7 5 O 0 6 6 2 7 ~ 90.5231108 O13 O
Testing of FRC 17
F I BERS C C 63- 4O
, kg/m
Rs.10 = 88 R 10.20 = 87
15- 4.5
%,IO - 85 Río. 20 - 75
Iio * 16.3
I I
I I
4 0.1mm+
8’ 10.58’
-FIBERS FP38-4.6kglm3
R5.10t 4 5 R10.20 = 3 8
h!
1/20 9.6
1- 10.2 \
--``,`,-`-`,,`,,`,`,,`---
I I
I I
I l
n
- 8 318 5.523 10.58
8 38‘ 5.58’ 10.58‘
40.1
I
DE F LECT ION
Fig. &Single-specimen load-deflection relationships for 300 x 100 x 100 mm beams showing
end-point deflections and derived I and R values for nominal deflection (broken) and net
deflection (solid) with 40 kglm3, O5 percent volume of CX 63 steel fibers (top) and 4.55 kg/m3,
05 percent volume of FP 38 polypropylene fibers (bottom)

A C 1 SP-155 75 m 0662949 0523309 T 5 T
18 Johnston
FIBERS C C 6 3 - 25 kg/m3
1
.
1
.... \
'. 1
1-
\ Y'
--L--,--
E -
I
l
I
I
I F 1 -
DEFLECTION
Fig. L E f f e c t of uncertainty about the nature of load-deflection relationship immediately

following first crack on I and R values
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 75 0 6 6 2 9 4 9 0523110 771 I
Testing of FRC 19
Fig. &Effects of fiber geometry and amount on i values based on nominal deflection (broken)
--``,`,-`-`,,`,,`,`,,`---
and net deflection (solid) for 450 x 150 x 150 mm beams with four different steel fibers
A C 1 SP-155 95 0662949 0523111 608
20 Johnston
100
i.,25
FIBER-CC63
MATRIX
Oi5
FIBER VOLUME FRACTION
0.Y5,Oi25 0;5
FIBER-€EI8
MATRIX
0.75
-%
I.?
-
0.25
FIBER
HE-50-0.
0.5
J
8 90 25MPa- O 0 25MPa-0. cw-60- o
30MPO - A A 30 MPo - A A
.Pc"
8
R
--``,`,-`-`,,`,,`,`,,`---
K 70
d
a
O
I-
2 60
LL
E 50
a
z
h o
E /
/
30
d'
20
0
$90 P
I
8
o6 80
K
8 70
s2
60
I
I-
<3
50
K
!- c/
u)
-
40
30 'O
I_
1 I I 1 1
25 40 55 25 50 75 20 40
FIBER C O N T E N T - k g /m3
Fig. 7-Effects of fiber geometry and amount on R values based on nominal deflection (broken)
and net defledion (solid) for 450 x 150 x 150 mm beams with four different steel fibers

Testing of FRC 21
16
.aH
8
O
L
U 1
4
H
X
w
n
f 12
ia
8
--``,`,-`-`,,`,,`,`,,`---
X
w 7
n
Z
.m
H
8 4.5
ln
I
X
3.5
FIBER CONTENT - kg/m3
Fig. &Effects of fiber type and amount on I values for 300 x 100 x 100 mm beams with steel
and polypropylene fibers (left) with an indication of the importance of specimen size for rigid
steel fibers
Provided by IHS under license (right)
with ACI
A C 1 SP-155 95 = 0662747 0523113 480
22 Johnston
100 I I I I I I I
FIBER FIBER FIBER FIBER
CC63-0. CC63-• 0 EE 18- AA
O 9 0
FP38-AA
FP64-vV -
d
ìE
a 80
O
O
I
Q 70
a
IY
O
5. 60
9
F
o
2
!i
50
40
OI :& SPECIMEN
30 - SPAN / CROSSECTION
0 A- 300/100mm
o A- 4 5 0 / t 5 0 m m
20
0
g-90
8
0 80
ui
0:
E
--``,`,-`-`,,`,,`,`,,`---
70
o
8 60
I
I-
(3
Ea 50
I-
V>
40
30
Fig. 9-Effects of fiber type and amount on R values for 300 x 100 x 100 mm beams with steel
and polypropylene fibers (left) with an indication of the importance of specimen size for rigid
steel
Provided by IHS under fibers
license (right)
with ACI
A C 1 SP-1.55 95 m 0662749 0523114 317 m
SP 155-2
Toughness of Fiber-Reinforced
High-Strength Concrete from
Notched Beam Tests
by D. Jamet, R. Getîu, V. S. Gopalaratnam, and A. Aguado
Svnomis: The toughness of fiber-reinforced concretes (FRC) is characterized

from notched beam tests. The tests are performed under CMOD control in a
servo-hydraulic machine to obtain the stable response of both the unreinforced
concrete and the FRC. Several toughness measures are defined in terms of the
experimentally obtained load versus crack opening (CMOD) curves. They give a
better indication of the fundamental behavior of the concrete, avoid the problems
associated with the approach based on the deflection of unnotched beams, and are
amenable to the incorporation of serviceability considerations (e.g., crack widths).
The effect of specimen size on the toughness is found to be significant in both the
matrix- and fiber-dominated regimes of the FRC behavior. In general, toughness
increases with specimen size and needs to be accounted for in the
characterization.
The study was conducted on beams of a 70 MPa compressive strength

silica fume concrete, with and without high-strength hooked steel fibers. It was
found that the incorporation of a low volume fraction (1%) of steel fibers is
--``,`,-`-`,,`,,`,`,,`---
sufficient to significantly decrease the brittleness of high strength concretes.
Kevwords: Beams (supports); cracking (fracturing); ductility; fiber

reinforced concretes; fibers; high-strenpth concretes; silica fume; steels;
tests; toughness

No reproduction or networking permitted without license from IHS 23
Not for Resale
A C 1 SP-155 95 = 0662949 0523115 253
24 Jamet et al
David Jamet is an Assistant Lecturer at the School of Civil Construction

(University of Valparaiso, Chile). The experimental work reported in this paper
was conducted during his stay at the Universitat Politècnica de Catalunya (UPC),
--``,`,-`-`,,`,,`,`,,`---
Barcelona, Spain, funded by the Instituto de Cooperación Ibero-Americano.
AC1 member Ravindra Géttu is Director of the Structural Technology Laboratory,

UPC, Barcelona. He is a member of AC1 Committee 446, Fracture Mechanics.
The current interests of Dr. Gettu include experimental techniques, high strength
concrete, fracture mechanics, toughness of FRC and concrete technology.
AC1 member Veiiore S. Gopaiambamis Associate Professor of Civil Engineering

at the University of Missouri-Columbia. He currently chairs AC1 Committee 446,
Fracture Mechanics. He is the secretary of AC1 Committee 544, Fiber Reinforced
Concrete, and chairs its Subcommittee on Testing. His current research interest
includes fracture mechanics of cement and concrete composites.
AC1 member Antonio Aguado is Chair Professor of Structural Concrete in the

Department of Construction Engineering, UPC, Barcelona. Currently he is the
Sub-director for Research at the School of Civil Engineering, Barcelona. The
research interests of Dr. Aguado include concrete technology, high strength
concrete, polymer concretes, fatigue, concrete dams and structural design.
INTRODUCTION
Toughness characterization is essential for the utilization of fiber reinforced

concretes (FRC) since it provides the basis for quantifying the benefits of
incorporatingfibers, such as the post-crack load-carrying capacity and the ductility
of the composite. It can also be used for relating the fundamental material
behavior to structural performance. Conventionally, FRC toughness is obtained
experimentally from a beam test. Such simple "engineering" measures can be
correctly employed only when they are correlated to basic parameters, such as
fracture toughness and fiber pullout characteristics.Moreover, effects of specimen
size and geometry have to be accounted for when extrapolating laboratory results
to larger structures. These effects depend on volumetric and planar energy
dissipation mechanisms, whose proportions vary with the structural dimensions
and crack configuration (1).
The present work studies the characterization of toughness based on notched

beam tests that are conducted under crack-mouth opening displacement (CMOD)
control in a closed-loop servo-hydraulictesting machine. This approach deals with
the planar energy dissipation at a predefined critical cross-section and directly
A C 1 SP-II55 95 W 0hh2947 0 5 2 3 L L h L 9 T
Testing of FRC 25
uses the displacements associated with this plane. The size effects on the matrix-
and fiber-dominated regimes of the specimen response are studied by testing three
different sizes of geometrically-similar beams.
RESEARCH SIGNIFICANCE
The toughness characterization of fiber reinforced concretes has several

objectives. For materials engineering purposes, toughness can be used to quantify
the effectiveness of fibers, the ductility of the composite and the resistance offered
by the material against crack propagation. In structural design, the toughness
concept has been employed in the design against brittle failure and in the
computation of load-carrying capacities at large deformations, and seems
promising. The significance of toughness is especially important for high-strength
concretes (HSC), which characteristically exhibit higher brittleness than normal
concretes. One of the most rational methods of reducing the brittleness
significantly is through the incorporation of steel fibers. In this paper, the
toughness of a 70 MPa-strength silica fume concrete, reinforced with hooked steel
fibers, is evaluated using the notched beam test. The aim is to use toughness as
a measure of the effectiveness of fibers in increasing the ductility and energy
absorption capacity of the concrete.
TOUGHNESS FROM NOTCHED BEAM TESTS
In the context of fiber reinforced concretes, toughness may be generally

defined as an experimentally obtained measure of the energy absorption capacity
of the composite. The most common test configuration used for this purpose is
the bending of prismatic specimens. A recent review (1) of the popular methods
for the determination of flexural toughness lists their merits, as well as problems
associated with some of them (e.g., References 2-4). One approach that avoids
many of those problems and seems to have a sound fundamental basis is the
quantification of toughness through tests of notched specimens. Several such
geometries and loading configurations have been proposed for concrete (5-9),
including the single-edge notched plate under tension, double-notched prism under
eccentric compression and the single-edge notched beam under three-point
bending.
--``,`,-`-`,,`,,`,`,,`---
In the present study, the three-point bend (3PB) specimen shown in Figure 1
has been adopted as the test geometry. Nevertheless, most of the discussion and
its implications also apply to other notched specimen geometries. The 3PB
specimen has several advantages: (i) it is relatively easy to fabricate and test;
A C 1 SP-155 95 W 0bb2949 0523117 0 2 6 W
26 Jamet et al
(ii) it deals with just one critical section and one propagating macro-crack, making
its analysis and modeling simpler; (iii) the notch localizes the crack along its
plane, reducing the variability of the crack path and non-symmetry of the
deformations,both of which can occur in unnotched third-point loaded beam tests;
(iv) a stable test can be performed, even for plain concrete, under crack-opening
displacement control and (v) it has been extensively studied and documented in
the fracture mechanics literature (see Reference (IO) for a review).
Since the crack-mouth opening displacement (CMOD) of the notched beam

can be measured accurately, prescribing it as the controlled variable in a closed-
loop testing machine leads to stable post-peak response. Moreover, toughness
based directly on CMOD, instead of beam deflection, avoidsthe errors related to
extraneous deformations and unstable control beyond the peak load (8, 1 1 , 12).
in the present work, several toughness measures that have been previously defined
on the basis of the deflection of unnotched beams are adapted for the CMOD of
notched specimens. As shown later, the beam deflection and CMOD are related
almost linearly in the post-peak regime, making it possible to convert deflection-
based measures to analogous CMOD-based ones, and vice versa. Furthermore, the
area under the experimentally obtained load-CMOD curve also reflects the energy
absorbed during failure.
MATERIALS USED
The base (unreinforced) concrete had a composition of cement:sand:gravel:

microsi1ica:water as 1:1.32:2.2:0.1:0.42, by weight. ASTM Type I (Spanish
Type I 45-A, rapid hardening) cement, crushed limestone aggregates and
silica fume slurry were used. DRAMIX@ZC 30/.50 collated steel fibers with
hooked ends (tensile strength of about 2000 MPa, elongation of 0.8-1.0%, 30 mm
in length and 0.5 mm in diameter) were added to this concrete. The base concrete
is hereafter denoted as HSC-0.0, and the concretes with 40 kg/m3 (volume
fraction, V f , of approximately 0.5%) and 80 kg/m3 (Vf= 1%) of fibers as
HSC-0.5 and HSC-1.O, respectively. All the concretes were prepared in a 250-liter
paddle mixer. Other characteristics of the concretes are shown in Table 1. Note
that different amounts of super-plasticizer were used in the three concretes since
the addition of fibers decreased the workability considerably. Also, the slump was
measured with a standard 300 mm cone and the air content with a mini air-meter.
The compressive strength was obtained using standard 150x300 mm cylinders
--``,`,-`-`,,`,,`,`,,`---
tested under stable conditions using circumferential deformation as the feedback

in a closed-loop servo-hydraulic system. This strength was observed to decrease
slightly due to the addition of fibers, which may be attributed to the increase in
air content.

A C 1 SP-1-55 9 5 IOb62949 0 5 2 3 3 3 8 Tb2 =
Testing of FRC 27
DETAILS OF BEAM "3
Beams of three different sizes, three specimens in each size, were cast from
each of the concretes. The specimens were geometrically similar (fig. l), with
thickness = 90 mm and depth (d)= 90, 180 and 320 mm. The beams were cast
in two layers normal to the loading plane and vibrated with a finishing screed.
Notches of length a, = 0.275d were cut with a diamond disc before testing. The
beams were tested at the age of 25-28 days in a 100 T closed-loop
INSTRON 8505 servo-hydraulic system. The tests were conducted at constant
CMOD rates that were chosen to give peak loads in the HSC-0.0 specimens at
approximately 3 minutes. The CMOD and net mid-span deflection (fig. 2) were
monitored with an INSTRON clip-on crack-opening displacement gage and a
Schaevitz LVDT, respectively. These measurements, as well as the load and
piston displacement, were recorded electronically using the INSTRON FLAPS
computer program. The data sampling was done at 0.6 seconddpoint initially, and
after substantial crack propagation was observed, the acquisition was slowed to
30 secondsìpoint.
TEST RESULTS AND SI2X EFFECT
Typical load-CMOD curves for the different concretes are shown in

Figures 3-5. It can be seen that the tendency toward a hardening-type curve
increases with fiber volume fraction and specimen size. The phenomenon was also
observed previously for concrete with shorter straight fibers (9). This fiber-
dominated size effect seems to be stronger in less ductile composites, where
smaller specimens exhibit softening after the first-crack or first-peak, while much
larger (geometrically similar) specimens undergo considerable hardening. This is
contrary to the accepted trend for unreinforced concrete where smaller specimens
exhibit more ductile post-peak responses (10). It obviously leads to a pronounced
effect of the specimen size and geomets, on toughness, and has to be accounted
for adequately. A quantitative analysis of other size effects can be made from
Table 2, where the maximum nominal stresses and relative displacements
conespondmg to first-crack, first-peak (or bend-over-point) and maximum load
are given. Note that the nominal stress is the maximum stress in the ligament (i.e,,
net section in the notch plane) calculated using elastic beam theory. The relative
displacements shown in Table 2 are the deflections and CMOD divided by the
--``,`,-`-`,,`,,`,`,,`---
beam depth. The first-crack load corresponds to the point at which the load-
CMOD curve exhibited a significant change in slope or a kink. The identification
was made graphically and is admittedly subjective. The first-peak or bend-over-
point is more easily determined, at least when non-hardening behavior is
observed, and is taken to be the end of the matrix-dominated response. The

maximum load is obviously the most objective but some tests had to be
terminated before it was reached due to practical reasons (i.e., large displacements
and long test durations).
From Table 2, it can be observed that the scatter of the stresses is much less
than those of the displacements. Also, the variations in the stresses generally
increase with Vf . The first-peak stress exhibits much less scatter than the first-
crack stress in ail the concretes. Strong size effects are observed in both the
matrix- and fiber-dominated responses for HSC-0.5. The stresses and relative
displacements decrease with an increase in specimen size, the latter much more
significantly. This is similar to that seen in HSC-0.0 as expected ( 1 O). However,
this effect is observed only at the maximum loads of HSC-1.0 and not in the
matrix-dominated response, implying that higher fiber volume-fractions may make
the size effect much milder.
Another interesting aspect seen in Table 2 and Figure 6 is the relation between
CMOD and deflection. The rate of change of deflection during the test decreases
progressively (see inset of Figure 6). especially after first-cracking, to ultimately
reach a constant value. This demonstrates why the test can always be controlled
--``,`,-`-`,,`,,`,`,,`---
in a stable manner with CMOD control. Moreover, in the post-crack regime, the
CMOD-deflection curve is linear with a deflection/CMOD ratio of approximately
0.65, for most of the specimens tested. This behavior may be attributed to rigid-
body rotation of the beam halves, indicating the presence of a predominantly
planar energy dissipation mechanism due to localization of the deformations.
Another implication is that CMOD will be a more sensitive parameter to base
toughness on since it increases much more than deflection, for a given change in
load.
CMOD-BASED TOUGHNESS MEASURES
Several toughness quantifiers based on CMOD, more specifically the area

under the load-CMOD curve, are developed and presented in this section. Most
of them are analogous to the toughness measures defined by existing standards in
terms of the load-deflection curves of unnotched beams.
Toughness Based on Absolute Load-CMOD Ama
JCI (2) and RILEM (13) recommendations define toughness as the work done
@e., area under the load-deflection curve) to reach a prescribed deflection. A
similar measure of toughness, r', , is defined here as the area under the
load-CMOD curve. The CMOD limit is prescribed as a fraction of the beam depth
A C 1 SP-355 95 I0 6 6 2 9 4 9 0 5 2 3 3 2 0 b L 0 I
Testing of FRC 29
dln, where values of 250, 500, 1000 and 2000 have been chosen for n. The
corresponding values of TR are tabulated for the three concretes in Table 3.
Obviously, r", increases with V, , for all n and for all d. This type of toughness
measure can only be used for comparing the values obtained with a specific
specimen size and geometry.
For better comparison, r", cau be divided by the ligament area to get a more
fundamental quantity that is analogous to fracture energy. This, denoted as Tn ,
is the area under the load-CMOD curve, until the prescribed CMOD limit dln,
divided by the ligament area. These values are given in Table 3 for different d
and n. Also evident from the table is that the value of increases with the
specimen size. This effect increases with V f , especially during post-crack
hardening where the larger specimens exhibit more significant increases in load-
carrying capacity than the smaller specimens. Also, TE increases considerably
with CMOD limit (Le., n), which should, therefore, be based on practical crack
--``,`,-`-`,,`,,`,`,,`---
width considerations.The increase in Tz with CMOD limit is greater for a larger

V, indicating the greater effectiveness of the fibers at larger deformations.
Towhness Based Onlv on the Fiber Conûibution
Figures 7 and 8 compare typical load-CMOD curves of the two FRCs with
those of HSC-0.0. The differencebetween the FRC and unreinforced concrete can
be attributed to the presence of the fibers, and can be used to quantify the fiber
performance. Such toughness measures, which explicitly deduct the toughness of
the unreinforced matrix, have been recommended by AC1 (3) and some German
standards (e.g., Reference 14). Following this approach, a relative toughness can
be computed as Tr = TE (for FRC) - r f f (for HSC-O.O), for the values of n used
earlier. These values presented in Table 4 show clearly the increase in toughening
with fiber effectiveness and the specimen size effects. This type of toughness
measure obviously has several advantages but would always require the testing
of a companion specimen of the base concrete. This could be considered as a
disadvantage.It can, however, be handled by defining an idealized curve for the
unremforced concrete based empirically on other tests (as in Reference 14) or by
determining it analytically using an approach such as nonlinear fracture
mechanics.
Nondimensional T o w h e s s Indices Based on Relative Disdacements
The ASTM Standard C 1018 (4) defines toughness indices that are equal to
the areas under the load-deflection curve of an unnotched beam until prescribed
multiples of the first-crack deflection divided by the area until first-crack. A
A C 1 SP-155 95 M 0662949 052312L 557 M
30 Jamet et al
similar approach was used by Barr and Hasso ( 5 ) for defining toughness based on
notched and unnotched specimens.
Bryars et al. (9) recently used a toughness index based on the load-CMOD
area of a notched beam defined in terms of the first-peak, instead of the first-
crack. They found that the value of such an index increases with the CMOD limit,
is practically size-independent and reflects the effectiveness of the fibers
satisfactorily. Two similar indices are evaluated here: Ti" defined as the ratio of
the area of the load-CMOD curve until rn times the first-crack CMOD and the
area until the first-crack; and ';"7 is defined similarly in terms of the first-peak,
instead of the first-crack. The results for rn = 5, 10 and 20 are given in Table 5.
From Table 5 it can be seen that all the indices reflect the fiber effectiveness
and are practically size-independent. Further testing is required to confirm whether
these indices are uifluenced by specimen geometry and dimensions. However, as
concluded by other researchers (1, S), the indices are not very sensitive to the
reinforcement parameters at small values of rn. This implies that tests have to be
continued until larger displacements are attained. Moreover, comparisons with
previous results of Bryars et al. (9) suggest that the sensitivity or the dependence
on the factor rn may depend on the material. In their study, conducted on a HSC
with shorter straight fibers, use of rn = 5 yielded 7'; values that were reasonably
sensitive to the increase in fiber length (i.e., fiber effectiveness).
--``,`,-`-`,,`,,`,`,,`---
CONCLUSIONS
1 . The load-CMOD cume obtained experimentally from notched beam tests

provides a good basis for the toughness characterization of fiber reinforced
concretes. This approach avoids the problems associated with the
characterization based on the deflection of unnotched beams and is amenable
to the incorporation of serviceability limits (such as crack widths). Since the
toughness based on CMOD is derived from the deformation of one critical
cross-section and a single crack, it is more closely related to the fundamental
behavior of the composite.
2. Several toughness measures have been analyzed in this study. They have been
based on (i) the absolute area under the load-CMOD curve, (ii) area between
the curves of unreinforced and fiber reinforced concretes and (iii)
nondunasional indices defined as ratios of the post- and pre-cracking areas.
All such quantities reflect the effectiveness of the fibers adequately and
warrant further research.

A C 1 SP-155 95 œ 0 6 6 2 9 4 9 0523322 493 œ
Testing of FRC 31
3. There is a significant effect of the size of the specimen on its behavior, in

both the matrix- and fiber-dominated regimes. This should be adequately
considered in toughness characterization.
4. The brittleness of high-strength silica fume concrete can be decreased

considerably with the addition of a low volume-fraction (1% or less) of
randomly distributed steel fibers.
Acknowledgements
This research was partially supported by Spanish DGICYT grants PB90-O598

and MAT93-0293 to the Universitat Politècnica de Catalunya, Barcelona. The
authors are grateful to D.Nemegeer of N.V.Bekaert (Belgium), J.Puig of Cementos
M o h s (Spain) and J.L.Rodríguez of GRACE (Spain) for providing the materials
used in this study. V.S. Gopdaratnam was supported, during his sabbatical stay
as Visiting Professor at the UPC, by the DGICYT (grant SAB93-0190) and the
University of Missouri-Columbia. The help of S.Carmona, who performed the
compression tests, is deeply appreciated.
REFERENCES
1. Gopalaratnam, V.S., and Gettu, R., "On the Characterization of Flexural

Toughness in FRC", Proc., US NSF - Univevsity of Sheffield Workshop on
Fibre Reinforced Cement and Concrete (Sheffield, U.K.), eds. R.N.Swamy
and V.Ramakrishnan, pp. 161-180, 1994.
2. JCI Standard SF-4, "Method of Tests for Flexural Strength and Flexural
Toughness of Fiber Reinforced Concrete", J q a n Concrete Institute Standards
f o r Test Methods of Fiber Reinforced Concrete, Tokyo, pp. 45-51, 1984.
3. AC1 Committee 544, "Measurements of Properties of Fiber Remforced

Concrete", ACZMater. J., V.85, No.6, pp. 583-593, 1988.
4. ASTM C 1018-92, "Standard Test Method for Flexural Toughness and First-
Crack Strength of Fiber-Reinforced Concrete Using Beam With Third-Point
Loading", Annual Book of Standards, ASTM, Philadelphia, V. 04.02,
pp. 510-516, 1992.
5 . Barr, B.I.G., and Hasso, E.B.D.,"AStudy of Toughness Indices", Mag. Concr.

--``,`,-`-`,,`,,`,`,,`---
Res., V.37, No.132, pp. 162-173, 1985.

A C 1 SP-355 95 M 0662749 0523323 32T M
32 Jamet et al
6. RILEM Committee 5O-FMC, "Determination of Fracture Energy of Mortar

and Concrete by Means of Three-Point Bend Tests on Notched Beams", Draft
Recommendation, Mater. Struct., V.18, No.106, pp. 285-290, 1985.
7. RILEM Committee 89-FMT, "Determination of Fracture Parameters (Kf, and

CTOD,) of Plain Concrete Using Three-Point Bend Tests", Draft
Recommendation, Mater. Struct., V.23, No.138, pp. 457-460, 1990.
8. Gopalaratnam, V.S., Shah, S.P., Batson, G.B., Criswell, M.E.,

Ramakrishnan, V., and Wecharatma, M., "Fracture Toughness of Fiber
Reinforced Concrete", ACI Mater. J , V.88, No.4, pp. 339-353, 1991.
9. Bryars, L., Gettu, R., Barr, B., and Ariño, A., "Size Effect on the Fracture of
Fiber-Reinforced High-Strength Concrete",Proc., Europe-US. Workshop on
--``,`,-`-`,,`,,`,`,,`---
Fracture and Damage in Quasibrittle Structures (Prague), E&FN Spon,
London, 1994.
10. Gettu, R., and Shah, S.P.,"Fracture Mechanics", in High Perfomance

Concrete and Applications, eds. S.P.Shah and S.H.Ahmad, Edward Arnold
Publishers, London, pp. 161-212, 1994.
1 1 . Trottier, J.-F., and Banthia, N., "Toughness Characterization of Steel-Fiber

Reinforced Concrete", J. Civil Engng. Mater., V.6, No.2, pp. 264-289, 1994.
12. Mindess, S., Chen, L., and Morgan, D.R., "Determination of the First-Crack
Strength and Flexural Toughness of Steel Fiber-Reinforced Concrete", Advn.
Cem. Bas. Mat., V.l, No.5, pp. 201-208, 1994.
13. RILEM Committee 49-TFR, "Testing Method for Fibre Reinforced Cement-
Based Composites", Mater. Struct., No. 102, pp. 441-456, 1984.
14. "Technologie des Stahlfaser-betonsund Stahlfaserspritzbetons"(Technology

of Steel Fiber Reinforced Concrete and Steel Fiber Shortcrete, in German),
Deutschen Beton-Vereins, pp. 3-18, 1992.
Unit Conversions
1 111111 = 0.0394 inches

1 m = 3.28 ft.
1 MPa = 145 psi
1 kg = 2.2 lbs
1 N = 0.224 Ibs-force
1 liter = 0.0353 CU. ft.
--``,`,-`-`,,`,,`,`,,`---
TABLE 1 - CONCRETE PROPERTIES
*1:2 mixture, by weight, of solid super-plasticizer and water.
TABLE 2 - MEAN VALUES AND COEFFICIENTS OF VARIATION OF STRESS, CMOD AND

DISPLACEMENT AT FIRST-CRACK, FIRST-PEAK, AND MAXIMUM 'LOAD
Stress Relative Relative Stress Rel. I Rel. defl. Stress Rel. Re1 defl.
CMOD deflection CMOD CMOD
(MPd (W-) (P&-) (ma) (P&-) (rd-) (MPa) (Y&-) (cid-)
HSC-0.0 90 4.53 0.156 0.156 5.28 0.252 0.237 5.28 0.252 0.237
f4.7K iL2.4% 112.4% *0.7% i6.7'h ì2.7% ì0.6% *6.7% +2.7%
I I 412
i6.6% 1
~ 0.120
19.6%
I 0.136
*20.2%
I 440
ì2.5%
1 0.167
ì20.0%
1 0.156 I 4.40
*2.5%
I O I67
*20.0%
1 0.156 I
i13.4% ì13.4% 513.8% S.4% +15.2%1 *9.0% *3.9% *62% *2.9%
320 3.75 0.138 0.117 6.09 0.675 ~ 0.514 6.94 2.81 I91
f22.2% +31.5% ì29.6% ì26.3% *%.i% 1I *51.2% ì24.7% *l4.0% *l4.8%
Relative CMOD = CMODíBeam depth

Relative deflection = Deflection/Beam depth
Stress = maximum nominal stress computed for net section
Nores; Coefficients of vaiation are. omitted when less than three data values were used.
Maximum load data are omitted when there was no clear maximum in the test range
A C 1 SP-155 95 O b b 2 î Y 9 0523325 L T 2 M
34 Jamet et al
TABLE 3 - AVERAGE TOUGHNESS VALUES FOR DIFFERENT CONCRETES BASED ON LOAD-

CMOD AREA
--``,`,-`-`,,`,,`,`,,`---
Note that values are omitted when the experimental CMOD range is smaller than the limit
TABLE 4 - FIBER CONTRIBUTION TO TOUGHNESS
Material T r (N/mm)
n = 500
HSC-0.5 0.025 0.089
0.085 0.253
320 1 0.042 1 0.163
HSC-1.0 90 I 0.007 I 0.042 0.149
180 I 0.028 I 0.116 0.364
320 I 0.057 I 0.220 ---
e = E (of FRC) - e (of base concrete)

A C 1 SP-355 75 IObb2747 0 5 2 3 3 2 6 037
Testing of FRC 35
TABLE 5 - MEAN VALUES OF NONDIMENSIONAL TOUGHNESS INDICES

Material d Index based on Index based on
--``,`,-`-`,,`,,`,`,,`---
T I c = (load-CMOD area until the CMOD value of mS, )/(area until 6, )

?"i = (load-CMOD area until the CMOD value of mSp )/(area until Sp )
6, = CMOD at first-crack, 6, = CMOD at first-peak
Values marked with * were obtained with less than three specimen data
b
S=2.5d
Fig. 1-Three-point bend geometry

A C 1 SP-355 75 = Obb27Y7 0523327 T75
36 Jamet et al
Rigid steei yoke

--``,`,-`-`,,`,,`,`,,`---
Fig. 2-Schematic of specimen and yoke used to measure deflection
2oooo
High strength concrete, HSC-0.0
15OOO
E
n
loo00
75
m
O
4
Large
d = 320 rnm
5000
Medium
d=l80mrn
Small
d=90mm
O
O 100 200 300 400 500
CMOD, A (microns)
Fig. &Typical load-CMOD curves for unreinforced concrete

A C 1 SP-L55 95 = 0662949 0523328 9 0 1 =
Testing of FRC 37
25000
High strength fiber concrete, HSC-0.5
Hookedend steel fibers (30 mm, V, = 0.5%)
2oooo
t
n
U- Medium ,
8
J
loo00 -
-----.- d=180mm I
- High strength fiber concrete, HSC-1 .O

36000 Hooked-end steel fibers (30 mrn, V,= 1%)
32000
28000
-k 2 m
Q
2oooo
U-
m
2 16Ooo
12000
8000
- d-9ümm
4Mxi
O-
O 400 800 1200 16W
CMOD, A (microns)
--``,`,-`-`,,`,,`,`,,`---
Fig. &Typical load-CMOD curves for concrete with fiber V, = 1.0 percent
A C 1 SP-155 95 I0 6 6 2 9 4 9 0523129 8 4 8
38 Jamet et al
1000 I
n
800 /
cn
c
g
.-
E 600
v
Lo
t
.=
O
o
400
a>
=Q)
n
200
o 50 100 150 200
O
O 400 800 1200 1600 2000
CMOD, A (microns)
Fig. &Typical deflection-CMOD curves for all concretes and all specimen sizes
25000
HSC-0.5: High strength fiber concrete (V, = 0.5%)

Large: d = 320 mm
Small: d = 90 mm
20000
HSC-0.5-Large
HSC-0.5-Small
O 100 200 300 400 500 600

--``,`,-`-`,,`,,`,`,,`--- CMOD, A (microns)
Fig. 7-Comparison of iypical load-CMOD curves of the small and large specimens of
unreinforced concrete and FRC with V, =Not 0.5
percent
for Resale
A C 1 SP-155 9 5 I0 6 6 2 9 4 9 0523330 5 b T
Testing of FRC 39
Small: d = 90 mm
t
n
O
4
3
HSC-1 .O-Small
O 100 200 300 400 500 600

CMOD, A (microns)
Fig. M o m p a r i s o n of typical load-CMOD curves of the small and large

specimens of unreinforced concrete and FRC with Vf = 1.0 percent
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0662’749 0523131i 4Tb I
SP 155-3
Comparative Toughness Testing of

Fiber Reinforced Concrete
by L. Chen, S. Mindess, D. R. Morgan, S. P. Shah,
C. D. Johnston, and M. Pigeon
SvnoDsis: Round-robin tests of the flexural toughness of fibre reinforced

concrete were camed out using six different testing machines in five different
laboratories. Six groups of beams, including a plain concrete control, two
different volumes of polypropylene fibres, and three different volumes of steel
fibres were tested in accordance with ASTM C1018, with special care taken to
exclude the “extraneous” deflections due to deformations at the specimen
supports. The results from each laboratory were used to compute the ASTM
Cl018 toughness indices Ili,Il0, Izo,130 and IsO, and the corresponding residual
strength factors RS,iO, RIO,^^, R20.30 and R30,50. In addition, the JSCE
Toughness and Toughness Factor were also computed. It was found that,
although the load vs deflection curves were inherently quite variable, in most
cases there was no significant difference amongst the participating laboratories,
except for those mixes with a very low toughness.
It was found that the ASTM C1018 toughness indices, particularly Ili
and Il0, did not discriminate very well amongst the different fibre contents or
different fibre types; the JSCE parameters were rather more successkl in this
regard.
--``,`,-`-`,,`,,`,`,,`---
Keywords: Fiber reinforced concretes; fibers; flexural strength;

polypropylene fibers; steels; strength; tests; toughness

Not for Resale
A C 1 SP-355 9 5 I0 6 6 2 9 4 9 0 5 2 3 3 3 2 3 3 2
42 Chen et al
BIOGRAPHIES
Mr. Lianrong Chen is a Ph.D. candidate in the Department of Civil Engineering

at The University of British Columbia. He obtained his BS and MS degrees in
Materials Science and Engineering from Tongji University and Wuhan
University of Technology, respectively, in China. His research interests include
mechanical properties and durability of concrete and fibre reinforced concrete.
AC1 Fellow, Sidney Mindess, is a professor in the Department of C i d

Engineering at The University of British Columbia. He is a member of AC1
Committee 370, Short-Duration Dynamic and Vibratory Load; and AC1
Committee 446, Fracture Mechanics. He is also a member of the Co-
ordinating Committee of RILEM.
AC1 Fellow, Dudley R. Morgan, is chief materials engineer for the AGRA
Earth & Environmental Group in North America. He is a member of AC1
Committee 544, Fibre Reinforced Concrete, AC1 Committee 506, Shotcrete
and AC1 Committee 234, Silica Fume, and has been extensively involved in
specifjing and testing toughness for fibre reinforced concrete and shotcrete
projects.
AC1 Fellow, Surendra P. Shah, is a Walter P. Murphy professor of Civil

Engineering and the Director of the National Science Foundation's Science and
Technology Centre for Advanced Cement-Based Materials. He is also the
editor-in-chief of the recently established Advanced Cement Based Materials
Journal, published by Elsevier in affiliation with ACI. He is post chairman of
--``,`,-`-`,,`,,`,`,,`---
AC1 Committee 544, Fibre Reinforced Concrete; and AC1 Committee 215,
Fatigue of Concrete; and is a current member of several AC1 Committees. He
received the AC1 Anderson Award in 1989.
AC1 Fellow, Colin D. Johnston, is a professor of Civil Engineering at the

University of Calgary. He was awarded the AC1 Wason Medal for materials
research in 1976, and received a Government of Alberta Achievement Award
for excellence in concrete engineering the same year. He was chairman of
ASTM Subcommittee C09.03.04 on Fibre Reinforced Concrete from 1980 to
1990, and is currently a member of AC1 Committee 544, Fibre Reinforced
Concrete.
AC1 Fellow, Michel Pigeon, is a professor of Civil Engineering at Laval

University. He is a member of AC1 Committee 201, Durability of Concrete;
and of R E E M Committee TC115 on High-Strength Concrete and TC117 on
Frost Resistance. He was president of the Quebec and Eastern Ontario
Chapter of AC1 in 1986-1987.

A C 1 SP-I155 9 5 II 0662949 0 5 2 3 1 3 3 2 7 9 I
Testing of FRC 43
INTRODUCTION
The primary reasons for adding discontinuous fibres to a plain concrete

matrix are to improve the post-cracking response of the concrete, i.e., to
improve its energy absorption capacity and apparent ductility, and to provide
crack resistance and crack control. For fibre reinforced concrete (FRC), the
concept of flexural toughness (which is a measure of the energy absorption) is
often used to characterize its performance. However, it has long been known
that the flexural toughness of FRC is sensitive to the way in which it is
measured (1-4). In addition, there are a number of uncertainties regarding the
way in which the flexural toughness should be determined, interpreted or used.
Probably the most common method of trying to quanti@ the flexural

toughness of FRC is that prescribed in ASTM C1018 (5). In this test, the
energy absorbed up to certain specified deflections is normalized by the energy
absorbed up to the point of first cracking; the resulting toughness indices and
residual strenHh factors are defined in Fig. 1. However, it has been shown
(1,4) that, for this test to be reproducible between laboratories, great care must
be taken to eliminate any “extraneous” deformations caused by deflections at
the specimen supports or machine deformations. Indeed, in the round-robin
test series reported in (1), it was found necessary to divide the tests into two
groups: those which excluded the extraneous deformations, and those which
included them.
In addition, it has recently been shown that the ASTM C1018

toughness indices are not independent of specimen geometry (3), unless the
specimens are in geometrical similarity. The specimen s i e was found to
idluence not only the toughness, but also the stress and deflection at first
crack, and the ultimate flexural strength. The toughness parameters decreased
with an increase in the span-to-depth ratio. They also decreased with a
decrease in the width of the specimen. The ASTM toughness indices also have
been shown to depend upon the way in which the point of “first crack” is
defined (4); when the load vs. deflection curves are examined carefully, it
becomes clear that there is no unique definition of first-crack. First crack may
be defined as the point at which the load reaches its first maximum point; it
may be defined (as per ASTM C101S) as the point at which the load-deflection
curve first becomes obviously non-linear; or it may be defined (4) arbitrarily as
the point from which a series of 20 consecutive data points (over a total
deformation of 0.01 mm or more) have a slope at least 5% less than the
average slope of the load vs. deflection curve between 45% and 70 % of the
peak load. These different definitions can lead to differences in computed
ASTM Cl018 toughness indices ranging from 10%at low fibre contents to
about 35% at higher fibre contents. Thus, there are concerns about the
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-I155 9 5 I0662949 0 5 2 3 3 3 4 L O 5
44 Chen et al
applicability of this test method for speciQing FñC,or for carrying out quality
control in the field.
A second commonly used method of evaluating FRC is that proposed

in JSCE-SF4 (6). This test method uses the area under the load-deflection
curve out to some specified deflection, i.e., the point at which the deflection
reaches U150 of the beam span (Fig. 2); a toughness and a toughness factor
are defined. Unlike ASTM C1018, which is supposed to be independent of
specimen dimensions, JSCE-SF4 clearly is dependent on the specimen
geometry. This method has the considerable advantage that the precise shape
of the load-deflection curve in the region surrounding the point of first crack is
not very important in defining the toughness. (This should not be taken as an
invitation to include extraneous deflections; these are specifically excluded in
--``,`,-`-`,,`,,`,`,,`---
the test specifications.) However, this method is less sensitive than that
prescribed in ASTM Cl018 as to whether the extraneous deformations have
been properly eliminated (and completely insensitive as to whether the first
crack deflection has been properly chosen). For instance, Fig. 3 shows the
load vs. deflection curves for three different FRC mixes, both including and
excluding extraneous deformations. These three sets of curves show quite
different behaviours. The calculated toughness parameters for these curves are
shown in Table 1, for the mixes described in Table 2. It may be seen that T,sc,
and FJSCEare essentially independent of how the deflections were measured;
the various ASTM C1018 parameters, on the other hand, can vary
considerably.
It has been suggested (1,3) that JSCE-SF4 is better able than ASTM
Cl018 to distinguish amongst FRC’s containing different fibre types and
different fibre volume fractions. This view, however, remains a controversial
one (2,7).
In the round-robin tests reported here, six different machines in five

different laboratories were used to cany out toughness tests on FRC, fiom
which both the ASTM toughness parameters and the Japanese toughness and
toughness factor could be determined. In all six test arrangements, great care
was taken to eliminate extraneous deflections, so that at least some of the
problems that arose in the previous round-robin tests (1) could be avoided.
EXPERIMENTAL PROCEDURES
SDecimen PreDarrtion
Unnotched specimens of dimensions 100x100~350mm were cast in

perspex moulds and moist cured for about 30 days. Six different concrete

A C 1 SP-155 95 I0662949 0523135 041 I
Testing of FRC 45
mixtures were cast, with 24 specimens prepared from each mixture. In

addition to a control mix without any fibres, three different contents of steel
fibres and two different contents of fibrillated polypropylene fibres were used.
For the FRC mixes, the dosages of water reducing admixtures and
superplasticizers were varied in order to maintain comparable workabilities.
The mix proportions and fresh concrete properties for these mixtures are given
in Table 2.
Test Protocol
After moist curing, four specimens of each type were shipped to each
of the participating laboratories, where they were tested at an age of 42 f2
--``,`,-`-`,,`,,`,`,,`---
days after casting. The tests were camed out in accordance with ASTM
C1018, but taking into account the proposed revisions to this standard
circulated to the members of ASTM Committee C09.03.04 on Fibre
Reinforced Concrete in March, 1993.
o The deflection measuring systems excluded all extraneous deformations.

o Either digital recorders or x-y plotters were used to capture the load-
deflection data.
o Full load-deflection curves were provided up to at least 2 mm beam
deflection (U150 of the span).
o The estimated “first crack” load was shown on each cuwe.
0 For each beam, the following values were calculated: ASTM C1018 IS,110,
120, 130, 150, R5,10> &0,30, and R30,50; JSCE-SF4 TJSCE and
toughness factor FjSCE.
ExRerimental Arranpements
The experimental arrangements in each laboratory are summarized in

Table 3.
Five of the loading systems were operated under stroke displacement

control; the loading system at Northwestern University was operated under
beam displacement control using a closed-loop testing system. Three different
deflection measuring systems were used, so that accurate measurements of the
mid-span deflection under third-point loading, excluding extraneous
deformations could be camed out. The schematics of the Japanese Yoke, Top
LVDT’s and Side LVDT’s are shown in Figs. 4-6.
ExDerimentri Results
Figures 7-12 show d of the load vs deflection curves obtained from

the six FRC mixes described in Table 2, without differentiating them by

A C 1 SP-155 9 5 I0662949 0523l13b T ô ô I
--``,`,-`-`,,`,,`,`,,`---
46 Chenetal
laboratory, except that the results obtained with the closed-loop testing system
are shown by thicker lines. Figures 13-18 show typical load-deflection curves
for each mix as obtained in each laboratory, so that the differences between the
various loading systems can be seen more clearly.
Table 4 gives the values of the various ASTM Cl018 toughness

indices, as well as the first-crack strength (fid)and the deflection at first crack (
öl,), as averages for each laboratory. Table 5 gives the values of the Residual
Strength Factors, also determined according to ASTM Cl018 (for pure elastic-
plastic behaviour, the residual strength factors would all be equal to 100).
Table 6 gives the corresponding values of the JSCE-SF4 toughness
parameters. It should be noted that the “theoretical” first crack deflection, 6,
may be determined from the expression,
flexural shear
component component
where
mid-span deflection
load at first crack
span length
modulus of elasticity (assumed to be 35,000 MPa)
Poisson’s ratio (assumed to be 0.2)
moment of inertia
beam depth
These values are also given in Table 7. It may be seen that these theoretical
first crack deflections agree within k20% with the measured values, except for
four of the thirty-six results (see Tables 4 & 7), which fall outside this range.
Of these four results, two are from Northwestern University, the other two
from the UBC MTS machine; in these cases, the measured first crack
deflections all appear to be low. These differences are small when contrasted
with the values that are obtained when the extraneous deformations are not
properly eliminated (Table i). However, the measured values are much more
variable than the theoretical values, probably because of the difficulties in
accurately measuring the very small deflections at first crack (4), even when
extraneous deformations are eliminated.

A C 1 SP-155 95 I0662949 0 5 2 3 3 3 7 914
Testing of FRC 47
DISCUSSION
The purpose of this round-robin series of tests was two-fold:
1) To determine the variability of flexural toughness measurements

between different laboratones; and
2) To see whether the commonly used flexural toughness parameters

could easily distinguish amongst different fibre types or fibre volumes.
Variabilitv in Touphness Measurements
From Figs. 7-12, it may be seen that there is a considerable spread in

the measured load-deflection curves. Given the experimental difficulties
inherent in such tests, and the variability in cast specimens, this is not
surprising. However, the variability is greatest for the low toughness (¡.e. low
fibre content) curves; it is less for the high toughness curves (those for 0.75%
and 1.27% by volume of steel fibres). When representative load-deflection
curves for each mix are compared for the six different experimental
arrangements (Figs. 13-18), the between-laboratory variability is not great,
particularly since all of the laboratories took great pains to eliminate
extraneous deflections.
Moreover, fi-om Tables 4-6, it may be seen that the calculated

toughness parameters did not vary too much from laboratory to laboratory,
though in general, the University of Calgary results tended to be on the low
side and the Northwestern University results tended to be on the high side
most of the time.
However, fiom Fig. 19, for plain concrete (Mix 1) and for the relatively
low fibre content mixes (Mixes 2, 5 and 6), the load falls very rapidly from the
maximum load to quite a low level immediately after cracking has been
initiated; this represents a region of instability. Because of this, the recorded
lines B-C, even with properly designed testing systems, do not represent the
true load-deflection response of the beams in the region B-C;they are largely
an artifact of the particular system. In fact, it is very difficult to measure the
‘ h e ” load-deflection curve in this region unless a closed-loop servo-
controlled testing system is used, such as the one at Northwestern University.
This may be the reason that the load-deflection curves for these four low
toughness mixes (Figs. 7, 8, 11, 12) exhibit more variability. This is manifested
in Tables 4-6, where the greatest variability amongst the different laboratories
in the ASTM toughness indices 13,, and Ise, and in the Japanese Toughness
TJSCE,occurs for these four mixes. It therefore may not be appropriate to
evaluate low toughness mixes on the basis of their load-deflection curves
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-I155 95 O662949 0523338 850
48 Chen et al
unless this instability is taken into account. This is because, for the low
toughness mixes such as Mix 5 (0.20% polypropylene fibres), the end point for
130for some specimens fell in the unstable part of the curve. Thus 15, Il0, bo,
and R5,i0,Rlqz0and even &o,3o may be affected by the instability.
It should be noted that the differences between the load-deflection

curves from X-Y plotters versus digital data acquisition systems lie mainly in
the shape of their unstable portions (See Fig. 20).
Differences in Fibre Tvue or Fibre Volume
These results are in general agreement with the conclusions reached by

Gopalaratnam et al. (1). From Table 4, the ASTM toughness indices I5 and Ilo
were not particularly sensitive in distinguishing amongst the different fibre
types or contents, though the very high fibre content mix (Mix 4, 1.27% by
--``,`,-`-`,,`,,`,`,,`---
volume of steel fibres) did tend to stand out. Only when using Izo, and more
particularly 130 and 150, did the ASTM toughness indices begin to discriminate
sensibly amongst the different mixes, even though the results presented in
Table 4 are not entirely consistent. From Table 4, it may be seen that, for the
UBC-Instron machine, the Izo, 130, and Is0 values for Mix 5 (0.20%
polypropylene fibres) are considerably higher than those for Mix 6 (0.50%
polypropylene fibres), which can not be correct. The reason for this is that
these values were calculated from the unmodified P-6 curves, which contain
irregularly shaped unstable portions. For such curves, the calculated toughness
indices are simply artifacts of the trace of the dropping pen (the recorded curve
in Fig. 21). The "modified curve in Fig. 21 is an extrapolation of the initial
slope of the line &er first crack, and is a better approximation of the real
behaviour. If the curves are modified (Fig. 21), the values for the two mixes
fall much closer together (values shown in parentheses in Tables 4-6). In
addition, some of this inconsistency may also have been due to differences in
estimating the location of the first crack.
From Table 5, it may be seen that the R-values do discriminate amongst

the different steel fibre mixes, but only was consistently able to
distinguish between the two polypropylenefibre mixes.
From Table 6, it would appear that the JSCE-SF4 toughness

parameters did distinguish reasonably well amongst the different mixes. This
may be due to the fact that, when the total area under the load-deflection curve
is measured, inaccuracies in the portion of the curve in the vicinity of the first
crack are not very important. Moreover, the Japanese toughness T,
appeared to be more sensitive than the ASTM toughness indices Is, Il0,boand
130in distinguishing between the two fibre types employed in these tests. For
instance, Figure 22(a) shows a comparison between the values of the Izo for the
six test arrangements for mix 2 (0.25% steel fibres) and mix 5 (0.20%

A C 1 SP-I155 9 5 I0 6 6 2 9 4 9 0 5 2 3 3 3 9 7 9 7
Testing of FRC 49
polypropylene fibres). On the basis of these data, there is no very clear

distinction between the two mixes. However, when the same comparison is
made on the basis of T,,,, (Fig. 22(b)), the difference in behaviour between the
two mixes is readily apparent.
Table 8 shows the relative sensitivity of d l the toughness parameters

calculated in this study in distinguishing between these same two mixes (Mix 2
and Mix 5). It may be seen that only the parameters obtained at high beam
deflections are particularly sensitive, most notably R30,50,T,,, FISCE, and to a
lesser extent &o,3o and 150. Note that Ise, here refers to a beam deflection of
about 1 mm, i.e. about one-half of the end-point deflection used in calculation
of the JSCE toughness parameters.
CONCLUSIONS
On the basis of these test results, it may be concluded that:
The ASTM toughness indices Is, Il0, and the corresponding Rs,lo, and
to a lesser extent Izo, are not particularly sensitive to either fibre
addition rate or fibre type; 130,Iso, and particularly are more
useful in this regard.
The JSCE toughness parameters TjSCE and FISCEand ASTM R ~ o , ~ o

appear to be better at distinguishing amongst the different mixes and
between different fibre types.
If the ‘kxtraneous” deformations (due to machine deformations and

deformations at the specimen supports) can be excluded during
deflection measurement, both the testing machine and deflection
measuring system have only a minor effect on the calculated toughness
parameters. However, the variations amongst different testing systems
may become significant for low toughness FRC due to the instability in
the load-deflection curve that occurs just after the peak load.
Because it is very difficult to measure accurately the entire load vs

deflection response for FRC with very low fibre addition rates, it may
not be suitable to evaluate these materials on the basis of their load vs
deflection curves if the instability referred to above is not properly dealt
with.
The results presented here illustrate the difficulties in achieving

completely consistent measurements amongst different laboratories,
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-355 95 Obb2949 0 5 2 3 3 4 0 409 W
50 Chenet al
even when every effort was made to test rigorously in accordance with
the testing procedures prescribed in ASTM C1O18.
ACKNOWLEDGEMENTS
This study was supported primarily by the Natural Sciences and

Engineering Research Council of Canada, through its support for the Network
of Centres of Excellence on High-Performance Concrete. The contributions of
the other participating laboratories (University of Calgary, Université Laval,
Northwestern University, and HBT AGRA) are also grateftlly acknowledged.
--``,`,-`-`,,`,,`,`,,`---
REFERENCES
1. Gopalaratnam, V.S., Shah, S.P., Batson, G.B., Crisweli, M.E.,

Ramakrishnan, V. and Wecharatana, M., Fracture Toughness of Fibre
Reinforced Concrete, AC1 Materials Journal, Vol. 88, No. 4, 339-353
(1991).
2. Kasperkiewicz, J. and Skarendahl, A., Toughness Estimation in FRC

Composites, Swedish Cement and Concrete Research Institute, CBI
Report 4:90, Stockholm, 1990,52 pp.
3. Chen, L., Mindess, S . and Morgan, D.R., Specimen Geometry and

Toughness of Steel Fibre Reinforced Concrete, ASCE J o u d of
Materials in Civil Engineering, in press.
4. Mindess, S., Chen, L. and Morgan, D.R., First Crack Strength and
Flexural Toughness of Steel Fibre Reinforced Concrete, Journal of
Advanced Cement Based Materials, Vol. 1, No. 5, pp. 201-208, 1994.
5. ASTM C1018-89, Standard Test Method for Flexural Toughness and

First Crack Strength of ñbre Reinforced Concrete (Using Beam with
Third-Point Loading), 1991 Book of ASTM Standards, Part 04.02,
American Society for Testing and Materials, Philadelphia, pp. 507-513.
6. JSCE-SF4, Method of Test for Flexural Strength and Flexural

Toughness of Steel Fibre Reinforced Concrete, Concrete Library of
JSCE, Japan Society of Civil Engineers, A58-61 (1984).
7. Johnston, C.D. and Authors, Discussion of Ref. 1, AC1 Materials

Journal, Vol. 88, No. 4, pp. 339-353 (1991).

A C 1 SP-155 95 I0662749 0523141 3 4 5
Testing of FRC 51
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0 6 6 2 9 4 9 0 5 2 3 3 4 2 281
52 Chen e t a l
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-355 95 H Obb29Y9 0 5 2 3 3 4 3 138 H
Testing of FRC 53
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 = 0662947 0523344 054 =
54 Chen et al
TABLE 4 - ASTM TOUGHNESS INDICES - AVERAGE VALUES
--``,`,-`-`,,`,,`,`,,`---
*Values in parentheses were recalculated using a modified P-6 curve (see

Fig. 21), to correct for the instability in the original P-6 plots.

A C 1 SP-155 95 IOb62949 0 5 2 3 3 4 5 T70
Testing of FRC 55
TABLE 5 - ASTM RESIDUAL STRENGTH FACTORS - AVERAGE VALUES

- Prop. No. uBcIuBc*I
MTS Instron UL ~ H B T ~ N L J ~
~~
1 2 1 58 I 82
3 82 90 1O0 65 94 74
R10,20 4 100 113 110 111 111 101
5 26 39 77(65) 51 16 55
6 34 68 53 50 39 71
3 78 85 96 66 92 82
R20,30 90 105 103 104 104 96
5 17 33 72(57) 29 6 53
--``,`,-`-`,,`,,`,`,,`---

Fig. 21), to correct for the instability in the original P-6 plots.

56
A C 1 SP-155
Chenet al
95 = 0662949 0523346 9 2 7 -
TABLE 6 - JSCE FLEXURAL TOUGHNESS PARAMETERS - AVERAGE VALUES
26.6 I 26.5 I 29.1 I 22.9 I

--``,`,-`-`,,`,,`,`,,`---
30.6 39.0 37.3 32.3

10.3 (9.2) 14.5 5.8 6.7
16.3 I 26.0 I 16.6 I 20.0 I
2.2 I - I 2.3 I 1.6 1

Fig. 21), to correct for the instability in the original Pa plots.
TABLE 7 - THEORETICAL FIRST-CRACK DEFLECTIONS, ó (rnrn)

A C 1 SP-It55 95 0 6 6 2 9 4 9 0 5 2 3 3 4 7 863
Testing of FRC 57
U
E
P-
rn
æ
U
U
E
vi
--``,`,-`-`,,`,,`,`,,`---
Ym
W
z
æ
c"
a
W
z
!z
n
5
WY
m
w
32
v,
Y
æ
æ
W
a
c
O
LL
O
E2
t
vi
Y
vi
Lu
z
c
2
00
I
w
2
m
2

A C 1 SP-355 95 IO662949 0523348 7 T T I
58 Chen etal
Flexural Load
First Crack I = ARE~A,,/AREAo,,

$0 = AREAOAEFm /AREA OA,
25.56 -- - --
~:
; ;s.SS -j R20.30 = 101130- lm) I
.- :36 -4
I
R30.50 = 5 ( l ~ 130)
-
I
SJ
fiB 10 ,I F I /H
I IJ I I L i
o \s is the first crack deflection. Net Midspan Deflection
Fig. 1-Definition of ASTM C 1018 (5)toughness parameters
Bending Load
I L. B and H are the span, width and height of the beam respectively.
L!tb is deflection of II150 of span (2mm when span is
6 t b = LI150
Deflection
--``,`,-`-`,,`,,`,`,,`---
Fig. 2-Definition of JSCE-SF4 (6) toughness parameters

A C 1 SP-15.5 9 5 I0bb2949 0 5 2 3 1 4 9 636 I
Testing of FRC 59
.. . . .. ..c . . .. . ..t.. . . .. ..s. - .. -2.. ...... a . . . . . ..- 8 . . --- -.

J - - .. . . .
Type 6: Mix 3 (SFRC 0.75%)

--``,`,-`-`,,`,,`,`,,`---
O 0.5 1 1.5
Mid-Span Deflection (mm)
Fig. &Effect of extraneous deformations on load-deflection curves
+
b
Specimen Pins ‘Leads’
Side elevation End elevation
Fig. 4-Schematic of Japanese yoke loading system

A C 1 SP-155 95 0662749 0523150 358
60 Chen et al
-L
r!
--``,`,-`-`,,`,,`,`,,`---
Specimen
I
Specimen
I
Base of Testing Machine Base of Testing Machine
I

Fig. &Schematic of top LVDT's loading system
Piece of Steel Connected with Beam

Steel bar
\
Specimen
o, 'O
A _u: I 0 '
#
Leads -LVDT
LVDT
Pins Leads
Fig. &-Schematic of side LVDT's loading system

A C 1 SP-155 95 II 0662947 0 5 2 3 1 5 1 2 9 4 I
Testing of FRC 61
O 0.2 O4 O6 O8 1 1.2 1.4 1.6 1.8 2
Net Midspan Deflection imm)
Fig. i-load versus deflection curves from all laboratories for Mix 1 (plain concrete); thicker
lines denote tests carried out with closed-loop testing system
--``,`,-`-`,,`,,`,`,,`---
"T i
Fig. L L o a d versus deflection curves from all laboratories for Mix 2 (0.25 percent steel fibers);
thicker lines denote tests carried out with closed-loop testing system
A C 1 SP-355 95 m 0662949 0523352 320 m
62 Chen et al
o 0.2 0.4 0.8 0.8 1 1.2 1.4 1.6 1.8 2
Net Midspan Deflection Immi
Fig. L l o a d versus deflection curves from all laboratories for Mix 3 (0.75 percent steel fibers);
thicker lines denote tesis carried oui with closed-loop testing system
--``,`,-`-`,,`,,`,`,,`---
O 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Net Midspan Deflection (mm)
Fig. lû-load versus deflection curves from all laboratories for Mix 4 (1.27 percent steel fibers);
thicker lines denote tests carried out with closed-loop testing system
A C 1 SP-155 95 0 6 6 2 7 4 7 0 5 2 3 3 5 3 O67
Testing of FRC 63
O 0.2 0.4 0.8 0.8 1 1.2 1.4 1.8 1.8 2
Net Midspan Deflection imm)
Fig. 11-Load versus deflection curves from all laboratories for Mix 5 (0.2 percent polypropylene
fiben)
25
20
15
Fig. 12-Load versus deflection curves from all laboratories for Mix 6 (0.5 percent polypropylene
fibers)
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-355 95 0662949 0523354 TT3
64 Chen et al
25
UC-MTS UBC-MTS UBC-lnstron UL-INstron HBT-MTS NU-MTS
'
--.- - -
- - Top LVDT'S Top LVDT'S
-- --
-----I
-- - -
........... --- ...
Yoke
--- --
Yoke Yoke
--
Side LVDT'S
-- --- --
-.-.-
--A--- --
20 2 L _I_ .
I L -0- _I_ a L -I- I
i -- -0- --
O 0.2 0.4 0.6 0.8 1 1.2 1.4

Midcpan Oeflection (mm)
--``,`,-`-`,,`,,`,`,,`---
Fig. 13-Typical load deflection curves from each laboratory for Mix 1 (plain concrete)
25
UC-MTS UBC-MTS UBC-lnstron UL-INstron HBT-MTS NU-MTS
- --- -..-..- -.-.-
_.
Top LVDT'S Top LVDT'S Yoke
-----. Yoke Yoke
*.........
Side LVDT'S
20
- _j_ Mix 2: SFRC (Fibre Volume Content: 0.25%)
I
; 15
All curves end at 1 /150of span.
..........................................
O
;- -;. -i
---- -- ----- -i. --- -i- -- -i-
-: :
-I - .~.
2
U
,o -- --- - ------ ------ ----- j -[- - - - - - -;- - - - - - -:.
C
m"
5
......................................................
a I I I I
O 0.5 1 1.5 2
Midspan Deflection (mm)
Fig. 14-Typical load deflection curves from each laboratory for Mix 2 (0.25 percent steel fibers)

A C 1 SP-155 9 5 M 0662949 0 5 2 3 3 5 5 93T
Testing of FRC 65
25
UC-MTS UBC-MTS UBC-lnstron UL-INstron HBT-MTS NU-MTS ;
...........
Yoke Yoke
--- -..-..-
Side LVDT'S
-.-.- -a-
I
--``,`,-`-`,,`,,`,`,,`---
20
A
5z 15
O
J
CD
C
10
C
Ill
m
'
5 _ _ . Mix
_ -3: _SFRC (Fibre Volume Content: 0.75%) -- -:- - - 7 .
All curves end at 1 Il 50 of span.
O 1 I l I
O 0.5 1 1.5 2
Fig. 15-Typical load deflection curves from each laboratory for Mix 3 (0.75 percent steel fibers)
25
20
o,
C
5 10
C
9>
m
5
a
0.5 1 1.5 2
Fig. 1 b T y p i c a l load deflection curves from each loboratory for Mix 4 (1.27 percent steel fibers)

A C 1 SP-155 95 S 0 6 6 2 9 4 9 0523156 ô76
66 Chen etal
UC-MTS UBC-MTS UBC-lnstron UL-INstron HBT-MTS NU-MTS :

. - -
-
Top LVDT'S Top LVDT'S Yoke Yoke
- - 1 - - 1
Yoke
......
..
Side LVDT'S ;
I.I.8 --- -o-..- -.-.-
Fig. 17-Typical load deflection curves from each laboratory for Mix 5 (0.2 percent polypropylene
fibers)
25
UC-MTS UBC-MTS UBC-lnstron UL-INstron HBT-MTS NU-MTS {
.--
-
-----. Yoke Yoke Side LVDT'S -;-
.......t. II --- -..-I.- -a-.-
i Mix 6: PFRC (Fibre Volume Content: 0.50%)

All curves end at 1/150 of span.
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ . _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
O I I I I
O 0.5 1 1.5 2
--``,`,-`-`,,`,,`,`,,`---
Fig. 18-Typical load deflection curves from each laboratory for Mix 6 (0.5 percent polypropylene
fibers)
A C 1 SP-155 '75 0 6 6 2 ï Y î 0523357 702 W
Testing of FRC 67
25
20
TI
15
10
Mix 2
5
Mix 5
O I
O o. 1 0.2 0.3 0.4 0.5 0.6

Net Midspan Deflection (mm)
Fig. 19-Typical load versus deflection curves for low toughness FRC mixes
20
15
z
Y
U
m
O
J 10
o
O 2
Fig. 2û-Load versus deflection curves for same specimen recorded by an X-Y plotter and by a
digital data acquisition system
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-15.5 95 IO b b 2 9 4 9 0.523158 b 4 9
68 Chen et al
20
- 15
5-
U
m
O
-
J 10
3
X
o,
h
5
O
O 0.2 0.4 0.6 0.8 1
Mid-Span Deflection (mm)
Fig. Ill-Effect of irregularly shaped unstable portions of the P-6 curves on calculated toughness
parameters
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 75 0 6 6 2 9 4 9 O523359 585 E
Testing of FRC 69
c
T - UBC-lnatron
16
14
O
Y 12
X
Q 10
-E
U
8
E 0
2 4
O
Mix 2 ISFRC 0.25%) Mix 5 (PFR C 0.20%)
14
u)
u)
QI 12
E
c
cn 10
a
O
c 8
Q
u)
e 0
E
m
Q 4
m
7
2
O
Mix 2 (CFRC 0.25%) Mix 6 (PFRC 0.20%)
@)
Fig. 22-Comparison of (a) ASTM C 1018 I, values, and (b) Japanese toughness values for: Mix
2 (0.25 percent steel fibers), and Mix 5 (0.20 percent polypropylene fibers)
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 75 I0662747 O523360 2T7 =
70 Chen et al
Further Analysis and Interpretation of Results
by Colin D.Johnston
The first three authors are sincerely commended for initiating and
coordinating this interlaboratory comparative testing program, which the writer
considers to have been a worthwhile exercise despite his reservations about
some of the results and conclusions. These reservations are based on an
analysis of the first-crack deflection data (Tables 4 and 7 in the paper) which
reveals characteristics that suggest significant inaccuracy or inconsistent
performances for at least two of the six deflection-measuring systems, and
consequently raises the possibility that conclusions regarding the ability of the
various toughness parameters to distinguish performance differences between
mixtures may have been reached using unreliable deflection measurements.
Obviously, the reliability of the data for toughness parameters and the validity
of the conclusions reached about them depends significantly on the quality of
the deflection measurements and the consistency achieved in their analysis.
Ouality of Deflection Measurements
As indicated in the paper and in ASTM C1018, the accuracy of the

measured first-crack deflection can be examined using the formula that
basically calculates a theoretical first-crack deflection from the first-crack load
P, a modulus of elasticity, and several constants for the beam in question.
Regardless of how the first-crack point is chosen, which as the paper states
may have varied between the labs, once P is identified the measured and
calculated first-crack deflections should correspond closely if the measurement
system is both accurate and consistent in its performance from one test to
another. These deflections are compared in Table A l and differences between
theoretical and measured values are given in parenthesis.
Since the six concrete matrices have essentially the same mixture
proportions, and are in any case nominally identical for any set of specimens
from a given mixture, the overall mean values of calculated theoretical and
measured deflections for all six mixtures should correspond, and the variability
about the means should be reasonably similar. Instead the following
discrepancies can be identified:
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 111 ûbb2949 0 5 2 3 1 b L 1 3 3 I
Testing of FRC 71
1. The UBC MTS, NU and HBT systems measure deflections that are on
average significantly lower than the theoretical values by respectively
26%, 14% and 7%. Agreement is good, within 2%, for the remaining
three systems. This reflects the accuracy of the systems independent
of how first crack has been identified. The mean of the differences
between theoretical and measured values in the last line of Table A l ,
reflects any tendency in the measurement system to consistently
underestimate or overestimate deflection. For a reliable system it
should be close to zero as it is for UC, UBC INS, and UL systems.
For single mixtures differences of 10 to 16 (30% to 46% of theoretical)
are worst cases and occur in the UBC MTS and NU data for Mix 5 and
Mix 6.
2. The consistency of each system from one mixture to another is

reflected in the range of the difference data. In this respect NU and
HBT are quiet variable and UBC MTS is slightly less so. The
coefficients of variation (C of V) for the mean measured deflection are
also high (14.8% to 34.4%) for these three systems. This reflects
inconsistency in the performance of the measuring systems from one set
of tests to another, that is overestimation (negative difference) as well
as underestimation (positive difference) of deflection. The remaining
three systems are more consistent.
Trends in Identification of First Crack
Given that the six concrete matrices are very similar and that the types and
percentages of fibers used are unlikely to greatly change the first-crack load
and the associated theoretical first-crack deflection, and that any changes in
mixture characteristics would be the same for all participating labs, the means
for first-class deflection (and load) should agree if all labs were using the same
approach to identify first crack. The coefficients of variation about the mean
should be similar if each participant was consistent in identifying first crack
from one mixture to another. Instead, the following discrepancies can be
identified:
1. On average for ail six mixtures, NU clearly identifies first crack at the
lowest theoretical deflection and therefore load, with UBC (both)
somewhat higher, and the remaining three in close agreement and
higher still.
2. NU is notably less consistent than the other five according to the

coefficient of variation of 13.7% (next highest is 7.1 %) for the mean
theoretical first-crack deflection and the corresponding load.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0662749 0 5 2 3 1 6 2 0 7 T
72 Chen et al
Nowhere are the discrepancies in the defining first crack more obvious,
and surprisingly so,than for the plain concrete (Mix 1 in Table 4 of the paper)
where the first-crack strength derived solely from load varies from a low of
3.93 MPa for Nu to 4.64 MPa for UBC MTS, and 5.23 to 6.1 MPa for the
remaining four participants (mean 5.7 MPa with C of V of 6%). Given the
fairly widely held view that first crack and ultimate loads are essentially equal
for plain concrete, it is surprising, at least to this writer, that Nu and UBC
MTS identify the first-crack load 31 % and 19% lower respectively than the
mean from the other four participants.
Performance Distinctions bv Fiber TvDe or Fiber Volume
In the paper, Table 8 is used to support the view and the associated
conclusion (item 2 in Conclusions) that the Japanese standard parameters TjscE
and FjscE appear to better distinguish performance differences between
mixtures. Yet the increase from Mix 5 to Mix 2 for R30,50 is 67%, higher
than the corresponding 49% for TJSCE and FjScE. Since the two Japanese
parameters are in fact directly proportional any ranking of performance is the
same for both, so only FjscE will be considered in the following analysis.
A comparison for ail five fiber-reinforced mixtures is presented in

Table A2, despite the fact that the all-lab mean values are influenced to some
unknown extent by the results from labs where deflection measurement
problems have been identified, and that the anomaly of higher values for Mix
5(0.2% fibers) than for Mix 6( 0.5% fibers) exists in the UBC INS data. It
shows that all parameters rank mixtures 4, 3 and 5 in the same order. Mixture
2 is ranked lower than mixture 6 by FJSCE and by R20,30 and R30,50, but the
difference is very smaU for these and most of the other ASTM C1018
parameters.
To compare ail parameters on the same basis, the value for the lowest
level of performance (Mix 5 with 0.2% polypropylene) is expressed as a
percentage of the value for the highest level of performance (Mix 4 with
--``,`,-`-`,,`,,`,`,,`---
1.27% steel) in the last line of Table A2. This percentage is about the same
for FJscE and R30,50, so the argument that the former is better than the latter
is not supported. Moreover, since the performance difference indicated by this
percentage is greater for R3?,50 than for R20,30 or it is probable that
R50,100 which has an end-point deflection comparable to the 2 mm end point
for FJscE would distinguish the performance difference better still. Given that
the end-point deflection for FjscE corresponds to a serviceability condition
with crack width of the order of 2 mrn, that FJscE does not address
serviceability conditions with lower end-point deflections and smaller crack
widths more appropriate for many applications, and that R30,50, R,0,30 and
even R10,20 do address serviceability conditions with lower end-point
deflections while distinguishing performance levels reasonably effectively,

A C 1 SP-155 95 B O b 6 2 9 4 9 0523163 Tob
Testing of FRC 73
ASTM C1018 offers versatility that is absent in JSCE-SF4. Another facet of

this versatility is the fact that the ASTM C1018 parameters are identifiable in
principle for ail sizes and shapes of specimen including the thin sections typical
of some applications that are not adequately represented by the 100 or 150 mm
deep beams required by JSCE SF-4. Identifiable in principle indicates that
they have the same meaning and significance, not that they are "supposed to
be independent of specimen dimensions" as stated in the paper. There are few
absolutes for concrete, and toughness parameters are no different from other
properties such as flexural or compressive strength that also depend on
specimen size and shape.
Summary
--``,`,-`-`,,`,,`,`,,`---
The writer agrees with much of the content of the paper and with the
essence of the conclusions other than item 2. Despite the criticai analysis
offered in this discussion, he believes that much can be learned from the
project. Obviously, achieving accurate and consistent deflection measurement
is more difficult than was thought at the outset. Refining the existing
provisions in ASTM Cl018 for identifying first crack and better
communicating their intent remain a priority. The fact that the effects of any
instability in the load-deflection curve immediately following first crack can be
important and can often be eliminated by the use of residual strength factors
in preference to toughness indices needs to be more widely recognized.
Finally, the reaiity is that engineers involved in design understand the
significance of residual strength better than toughness or toughness indices for
most applications, so specifications and test methods should move towards
more emphasis on residual strength at an end-point deflection serviceability
condition appropriate to the application. In this regard, a residual strength
factor based on the average load over a portion of the load-deflection curve
immediately preceding the specified end-point deflection (e.g. R30,50or
R,,,,,) is more meaningful than a factor based on the average load over the
whole curve including the portion prior to first crack (e.g. FJscE>.

A C 1 SP-15.5 95 W Ob62749 0.523364 942 =
74 Chen etal
TABLE A l - COMPARISON OF CALCULATED THEORETICAL (BOLD TYPE) AND MEASURED

(NORMAL TYPE) FIRST-CRACK DEFLECTIONS (DECIMAL OMITTED) IN mrn, WITH DIFFERENCES IN
PARENTHESIS
Mix uc UBC UBC UL HBT Nu

MTS INS
-
1 042 032 036 038 O40 027
II
038 (4) 024 (8) 029 (7) O41 (-3) 032 (8) 022 (5)
2 038 037 036 039 040 027
I,
039 (-1) 032 (5) 036 (O) 036 (3) 038 (2) 032 (-5)
3 040 034 036 039 O40 034
II
4
II
CON 6.4% I 7.1% I 4.6% 15.3% 1 4.8% I 13.7%

-
Mean O40 026 035 O41 039 027
cow 7.7% 25.5% 11.4% 9.1% 14.8% 34.4%
-
Max.
Min.
Range
Mean
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 I 0662949 0523365 8 8 9
Testing of FRC 75
TABLE A2 - COMPARISON AND RANKING OF MIXTURES BY ALL-LAB MEAN VALUES OF

TOUGHNESS PARAMETERS AND PERFORMANCE DIFFERENCE' (BOLD)
I30
18.1í3)
25.8í2)
3 1.4(1)
15.9(5)
18.0(4)
for Mix 4 (highest performance).

-0loOJJ
a percentage o v ue
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 I0662949 0523166 7 1 5
SP 155-4
Measuring Toughness Characteristics of

SFRC - A Critical View of ASTM C 1018
by D. E. Nemegeer and P. C. Tatnall
SvnoDsis: ASTM C 1018, Standard Test Method for Flexural

Toughness and First-Crack Strength of Fiber-Reinforced Concrete (Using
Beam With Third-Point Loading), is conceived to produce toughness
parameters independent of the dimensions of the test specimen. This
seems not to be true. Additionaly, the toughness indices that are required
to be reported are shown not to be sensitive to the type and amount of
fibers used, and thus do not provide a usable value for characterizing
flexural toughness. Furthermore, since the calculation of the toughness
index values are directly related to the first-crack deflection
measurements, a value which is difficult to determine, these values
become dependent on the testing equipment used. Proposals for revision
of ASTM C 1018 are presented to address these concerns.
--``,`,-`-`,,`,,`,`,,`---
Kevwords: Cracking (fracturing); fiber reinforced concretes; fibers; load-
deflection curve; toughness a;

Not for Resale
A C 1 SP-155 95 0662947 0 5 2 3 3 6 7 6.51
78 Nemegeer and Tatnall
AC1 member Dirk E. Nemegeer is Manager of Building Products

Development with N.V. Bekaert, S.A., Belgium. He received a degree in
civil engineering from the University of Ghent and was closely involved in
implementing Battelle’s Wirand Concrete concept in the Bekaert Group.
He is a member of AC1 Committees 506, Shotcrete, and 544, Fiber
Reinforced Concrete.
AC1 member Peter C. Tatnall is Technical Service Manager, Dramix

Steel Fiber Department, Bekaert Corp., Marietta, GA. He serves as Chair
of ASTM Subcommittee C09.42, Fiber Reinforced Concrete. He also
serves as Chair of AC1 Committee 544’s Subcommittee on Steel Fiber
Reinforced Concrete, and is a member of Committees 302 and 360 on
Slabs-on-Grade, and 506.
Introduction
The ASTM C 1018-92, Standard Test Method for Flexural

Toughness and First-Crack Strength of Fiber-Reinforced Concrete (Using
Beam with Third-Point Loading) is well known and often used as a
reference, both in the United States and abroad (1). Toughness, that is
the energy absorption capacity or the post-cracking behavior, is indeed
seen as the most important physical characteristic of Steel Fiber
Reinforced Concrete and Shotcrete. Measuring and reporting this
characteristic in a correct way is thus of vital importance.
History and Literature
In the current edition of C 1018 published May,l992, the toughness

indices I, and I,, must be determined as the ratio between the area under
the load-deflection curve (L- D curve) up to respectively 3 times 6 and 5.5
times 6 and the area under the same L- D curve up to first crack
deflection, 6. When requested, the toughness index I,, may also be
determined in an identical way but using the area under the L- D curve up
to 10.5 times 6. In addition the residual strength factor, R,,,,,must be
determined as 20 (I,, - I,) and the residual strength factor
R,,,, = 10 (I2, - I,,) may also be determined. Different studies (2 - 8) have
demonstrated that the toughness indices I, and I,, are not sensitive to fiber
type and fiber volume fraction. Reference 7 indicates only I, as not
sensitive, but in this study beam sizes were 800 x 150 x I 0 0 mm, span
750 mm, (31 x 6 x 4 in., span 30 in.) giving a real first-crack deflection of
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-1.55 95 S 0662949 0523168 5 9 8
Testing of FRC 79
--``,`,-`-`,,`,,`,`,,`---
0.24 mm (0.0094in.), while all the other references describe tests with
beam sizes of 100 x 100 x 350 mm (4 x 4 x 14 in.), span 300 mm (12 in.),
or 150 x 150 x 500 mm (6 x 6 x 20 in.), span 450 mm (18 in.) leading to a
real first crack deflection of approximately 0.04 mm (0.0016 in.) or 0.06
mm (0.0024 in.) respectively. The fact that a larger pari of the load-
deflection curve is considered explains the better sensitivity of the I,,
value.
The other references also indicate that using a larger pari of the
load deflection curve I>,, I,, or more, or even the Japanese definition of
toughness (9),going to a deflection of 1/150 of the span length, is much
more capable of distinguishing the differences in toughness due to fiber
type and/or fiber volumes. Several authors also mentioned the difficulty
of measuring accurately the first-crack deflection and the influence of
including extraneous deformations in the determination of toughness
indices (2 - 4,8,10,11). The consequence is that it is very difficult to
determine I, and I,, correctly. Although these conclusions were known at
the time of the latest revision of ASTM 1018, the requirement to determine
I, and I,, was maintained. Indeed, a proposal to omit I, and R,.,,,make I,,
mandatory and to make I, and R,,,3, optional was voted down.
Commercial aspects may have played a more decisive role than scientific
reasons for defeat of this proposal. It may be desirable to modify the
standard so that it is only valid for the case in which a minimum toughness
is provided (no need for substantially brittle materials to look for
toughness values).
But why are the indices chosen as they are? In the original
C 1018 edition published in 1984, the same values, I, and I,, ,were
specified as obligatory, but here I, was also specified as optional
(deflection of 15.5 times 6).
This edition was based on previous work done by Henager (12)

and Johnston (13). Henager found that for the specimen size used, 100 x
100 x 350 mm (4x4~14in.), span 300 mm (12 in.), a limiting deflection
criterion of 1.9 mm (0.075 in.) was the most acceptable. He notes that
this criterion corresponds to about 15 times the first crack deflection (Fig.
1). This deflection corresponds very well to the criterion as accepted in
the Japanese definition of toughness, namely 1/150 of the span, ¡.e., 2.00
mm (0.079 in.) for a span of 300 mm (12 in.). However, this means that
the measured first-crack deflection was approximately O. 127 mm (0.005
in.), thus clearly including a high amount of extraneous deformation. In
fact, the calculated first-crack deflection starting from a first-crack flexural
strength of 6 MPa (857 psi) and a E-modulus of 30 GPa (4.35 x I O 6 psi) is
0.038 mm (0.00015 in.), less than 1/3 of the measured value.

A C 1 SP-155 95 0662747 0523167 4 2 4 =
Also, the tests done by Johnston (13) show measured first-crack

deflections of 0.11 to 0.14 mm (0.0043 to 0.0055 in.). See Fig. 2.
However, the calculated first-crack deflection is in the order of 0.04 mm
(0.0016 in.), about 3 times less. The included extraneous deformations
increase with increasing load. While the load after first crack decreases
or only marginally increases, it means that the measured deflection after
first crack is more accurate or is even somewhat smaller than the real
deflection (in the case in which the extraneous deformations are elastic).
--``,`,-`-`,,`,,`,`,,`---
This means also that the 1984 proposed I, toughness index was
indeed calculated using a pari of the L- D curve after first crack equal to 2
times the measured first-crack deflection and thus at least 6 times the real
first-crack deflection. For I,, , it was approximately 15 times the real first-
crack deflection, and for I,, , the proposed end of the test, it was more
than 40 times the real first-crack deflection. This explains why in these
1984 tests it was possible to distinguish the influence of fiber type and
fiber amount with the I, and I,, indices.
Since then the importance of accurately measuring the real first-

crack deflections, excluding all extraneous deformations, became clear
and in most tests measures are taken to correctly record the net
deflections as required by C 1O 1 8-92.
This leads to two consequences:
I. The prescribed toughness values I, and I,, for the small beams
100 x 100 x 300 mm or 150 x 150 x 450 mm (4x4~12or 6 x 6 ~ 1 8
in.) only consider a very small part of the L - D curve and are not
at all capable of distinguishing the influence of fiber type and
fiber amount on toughness. This is especially true since the
correct determination of the L - D curve immediately after first
crack is often very difficult.
2. Because the net deflections are very small, every error in

measuring deflections becomes a relatively important error.
As mentioned previously, it is not easy to accurately measure the

first-crack deflection. In the case of low toughness, it becomes very
difficult or even impossible to record the L - D curves immediately after
first crack mainly due to the release of stored energy in the testing
machine at the moment of cracking. The calculation of the I, and I,,
values using this part of the L -
D curve is therefore more than
questionable. ASTM ClO18-92 defines the first crack as the point on the
L - D curve at which the form of the curve first becomes nonlinear. The
difficulties in determining the correct load-deflection curves immediately
after first crack are illustrated in the following figures:
A C 1 SP-155 95 m 0662749 0523170 146 m
Testing of FRC 81
Fig. 3 - The L - D curve plotted on a large scale makes it possibe

to record first-crack deflection for unreinforced concrete.
With that particular type of equipment it was possible to
calculate an I, value of 3.4 for unreinforced concrete,
which by definition has toughness indices equal to 1.
Fig. 4 Shows the L - D curve recorded with two different

systems:
curve a: recorded with a digital system.
curve b: plotted directly with a pen.
In neither case is the deflection recorded during crack
formation correct.
Fig. 5 Shows a L - D curve for an apparently very tough

material, plotted on an expanded scale. I, is 6 ; I,, is 1 I
A material with post-cracking strengthening.
Fig. 6 Shows the same L - D curve as in Fig. 5 on a normal

scale. Certainly not a material demonstrating post-
cracking strengthening behavior.
Fig. 7 Shows a L - D curve on a normal scale for a very tough

FRC. The behavior can not be described by using the I,
and I,, toughness indices.
New Proposal
Reauirements:
1. The parameters describing toughness should be independent of first-

crack deflections.
2. To make these parameters dimensionless, they may be related to the

first-crack strength of the fibrous concrete, or even better, to the
flexural strength of the unreinforced concrete with the same
composition.
3. If the first-crack strength is used, it must be determined

unambiquously.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-255 95 0bb2949 0523171 0 8 2
4. Although the toughness can be expressed in dimensionless

parameters, it is not certain that the dimensions of the test specimens
have no influence. Therefore, fixed dimensions must be used for
comparison. At the same time test programs should be undertaken to
find the exact influence of the dimensions of test specimens on these
toughness characteristics.
--``,`,-`-`,,`,,`,`,,`---
5. Toughness values should only be reported if minimum toughness

parameters can be found.
6. The toughness parameters must be based on that part of the L - D

curve where the recording of the curve is not disturbed by the starting
of a crack and the release of stored energy in the test equipment. The
test equipment must be stiff enough to be sure that the release of the
stored energy only marginally influences the first-crack width.
Practical ProDosals for Corrections of C 1018:
1. The first-crack strength should be determined in line with what is done

in several European standards or recommendations (14 - 17). The
definition of first crack should be the point on the L - D curve where the
form of the curve sharply deviates from a straight line. When this point
can not be exactly determined (since there is always a slight curvature
which can always be seen, even for unreinforced concrete if the
magnification is high enough) the first-crack load is determined by
convention as follows:
A straight line is drawn at a deflection of x mm, (x is chosen as a

function of the dimensions of the test specimen) parallel to the
initial straight part of the L -.Dcurve as shown in Fig. 8. The first
crack is determined by convention at the point in the L - D curve
before the intersection with this straight line where the curve shows
a drop in load, and if this point does not exist, at the intersection
itself.
“x” may be taken as the deflection at first crack calculated from the
formula:

A C 1 SP-155 95 0662949 0 5 2 3 1 7 2 T L q
Testing of FRC 83
where: EI is the flexural rigidity.

L is the specimen span.
b is the specimen width.
d is the depth of the specimen.
p is Poisson’s ratio.
P is the load at first crack, or = obd2/L
(T is the first-crack flexural strength.
then for nominal values: D = 5 MPa (725 psi) , and
E = 30 Mpa (4.35 x l o 6psi)

We propose the following “x” dimensions for specimen sizes listed:
x = 0,04 mm (0.0015 in.) for 100 x 100 x 300 mm (4x4~12in.).

x = 0.06 mm (0.0022 in.) for 150 x 150 x 450 mm (6x6~18in.).
x = 0.10 mm (0.0040 in.) for 150 x 150 x 600 mm (6x6~24in.).
x = 0,20 mm (0.0080 in.) for 150 x 100 x 750 mm (6x4~30in.).
--``,`,-`-`,,`,,`,`,,`---
Fig. 8 illustrates how first crack load is determined using the above
definition. A large scale is used to illustrate the method. It can be
seen that the possible error, if any, is relatively small and will have
only a limited influence when used to determine relative toughness
(the deflection at first crack is no longer used as a parameter).
2. The toughness characteristics must be based on fixed deflections

which depend on the dimensions of the test beams. Increases in
crack width will in most cases be proportional to the increase in
deflection (an exception is when multiple cracking is formed). In
general it is very difficult, if not impossible to record in a correct way
the load-deflection diagram immediately after the crack is formed.
Therefore we propose two toughness parameters as described below
and illustrated in Fig. 9:
For a beam with dimensions 150 x 150 x 450 mm ( 6 x 6 ~ 1 8in.):
a) R,,,,,5 for the deflection interval of 0.5 mm (0.02 in.) to 1.5 mm (0.6
in.), and
b) R,,,,, for the deflection interval of 1.5 mm (0.6 in.) to 3.0 mm (1.2
in.).
The dimensionless toughness parameters are calculated as the ratio

between the mean value of the load (or calculated equivalent stress) in
A C 1 SP-155 75 O b b i 9 4 9 0523373 955 m
84 Nernegeer and Tatnall
the considered deflection interval, and the value of the first-crack load
(or equivalent first-crack stress). These values are comparable with
the previous Residual Strength Values, Rx,y,as defined in ASTM
C1018. They also can be used in relation to the serviceability
requirements of the structure
3. To obtain comparable values it is recommended that fixed values be

used for specimens. The proposal cross section is 150 x 150 mm,
span 450 mm (6x6~18in.). At the same time, test programs should be
started to examine the influence of the specimen dimensions on both
first crack and toughness parameters.
4. To avoid further confusion regarding the real measurement of

toughness parameters of fiber reinforced concrete or shotcrete (¡.e. its
energy absorption capacity), it should be stated that no toughness
values are reported if R o,5.1,5, or R 1,5.3,0 are less than 0.3.
5. The argument that the test requires too much time if large deflections
must be measured can easily be solved by increasing the rate of
increase of deflection. Increasing this rate to 0.5 mmlmin. (0.02
in./min.) has almost no influence on test results (9).
References
--``,`,-`-`,,`,,`,`,,`---
1. “ASTM C 1018-92, Standard Test Method for Flexural Toughness and

First-Crack Strength of Fiber Reinforced Concrete (Using beam with
Third-Point Loading)”, ASTM Standards, Pari 04.02, Concrete and
Aggregates, 1993, pp. 514-520.
2. Gopalaratnam, Vellore S., Shah, Surndra P., Batson, Gordon B.,

Criswell, Marvin E., Ramakrishnan, V., and Wecharatana, Methi,
”Fracture Toughness of Fiber Reinforced Concrete AC/ Materials
‘I,
Journal, V. 88, No. 4, July-August 1991, pp. 339-353.
3. Johnston, C.D., Discussion of “Fracture Toughness of Fiber

Reinforced Concrete,” by V.S. Gopalaratnam, et.al., AC/ Materials
Journal, V. 89, No. 3, May-June 1992, pp. 304-309.
4. Banthia, Nemkumar and Trottier, J.-F., Discussion of “Fiber Type

Effects on the Performance of SFRC”, by Parviz Soroushian and Ziad
Bayasi, AC1 Materials Journal, V. 89, No. 1, January-February 1992,
pp. 106-107.

A C 1 SP-155 7 5 H Ob62949 0 5 2 3 1 i ï 4 891
Testing of FRC 85
5. Banthia, Nemkumar and Trottier, Jean-Francois, Wood, David and

--``,`,-`-`,,`,,`,`,,`---
Beaupre, Denis, ?Influence of Fiber Geometry in Steel Fiber

Reinforced Dry-Mix Shotcrete,? Concrete International, V. 14, No. 5,
May 1992, pp. 24-28.
6. Wood, David F., Banthia, Nemy, and Trottier, J.-F., ?A Comparative

Study of Different Steel Fibers in Shotcrete,? Engineering Foundation,
Shotcrete for Underground Support VI, Niagara-on-the-Lake, Canada,
May 2-6,1993, pp. 57-66.
7. Johnston, C.D., and Skarendahl, A., ?Comparative Flexural

Períormance Evaluation of Steel Fiber-Reinforced Concrete According
to ASTM C 1018 Shows Importance of Fiber Parameters,? RILEM
Materials and Structures, Vol. 25, No. 148, May 1992, pp. 191-200.
8. Mindess, S., Chen, L., and Morgan, D.R., ?Determinationof the First-
Crack Strength and Flexural Toughness of Steel Fiber Reinforced
Concrete,? To be published in ACBM Journal, 5th issue.
9. ?Method of Test for Flexural Strength and Flexural Toughness of Steel

Fiber Reinforced Concrete, JSCE-SF-4,? Concrete Library of JSCE,
Japan Society of Civil Engineers, No. 3, June 1984, pp. 58-66.
1O. Johnston, C.D., ?Toughness of Steel Fiber-Reinforced Concrete,?

Proceedings of a U.S.-Sweden Joint Seminar, Swedish Cement and
Concrete Institute, Ed. S.P. Shah and A. Skarendahl, Elsevier Applied
Science, London, 1986, pp. 333-360.
1I.
El-Shakra, Zeyad M., and Gopalaratnam, Vellore S., ?Deflection
Measurements and Toughness Evaluations for FRC,? Cernent and
Concrete Research, Vol. 23, 1993, pp. 1455-1466.
12. Henager, C.H., ?A Toughness Index of Fiber Concrete?. Testing and

Test Methods of Fiber Cement Cornposites, RILEM Symposium, The
Construction Press, Ltd. Sheffield, England, April 1978 , pp. 79-86.
13.Johnston, C.D., ?Definition and Measurement of Flexural Toughness

Parameters for Fiber Reinforced Concrete,? ASTM, Cernent Concrete
and Aggregates, CCAGDP, Vol. 4, No. 2, Winter 1982, pp. 53-60.
14.Belgian Standard NBN Bl5-238. ?Tests on Fiber Reinforced

Concrete. Bending test on prismatic samples,? Institut Belge de
Normalisation (IBN), Bruxelles, September 1992, 9 pp., (In Dutch and
French) .

A C 1 SP-355 75 W Ob62747 0 5 2 3 3 7 5 7 2 8
15. Normalisation Francaise P18-409. “Concrete with steel fibers - flexural

strength test,” Association Francaise de Normalisation, Paris, April
1993, 8 pp., (In French).
16.“Merkblatt : Grundlagen zur Bemessung von Industriefussböden aus

Stahlfaserbeton,” Deutschen Beton-VereinE.V., Wiesbaden, February
1991, 14 pp. (In German).
17.Research Committee Cur Recommendation 35 (Determination of

flexural strength, flexural toughness and equivalent flexural strength of
steel fiber reinforced concrete), Appendix to Cernent, 1994, No. 2,
February, 12 pp., (In Dutch).
:
<II
8
2
3
O
n z
Fig. l - û r i g i n a l toughness criterion by Henager
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 9 5 W 0662997 0 5 2 3 3 7 6 664 W
Testing of FRC 87
.x -
o
c
-
o
L
o -
o
--``,`,-`-`,,`,,`,`,,`---
r
-I
a
/

A C 1 SP-1.55 95 0662949 0523177 S T O
aa Nemegeer and Tatnall
A
Load F(W
3.0 -
2.0 -
1.0 -
O
'6' 9 6 . 1 5.56.2 0.3
Defiection (rnm)
Fig. 3-L-D curve for unreinforced concrete
Load f F(kN)
1 2
--``,`,-`-`,,`,,`,`,,`--- Deflection (mrn)
Provided by IHS under license with ACI Fig. 4-1-D turves recorded with two sytems
A C 1 SP-355 95 Ohh29Y9 0523378 437 =
Testing of FRC 89
First crack
21.2 kN
20
10
I I
I
I I I
*
I I I
O I , I
02 0.3
-s,+ w Deflection (m)
Fig. !i-L-D curve for apparently very tough concrete (large scale)
0 1
1.o 2.0
--``,`,-`-`,,`,,`,`,,`---
DekUOn (mm)
Copyright American Concrete Institute Fig. &Same L-0 curve as Fig. 5 at normal scale
A C 1 SP-155 95 Ob62949 0 5 2 3 3 7 9 373
7
70
-i
50
45
.^ I
Beam
i50 X 150 X 500 (450) mm3
,..-.
'- .'
Daflarttvi immi
y
16i1
o
0X-J o! 1!0 115 2!0 2!5 3!0
Fig. 7-1-D curve for very tough FRC
Load
fc A
fc B
fc c
fc D
. .
0.1 O 2 03 0.4 0.5 Deflection (

m m)
0.04
Fig. L P r o p o s e d determination of first crack

--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-L55 95 m 0662949 0523380 095
Testing of FRC 91
4
--``,`,-`-`,,`,,`,`,,`---
Frn. 0.5 - 1.5 h,1 3 - 3,O

Ro,5 - 1.5 = R1,5 - 3.0 =
FC Fc
Fig. 9a-Proposed toughness parameters Ro545 and R15-3,0-strain hardening
Frn, 0,5 - 1.5 Fm, 1 3 - 3,O

R0,S - 1.5 = R1.5 - 3.0 =
FC FC
Fig. 9b-Proposed toughness parameters %5.i5 and R15-3,0-strain softening

A C 1 SP-155 95 I0662949 05231ô1 T2L
SP 155-5
Experimental R-Curves for Assessment

of Toughening in Fiber Reinforced
Cementitious Composites
by E. Mobasher, C. Y. Li, and A. Arino
--``,`,-`-`,,`,,`,`,,`---
SvnoDsis: Procedures to obtain the experimental R-Curves using a

compliance calibration technique are revisited. R-Curves provide a
convenient means to study the process of fracture and the brittle-
ductile transition in materials. Single edge notched beam specimens
are tested under closed loop crack mouth opening control. The
procedure t o obtain the R-Curves using loading/unloading compliance
and the residual displacements are discussed. An elastically
equivalent toughness KR as a function of crack extension is defined
t o compare the R-Curves with the available data in the literature.
The developed test method is applied to FRC composites with up t o

8% by volume of short, chopped alumina, carbon, and polypropylene
(PP) fibers. Significant strengthening of the matrix due t o the
addition of short carbon and alumina fibers are observed. R-curves in
these composites are characterized by an increase in the steady state
fracture toughness. In PP-FRC composites, energy dissipation due to
fiber pullout increases the ascending rate of the R-curve well after
the main crack has formed. The work of fracture is computed from
the cyclic loading unloading tests and the results are compared with
the R-Curves.
Kevwords: Aluminum oxide; carbon; composite materials; concretes;

cracking (fracturing); fiber reinforced concretes; fibers; fracture
mechanics; mortars (material); polypropylene fibers; R-curves; toughness

Not for Resale
A C 1 SP-155 9 5 0662949 0 5 2 3 3 8 2 968
94 Mobasher, Li, and Arino
AC1 member Barzin Mobasher, Ph.D. is an assistant professor at the

Department of Civil Engineering, Arizona State University. His
research activities include fiber reinforced cement based composites,
toughening mechanisms, durability of concrete, and experimental
mechanics.
Cheng Yu Li is a Ph. D. candidate at the Department of Civil

Engineering, Arizona State University. His research interests are in
the areas of fiber reinforced cement based composites and bridge
dynamics.
--``,`,-`-`,,`,,`,`,,`---
Antonio Arino is a research assistant at the Department of Civil
Engineering, Arizona State University. His research interest is in the
area of high strength fiber reinforced concrete.
INTRODUCTION
Study of fracture in cement based composites suggests a

strong interaction between the microstructure of the material and the
process zone behind the crack tip. In fiber reinforced composites,
wake processes that are affected by the presence of fibers include
crack deflection, bridging, and pullout of fibers. It is generally
accepted that the energy dissipation in the crack wake region is a
dominant toughening mechanism which enhances the load carrying
capacity. Since the process zone depends on specimen size,
material, loading geometry, and the method of measurement, single
parameter fracture toughness criteria fail t o address the failure
processes objectively [ I ] . There is a need to better characterize the
fracture test results of FRC materials.
The loading path in the response of a specimen under closed

loop testing with displacement or crack opening control conditions
consists of a linear and a nonlinear response before the maximum
load. The descending branch is commonly called the strain softening
zone, and represents crack opening and growth under steady state
conditions. Determination of the crack length during the steady
state propagation regime permits quantitative modeling based on
nonlinear fracture mechanics.
Nonlinear nature of energy dissipation in FRC materials causes

hysteresis in the loading-unloading experiments. Linear elastic
A C 1 S P - 1 5 5 95 Obb2747 0523383 8 T 4
Testing of FRC 95
fracture mechanics (LEFM) does not address such inelastic processes

which are due to microstructural effects and inhomogeneities. Since
the crack propagation occurs before the peak load has reached,
definition of the stress intensity factor based on the original notch
length would lead to material parameters that are size and specimen
geometry dependant.
CRACK RESISTANCE APPROACH
The crack growth resistance or the R-Curve behavior of quasi-

brittle materials has been investigated in detail. Lenain and Bunsell
[2]used R-Curves for asbestos cement mixtures. Mai obtained R-
Curves using experiments on three-point bend and grooved double-
cantilever beams [31. Wecharatana and Shah [4]applied R-Curves t o
cement-based matrices and fiber reinforced composites. Foote, Mai &
Cotterell 151 studied R-Curves in strain softening materials. Sakai et
al. [61 and Hsueh & Becher [71 used R-Curves in the study of fracture
in ceramics.
R-Curve models integrate the closing pressure of fibers in the

process zone as a toughening component of the matrix material.
There are no standardized procedures to determine the R-Curves.
Approaches which are based on the energy principle and the
unloading-reloading methods have been quite convenient for
evaluating nonlinear fracture toughness parameters as functions of
crack length [81. These ideas relate the energy dissipation in the
process zone to an effective elastic crack length.
The strain energy release rate, G, is the source of total energy

available for crack extension. Once it reaches a critical value G, , an
--``,`,-`-`,,`,,`,`,,`---
instability condition is reached and crack propagation occurs. This is

shown as the horizontal line in Fig. l a . To characterize fracture
toughness using a single parameter G, , only the peak load of a
notched specimen tested under mode I condition is required. Quasi-
brittle materials dissipate energy due to frictional sliding, aggregate
interlock, and crack surface tortuosity. After an initiated crack begins
t o propagate, the dissipating mechanisms evolve. The increase in the
apparent toughness can be related to the stable crack growth by
means of an R-Curve. This is shown in Figure 1 .a for quasi brittle
and FRC materials [91. The condition for stable crack growth is:

A C 1 SP-155 95 = Ob62949 05231B4 730
The condition for crack instability can be defined as:

--``,`,-`-`,,`,,`,`,,`---
G(a,) =R(a,) - - --
aa ¿%I
a a= a, (2)
A formulation by Ouyang et al. t91 suggests that R-Aa curves

are an extension of the elastically equivalent fictitious crack models
(linear elastic materials) to the actual specimen with a cohesive zone.
Incorporating the cohesive zone into the R-Curves enables one t o
apply linear elastic approaches t o materials with an effective fracture
toughness, K',, , or K',, [IO]. Since the R-Curve is material,
geometry, and size dependent, it can conveniently be used t o study
stable crack growth and toughening. The equivalent elastic approach
proposed here allows one to use a linear analysis instead of a
nonlinear structural analysis. It can also predict the load-
displacement, and toughness effects of various fiber types.
The ComDliance ADDrOaCh
A possible means of determination of the R-Curve is through

the use of G, the strain energy release rate which is obtained using a
compliance approach. When a notched specimen exhibits
infinitesimal crack growth under constant load or displacement
conditions, changes in the load-deformation response before and after
the crack propagation are observed. The compliance, C, defined by
the inverse of the slope of the load displacement curve is a function
of the crack size. This is shown in Figure 1.b.
Using the LEFM formulas, the instantaneous compliance can be

compared to the initial compliance. The initial compliance of the
specimen is used to calculate the young's modulus of the material.
This modulus is used with the unloading compliance, Ca", in a non-
linear equation t o solve for an effective crack length according t o
equation 3:
6 S ( ao + Au) V(a) ao+Aa
fia) = E - =o , a =- (3)
Ce, b2 t b
A C 1 SP-155 95 111 0 6 6 2 9 4 9 0 5 2 3 1 8 5 b 7 7 m
Testing of FRC 97
V(a) = 0.76 - 2.28a + 3 . 8 7 ~- ~2.Wa3+

~ - (4)
( i -al2
+
By measuring (P, a =a, Aal values at successive intervals of crack
growth, the compliance-crack length relationship is constructed.
The stress intensity factor at the tip of the effective crack may be
obtained using Equations 5 and 6, and reported as the R-Curve, KR.
This definition of the R-Curve is referred to as a modified LEFM
approach:
--``,`,-`-`,,`,,`,`,,`---
1.99-0((1-a)(2.15-3.93a +2.7a2)
F(a) = (6)
Jx(i+2a)(1- ~ r ) ~ n
Alternatively, one can compute the rate of change of

compliance as a function of effective crack length. This rate can be
obtained using polynomial curve fitting to the compliance data, and
differentiating the resulting curve. Differentiation can also be
achieved using a local algorithm such as the cubic spline method. By
assuming crack growth under constant load, the energy release rate
due to the incremental crack growth can be obtained as:
G ( u ) = 1 -
- a c p 2
2 t au
Both the stress intensity approach shown in equation 5, and

the compliance rate shown in eq. 7 can be used to obtain the strain
energy release rate. The first method provides a local measurement
of the energy release rate and the second method provides an
averaged value. Both methods are theoretically equivalent under
LEFM assumptions, and depending on the nature of the experimental
data, the compliance approach underestimates the R-Curve due to the
averaging effect.
In the presence of residual displacements, additional terms are

needed t o account for the rate of change of inelastic displacement
with respect to crack growth. Both Wecharatana and Shah, in
addition to Mai and Hakeem [ l II, have proposed additional energy
A C 1 SP-155 95 I
I 0 6 6 2 9 4 9 0 5 2 3 3 8 6 503
98 Mobasher, ti, and Arino
terms as:
1 8%
G * ( u )= G ( u )+ - -
2 t aa
Equation 8 includes the effect of residual displacements in the
strain energy release term. This parameter is used t o define an
elastically equivalent effective toughness, KR, referred t o as the total
toughness:
KR (U) = JE;G'o (9)

--``,`,-`-`,,`,,`,`,,`---
where E', = E, / (1- v:) for plane strain and E, for plane stress. E,
and v, represent the elastic modulus and the poisson's ratio of the
composite.
The present approach utilizes the entire loading history as

opposed t o only the peak load and the effective compliance at that
point. The R-Curve is dependant on the size and geometry of the
specimen and represents material's resistance t o initiation and
propagation of cracks. It has been previously shown that R-curves
can be used t o compare the effectiveness of various ductile and
brittle fiber reinforcements [ 121.
EXPERIMENTAL PROGRAM
The present approach was evaluated for a wide range of FRC

composites. Mortar specimens reinforced with various volume
fractions of alumina, carbon and Polypropylene (PP) fibers were
prepared. High purity Kaolin based alumina-silicate fibers were
obtained .from Carborandum Corp., Niagra Falls NY (Fiberfrax@)[13].
Carbon Fibers were obtained from Ashland Petroleum Company,
Ashland Kentucky (Carboflex@). Polypropylene fibers were Krenit
fibers obtained from Danaklon, Denmark [ I 41. Physical properties of
the fibers are shown in table 1. Note that the average diameter of
ceramic fibers is less than I p m that is significantly smaller than
average anhydrous portland cement particles. A detailed description
of the use of these fibers is provided elsewhere [151.
The mortar matrix composed of type 1/11 portland cement, sand,

and silica fume with weight proportions of 1 :0.8:0.15. A constant
water to cementitious solids (cement silica fume) ratio of 0.3 was +
used for all specimens. Specimen dimensions were 2 5 x 7 5 ~ 3 2 5mm.
A C 1 SP-355 95 a Of=b2949 0 5 2 3 3 8 7 4 4 T
Testing of FRC 99
After 28 days of curing, cyclic loading-unloading tests were carried

out on three point bending specimens. A test span of 305 mm (12
inches) with constant notch length of 19 mm ( 0.75 inch) was used.
LOAD-DEFORMATION RESPONSE OF FRC MATERIALS
Addition of short brittle fibers can significantly increase the

strength of cement based composites. Cyclic load-deflection
response of a specimen with 1% alumina fibers is compared with
plain mortar in Fig. 2. The shapes of these curves are similar and
show the increase in load carrying capacity of the composite. The
response is quasi-linear up to the peak load and the specimens retain
most of their tangential stiffness. Brittle response beyond the peak
load may be attributed to the short length of chopped fibers that are
unable t o affect the post-peak macrocracking in the composites.
Whiskers increase the ultimate strength by stabilizing the microcracks
that occur before the peak load.
At nominal volume fractions, polypropylene (PPI fibers increase
--``,`,-`-`,,`,,`,`,,`---
the ductility by bridging the matrix cracks. Load deflection response
of a specimen reinforced with 4% PP fibers that are 12 mm long is
shown in figure 3. The nonlinear effects due to inelastic deformation.
and the toughness increase are significant. Work of fracture was
defined as the area under the entire load-deflection curve. i n
comparison to the plain matrix (shown in Fig. 21, it is observed that
the work of fracture increases as much as fifteen times. The ductile
response beyond the maximum load is due t o the closing pressure
exerted by the fiber pullout. Note that the response of the beam
under the load may be compared to an elastic-plastic solid.
DISCUSSION OF RESULTS
Changes in the compliance of the brittle and ductile specimens

as a function of crack opening displacement are shown in Fig. 4. In
both cases, the compliance increases due to crack growth, and the
value measured from the loading portion of the curve is higher than
the unloading portion. This difference may be viewed negligible in
the case of mortar and the brittle fiber composites. There is a
significant difference in the loading-unloading cycles of PP fiber
composites. This points to the energy dissipation during the cyclic
deformation process. It is furthermore shown that the compliance
increases much faster in the plain matrix as compared t o the
A C 1 SP-155 95 = 0662949 0523188 3 8 6
composite with PP fibers. As shown in Figure l.b, the initial portion

of the load displacement response during unloading was used in
compliance measurement.
In the fracture of brittle materials, energy dissipation due t o

inelastic processes is assumed to be negligible. This assumption
neglects the effect of inelastic displacements, and leads t o the use of
either equation 5, or 7 as measures of R-Curve. Two different
--``,`,-`-`,,`,,`,`,,`---
methods are used to derive the KRat various crack lengths for four
replicate FRC specimens with 1% Alumina fibers. One approach is
based on the use of equation 5 (Modified LEFM approach), and the
other is based on the use of equation 7 followed by equation 9
(compliance method). These methods are compared in Fig. 5. The R-
Curve increases considerably beyond the maximum load and
asymptotically approaches a constant toughness level. This may be
viewed as a steady state crack growth condition. The KRcurve at
this level is as much as 70% higher that its value at the maximum
load. The differences between the t w o methods are well within the
scatter of the test results.
The approaches discussed above neglect the inelastic

deformation effects. In order to account for this effect, inelastic
deformations were measured from the loading-unloading responses
and plotted as a function of crack extension. Figure 6 presents the
change in the inelastic CMOD as a function of crack extension
computed based on the definition of Figure 1.b. Note that with PP
fiber composites, significant nonlinear deformation takes place after
the peak load. For each specimen, the derivatives of the inelastic
deformation with respect to crack length (figure 61 were obtained.
Results were used in the computation of the additional term to the
energy release rate. The contribution of inelastic deformation t o the
total energy release rate were calculated using equations 8 and 9.
Variations in the KR-Aacurve for several specimens with

various fibers are shown in Fig. 7. For each composite, t w o curves
which represent the modified LEFM (no inelastic deformation, Eq. 5),
and the total KR (with inelastic deformation, Eq. 8 and 9) are shown.
It is observed that the inelastic energy dissipation comprises a
significant portion of the toughness of all composites. This value
may not be neglected even for the case of mortar specimens.
Although the general shape of the two curves are quite similar for
brittle fiber composites, total toughness is almost twice the modified
LEFM approach. The total KRresponse suggests an initial rise in the
curve followed by a flat portion which is characteristic of steady
state crack growth region.

Testing of FRC 101
The total KR-Aacurves for high volume content composites

with 8% carbon and 4% PP fibers may also be compared with a
mortar specimen in Fig. 7. Significant strengthening of the
composite due to the use of short carbon fibers is observed. The
fracture toughness increases by as much as 100% for 8% carbon
fibers. Note that the main toughening effects are due t o an increase
in the ascending rate of the response. R-Curves for PP fiber
composites are quite close to the matrix’s response during the initial
loading cycles. During the steady state cracking process of matrix,
fiber pullout toughening results in a significant increase in the slope
of the R-Curve. The rise in the R-Curve shows the ductility offered
by the PP fibers. Note that these specimens fail t o reach a steady
state condition. This indicates that the test results may be
significantly affected by the size of the specimens. Study of size
effect in these composites would be necessary to further evaluate the
fracture process.
Work of Fracture Analysis
R-Curves were defined using the strain energy release rate and
an elastically equivalent LEFM model. The energy release rate is a
fraction of the total potential energy of the system that is not
consumed due to the irreversible processes. In the present approach,
the contribution of the inelastic deformation to the total toughness
was measured. The work of fracture method however, includes the
energy dissipation due t o the crack growth and other dissipating
mechanisms in the process zone.
Work of fracture, ywo,, defined as the total area under the

--``,`,-`-`,,`,,`,`,,`---
load-deflection curve normalized with respect to the cracked ligament

has been used to assess the degree of toughening for various fiber
composites[l6]. For an elastic-brittle material, the work of fracture is
equivalent to the energy absorbed in the crack propagation. This
measure is also used by ASTM (21018-89 [I71 as a convenient way
t o measure toughness.
Fig. 8 shows the procedure to calculate the work of fracture

from the cyclic load deflection plots. By integrating the load
deflection response of the specimens from cycle t o cycle, energy
dissipation can be obtained. Two parameters are measured for each
cycle. The cyclic energy was defined as the amount of energy
dissipated during a cycle of unloading and reloading, (¡.e., from point
A to 6). It was assumed that no crack growth takes place in this
cycle. Crack extension was assumed to take place during the path
from point B t o C.
A C 1 SP-355 75 = 0 6 6 2 7 ~ 70523370 T 3 4
The total energy, ywof, was obtained by summing the area

under the load-deflection responses from points A to C for each cycle
of loading-unloading. As shown in Fig. 9a, the matrix phase has a
negative curvature due to a diminishing load carrying capacity.
Similar plots are shown for PP-FRC in Fig. 9b. Note that the
toughness of PP-FRC composites is an order of magnitude higher
than the matrix phase. Contrary to the plain matrix, ductile
composites have a positive curvature since the energy absorption
increases at the later stages of the loading process. For PP-FRC, a
significant portion of the energy dissipated ( as much as 1/3 1 takes
place during the cyclic region. Fig. 1 0 presents the energy absorption
results for various fiber composites discussed. Note that the cyclic
component of the energy in the PP fiber composites is significantly
higher that the fracture energy of the matrix phase.
CONCLUSIONS
This study indicates that experimental evaluation of the R-

Curves based on loading /unloading responses of specimens tested in
three point bending can provide significant information regarding the
fracture process in cement based materials. R-Curves were obtained
for brittle mortar specimens in addition t o composites made with
ductile fibers. It is observed that even in mortar specimens, the
inelastic deformation contributes significantly t o the energy
absorption processes. Neglecting this term results in a drastic
underestimation of the R-Curve. Short-brittle fibers increase the
strength of the composite considerably while increasing the
ascending rate of the R-Curve. PP-FRC materials show an increase in
the ascending rate of the R-Curve response well after the cracks have
formed. Comparison of the work of fracture with the R-Curves
shows that the results agree with the work of fracture method of
calculation of the toughness.
Acknowledgements
Authors acknowledge the Research Initiation Award from the

--``,`,-`-`,,`,,`,`,,`---
National Science Foundation (Grant No. 82-MSS9211063 , Program

Director Dr. Ken Chongl
References
1. Mindess, S.,"The Fracture Process Zone in Concrete,"
Toughening Mechanisms in Quasi-Brittle Materials, S.P. Shah
(ed.), 1991, Kluwer Academic Publishers, pp. 271-286.
A C 1 SP-155 9 5 I0662949 0523191 9'70 I
Testing of FRC 103
2. Lenain J. C., and Bunsell A. R., J. of Mat. Sci., Vol. 14, pp.
31 2-332 (1979).
3. Mai Y. W., "Strength and Fracture Properties of Asbestos-

Cement Mortar Composites," Journal of Materials Science,
Vol. 14, 1979, pp.2091-2102.
4. Wecharatana M., and Shah S . P., ASCE, J. of Eng. Mech.,

Vol. 109, 1983, pp.1231-1245.
5. Foote M. L., Mai Y. W., and Cotterell B., J. of Mech., Phys.,

and Solids, Vol. 34, No. 6, 1986, pp. 593-607.
6. Sakai M., Yoshimura J., Goto Y. and inagaki M., J. of Am.

Ceram. Soc., Vol. 71, No.8. pp. 609-61 6.
--``,`,-`-`,,`,,`,`,,`---
7. Hsueh C. H., and Becher P. F., J. ofAm. Ceram. Soc., 1988,

Vol. 71, NO. 5, pp. 234-237.
8. Sakai, M., and Bradt, R.C.,"Graphical Methods For

Determining the Nonlinear Fracture Parameters of Silica and
Graphite Refractory Composites," in Fracture Mechanics of
Ceramics, Vol. 7 Edited by R.C. Bradt, A.G. Evans, D.P.H.
Hassleman, and F.F. Lange, Plenum Press, New York, 1986,
pp.127-42
9. Ouyang, C., Mobasher, B., and Shah, S . P., "An R-Curve

Approach for Fracture of Quasi-Brittle Materials," Engineering
Fracture Mechanics, Vol. 37, 1990, pp. 901 -913.
1o. Karihaloo, B.L., Carpinteri, A., and Elices, M., "Fracture

Mechanics of Cement Mortar and Plain Concrete," Journal of
Advanced Cement Based Materials, 1993, 1, pp. 92-105.
11. Mai, Y.W., and Hakeem, Slow Crack Growth in Cellulose Fibre
Cements," Journal of Materials Science, 19, ( 1984) 501-508.
12. Mobasher, B., Ouyang, C. S . , and Shah, S . P.,"Modeling of

Fiber Toughening in Cementitious Composites using an R-
Curve Approach", Int. J. of Fracture, 50: 199-219, 1991.
13. Mobasher, B., and Li, C. Y.,"Fracture of Whisker Reinforced

Cement Based Composites," Proc., Int. Symp., Brittle Matrix
Composites 4, (BMC4) Cedzyna, Poland, Sep. 1994, pp. 11 6-
124.

A C 1 SP-355 95 IOb62949 0 5 2 3 3 9 2 8 0 7
14. Krenchel H., and Stang, H., "Stable Microcracking in

Cementitious Materials," Proc., lnt. Symp., Brittle Matrix
Composites 2,(BMC2) Cedzyna, Poland, Sep. 1988, pp. 20-
33.
15. Mobasher, B., and Li. C.Y.,"Mechanical Properties of Hybrid

Cement Based Composites," Manuscript in Review, AC1
Materials Journal, 1994.
16. Gopalaratnam, V.S., Shah, S.P., Batson, G.B., Criswell, M.E.,

Ramakrishnan, V., and Wecharatana, M.," Fracture Toughness
of Fiber Reinforced Concrete", AC1 Materials Journal, Vol. 88,
NO. 4, 1 9 9 1 339-353.
~ ~
17. "Standard Test Method for Flexural Toughness and First-Crack

Strength of Fiber Reinforced Concrete (Using Beam With Third-
Point Loading) (C 1O1 8-89)," 7989 Book of ASTM Standards,
Part 04.02, ASTM, Philadelphia, pp. 499-505.
TABLE 1 - PHYSICAL PROPERTIES OF FIBERS
Fiber Lf d' Ultimate Elastic Density

(mml (,um) Strength Modulus (g/cm3) --``,`,-`-`,,`,,`,`,,`---
íMPa) (GPa)
alumina 0.762 2.5 1725 105 2.70
carbon 1 .o 25 1800 230 1.90
PP 12.0 35x250 340-500 8.5-1 2.5 0.91

A C 1 SP-355 9 5 0 6 6 2 9 4 9 0523393 743
Testing of FRC 105
Fiber Reinforced Material
Brittle Matrix (LEFM)
Crack Extension
7
%= ao+Aai =u a
Fig. la-R-curves representing resistance to crack growth for brittle, quasi-brittle, and fiber
reinforced materials. The instability condition is the point of tangency of R-curve with the strain
energy release rate, G
CMOD H H
S = 4b
Fig. 1b-Definition of compliance and inelastic residual deformation of material
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0 b b 2 9 4 9 0523374 bBT
900 I
--``,`,-`-`,,`,,`,`,,`---
Deflection, mm
Fig. 2-Cyclic load-deflection response of plain mortar and specimen reinforced with 1 percent
alumina fibers

A C 1 SP-155 7.5 I0 6 6 2 9 4 9 0523395 5 L b M
Testing of FRC 107
1000 I
4% PP
800
--``,`,-`-`,,`,,`,`,,`---
600
400
200
-
O 1 1 1 1 1 1 1 1 1 ( 1 ' 1 1 1 1 ' 1 1
Fig. 3-Cyclic load-deflection response of specimen reinforced with 4 percent polypropylene fibers

A C 1 SP-355 95 0662749 0 5 2 3 3 9 6 452

--``,`,-`-`,,`,,`,`,,`---
- -A- 4% PP (Loading cycle)

4% PP (Unloading cycle
- - Mortar (Loading)
* , Mortar (Unloading)
O
0.0
1
0.2 0.6 0.8 1 .o 0.4
CMOD, mm
Fig. &Loading and unloading compliance of mortar and specimen reinforced with 4 percent
polypropylene fibers as function of crack mouth opening displacement

A C 1 SP-355 95 m Obb2747 0523377 377 I
Testing of FRC 109
40 - * *
-
O O *
I
O 1 1 1 1 ~ 1 ~ ' ~ ~ ~ '
O 10 20 30 40
Crack Extension, mm
Fig. E-R-curve responses based on two methods of modified LEFM approach and compliance
approach
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-L55 9 5 Obb2949 0523198 225 =
I.5
E
E 1.0
ai
O
2
o
o
.-
ci
(B
-a
(II
0.5
-t
--``,`,-`-`,,`,,`,`,,`---
o.oq-p-l , , ~
I I I I I l I I I l I I
O 10 20 30 40
Crack Extension, mm
Fig. b l n e l a s t i c deformation as function of track extension obtained using compliance calibration
technique

A C 1 SP-155 9 5 .8 0 6 6 2 9 4 9 0 5 2 3 3 9 9 I b L I
Testing of FRC 111
300
-- e- Mortar, Modified LEFM
Mortar, Total
-
* -
-*
8% Carbon, Modified LEFM
--
8% Carbon, Total
200 - - - 4% PP, Modified LEFM
- 4% PP, Total
loo/
--``,`,-`-`,,`,,`,`,,`---
O 10 20 30 40
Crack Extension, mm
Fig. 7-R-Aa curves for mortar and composites with 8 percent carbon and 4 percent PP fibers.
The energy term due to inelastic deformations is a significant portion of the R-curve

A C 1 SP-155 95 = 0662949 0523200 703

--``,`,-`-`,,`,,`,`,,`---
Deflection
Fig. &Definition of cyclic energy and total energy dissipation during crack growth experiments.
The work of fracture is obtained by summation of incremental values

A C 1 SP-155 95 I0662947 0 5 2 3 2 0 1 6 4 T
Testing of FRC 113
0.025 3,
Mortar
0.020
+ Total
,* cyclic
E 0.015
E
t
c
e 0.010
0.005
0.0 o.1 0.2 0.3

CMOD, mm
ta)
0.60 1
I /
4% PP
* Total
Cyclic
E 0.40
I /
??
L
O
0.20-
1 #
-
0.00
0.00 0.25 0.50 0.75 1.00 1.25
CMOD, mm
--``,`,-`-`,,`,,`,`,,`---
(b)
Fig, 9-Work
Copyright American Concrete Institute of fracture as function of crack mouth opening for (a) mortar, and (b) 4 percent
PP fiber
No reproduction or networking reinforced
permitted composite
without license from IHS Not for Resale
A C 1 SP-155 95 Ob62949 0523202 5 ô b W
E
???1
350
--``,`,-`-`,,`,,`,`,,`---
z?
Mortar \
Fig. 10-Cornparison of total work of fracture and its cyclic component for various fiber
composites

A C 1 SP-155 95 IO b 6 2 9 4 9 0 5 2 3 2 0 3 q12 I
SP 155-6
Test Methods for Durability of Polymeric

Fibers in Concrete and UV Light Exposure
by P. Bolaguru and K. Slatturn
Svnousis: Two test methods are presented that can be used for evaluating
durability of polymeric fibers subjected to: alkaline environment present in
concrete, and UV light exposure. The test methods were used to evaluate three
polymeric fibers namely: nylon, polypropylene, and polyester. Durability of the
fibers in an alkaline environment was ascertained by measuring the flexural
toughness of fiber reinforced concrete specimens that had been aged in lime
saturated water maintained at 50°C. The UV light exposure test was conducted at
a temperature of 65°C with intermittent water spray. The wet spray was used to
simulate conditions in the field. Durability of the fibers was determined by
measuring the retained tensile strength of the fibers after light exposure and by
observing the surface characteristics of fibers under a microscope.
The test results indicate that nylon and polypropylene fibers are durable in
alkaline environment present in concrete. The nylon fibers, which were light
stabilized, were determined to be stable under UV light exposure. Polypropylene
fibers deteriorated under W light, and the deterioration of the polypropylene
single filament fibers was more rapid than for the fibrillated fibers. Hence these
fibers should not be used in applications where the fiber contribution is needed at
cracked-exposed sections.
Kevwords: Accelerated aging; cracking (fracturing); durability; esters;

fiber reinforced concretes; flexure; nylon fibers; polypropylene fibers;
scanning electron microscope; tensile strength; toughness; ultraviolet light
--``,`,-`-`,,`,,`,`,,`---

Not for Resale
A C 1 SP-155 9 5 IOb62949 0523204 359
116 Balaguru and Slattum
P. N. Balagum is professor of civil engineering at Rutgers in New Jersey. He

received his Ph.D. in structural engineering from University of Illinois at Chicago in
1977. He is a member of ACI, ASCE, ASTM, & TRB. He is widely published in
the field of concrete including a book published in 1992: Fiber Reinforced
Cement Composites.
K.L. Slattum obtained a B.S. in Chemistry from Hampden-Sydney in Virginia and

an MBA from the University of Richmond in Virginia. He works Alliedsignal
Fibers Division Technical Center in Virginia. He has 10 years laboratory
experience and is now a product specialist in the Industrial Fibers Group. He has
received Concrete Technologist certification from NRMCA
--``,`,-`-`,,`,,`,`,,`---
INTRODUCTION
Polymeric fibers have been used in concrete for the past 15 years. The
fibers include: nylon, polyester, polyethylene, polypropylene, and polyvinylalcohol
[1,2]. In most cases, these fibers are used at low volume fractions to reduce the
shrinkage cracking during the initiai and final setting period. For this type of
application, the fiber content is normally less than 1 kg/m3. Recently polymeric
fibers are being used at higher volume fractions for repairs, tunnels, and canal
linings. At these higher volume fractions, fiber contribution to ductility is utilized.
Even under low volume fractions, fiber contribution to ductility is considered as a
factor for choosing the fiber type.
If fiber contribution to ductility of hardened concrete is the reason for use

of polymeric fibers, then the long-term performance of these fibers should be
ascertained. The primary concerns associated with long-term performance are:
durability of fibers in concrete environment, durability of fibers that are exposed to
elements either due to cracking or partial deterioration of concrete, capability of the
composite to retain its original properties, resistance to freezing and thawing, and
frost resistance. This paper focuses on the durability of fibers in concrete
environment and the durability of fibers exposed to the elements. The need for
durability of fibers in concrete is self evident. First, if fibers deteriorate the
composite behavior will be like that of plain concrete. Second, the voids or
channels lefi by deteriorated fibers may affect the long-term durability of the
concrete
The fibers exposed to the elements should be durable in order to sustain the
composite action. For example, the fibers in a cracked overlay will be exposed to
light and rain. If these fibers degrade due to the exposure to light, then the
composite action will be lost. It could be argued that most fibers are protected by

A C 1 SP-155 9 5 I0 6 6 2 7 4 7 0 5 2 3 2 0 5 295
Testing of FRC 117
concrete. But the need for fiber contribution is critical at the cracked section to
reduce crack widths and reduce long-term crack opening. In fact, the fibers are not
needed in hardened concrete if it can be assured that concrete will not crack.
Hence, the fibers exposed at the cracked sections should be durable so that they
can transfer forces across the crack. “Although UV radiation makes up
approximately 5% of the total sunlight, it causes almost all of the damage to
durable materials” [3]. Some of the most common attacking elements are UV light
and moisture. UV light component is strong both in sunlight and artificial light.
In certain instances, exposed fibers on the surface of finished concrete are

left to degrade themselves in the light. But the degradation of the fibers could
continue below the surface creating voids or channels near the surface of the
concrete. This could lead to concrete durability problems because of easy
permeation of liquids and potential for freeze-thaw damage.
This paper presents the test methods and results for popular polymeric
fibers for both durability tests involving alkaline environment and W light. The
alkaline durability test was adopted from the method used for glass fibers, and the
UV test method was adopted from an existing ASTM test used for evaluating the
deterioration of geotextiles.
--``,`,-`-`,,`,,`,`,,`---
DURABILITY TEST FOR ALKALINE ENVIRONMENT
The popular test for evaluating the fibers in concrete is the accelerated
aging test in which fiber reinforced concrete samples are stored in lime-saturated
water maintained at elevated temperature levels. The lime-saturated water prevents
the leaching away of naturally occurring lime. The elevated temperature levels
varying from 50 to 80°C accelerate the aging process. For example, 1 day
immersion of the sample in lime-saturated water maintained at 50°C has been
shown to be equivalent to 101 days of natural weather exposure in the United
Kingdom with a mean annual temperature of 10.4OC [1,4,5]. Typically, test
specimens stored in water baths are tested either in tension or flexure at various
time intervals to determine strength and ductility. Normally the accelerated tests
are run up to 52 weeks.
The deterioration caused by accelerated aging is compared with

deterioration caused by natural aging process to obtain equivalency values. For
example, Fig. 1 shows the modulus of mpture values at various stages of aging for 3
accelerated aging 50, 60 and 80°C. When the accelerated aging curves at various
temperatures are horizontally displaced to the right, they tend to coincide with
natural weathering with good accuracy, Fig 2. Hence, it can be concluded that the
results obtained at various accelerated aging temperatures correlate well with
natural weathering conditions.

A C 1 SP-155 95 Ob62949 0523206 121 9
DURABILITY TEST FOR EXPOSED FIBERS
No standard method exists for evaluating the durability of fibers used in

concrete that are exposed to the elements. However, a number of standard test
methods exist for evaluating textile fibers. Two types of laboratory
weatherometers exist for testing materials: sunshine carbon arc & xenon arc. The
sunshine carbon arc is not preferred because unlike the sun, it emits significant
energy at less than 260 nm wavelengths which is well below the normal cut off of
295 nm for sunlight. This excessive amount of short wavelength radiation make
conditions in the carbon arc unrealistic compared to outdoor results. The xenon
arc more closely simulates the entire spectrum of sunlight [ 3 ] .
Table 1 presents some of the xenon arc methods that could be used for testing
fibers. The American Association of Textile Chemists and Colorists (AATCC) test
number 16E is intended for determining the colorfastness (fade-resistance) of
textile fibers to light. AATCC 16E is not a good test for concrete reinforcing fibers
because it lacks the shorter W part of the spectrum that is present in sunlight. The
Society of Automotive Engineers (SAE) JI885 is an accelerated exposure test for
evaluating automotive interior trim components which are often polymeric in
composition. While nylon, polyester, and polypropylene are often used in
automotive interiors, the SAE JI 885 test is too severe for concrete reinforcing
fibers as the test temperature is 89°C.
ASTM D 4355 [8] method, which is used to evaluate the deterioration of

geotextiles exposed to light and water, appears to be the best existing method for
testing concrete reinforcing fibers. The exposure conditions seen by geotextiles
and fiber-reinforced concrete are very similar. This test subjects the fibers to a
wavelength spectrum closely simulating that of sunlight. The irradiance level of
this test at 0.35 W/m2 “is more like winter sunlight,” while 0.55 W/m2 irradiance
level “compares well with summer” [ 3 ] . While the higher irradiance level would be
preferred, the lower level is often selected for convenience. The fiber samples are
exposed to water as might be expected in an exterior application. The test
temperature of 65OC is in the temperature range that concrete would reach during a
summer day. Finally this test was selected as the construction industry and fiber
producers are already familiar with the test procedure and interpretation of the
results.
In the ASTM D 4355 test method, the fibers are exposed to UV light for
specified time intervals then tested for percent breaking strength retained. The
fibers were also examined using Scanning Electron Microscopes (SEM) to
determine the extent of deterioration.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 I0662949 0 5 2 3 2 0 7 068 m
Testing of FRC 119
An experimental investigation was undertaken to study the durability of

nylon, polypropylene, and polyester fibers because these fibers are widely used in
concrete construction. The first phase of the investigation dealt with the durability
of fibers in the alkaline environment of concrete. Since polyester fibers did not fare
well in this test, they were omitted from the second phase dealing with durability
under UV light.
For the first phase dealing with the durability of fibers in concrete, fiber type
was the only independent variable. The concrete was proportioned to obtain a 28
day compressive strength of about 20 MPa. The durability was studied using
--``,`,-`-`,,`,,`,`,,`---
flexural test specimens. Both flexural strength and flexural toughness were
measured at various stages of accelerated aging.
For the second phase, nylon, fibrillated polypropylene, and single filament
polypropylene fibers were tested using ASTM D 4355 test method. Fiber strengths
and surface characteristics of fibers after various exposure periods were used to
evaluate the durability.
MATERIALS, MIX PROPORTIONS, AND SPECIMEN PREPARATION
Materials
The constituent materials for the beam specimens consisted of ASTM
Type I cement, natural sand, crushed stone, tap water, and air entraining
admixtures. The aggregates met the ASTM gradation requirements. Ail the three
types of fibers were 19 mm long. Nylon and polyester fibers were made of single
filaments whereas polypropylene fibers were fibrillated.
For the UV durability test, polypropylene fibers were tested both in

fibrillated and single filament form. Nylon fibers were in single filament form, All
of the fibers tested are commercially marketed for use in concrete.
Mix Proportions
The matrix composition for flexural specimens consisted of 307, 813,
1068, and 177 kg/m3 of cement, sand, coarse aggregate, and potable water
respectively. In addition, 500 and 709 mum3 of high range water reducing and air
entraining admixtures were used. A high fiber content of 4.75 kg/m3 was used in
order to obtain well-defined and repeatable load-deflection responses. The well-
defined load-deflection curves were needed in order to compute toughness indices
explained in a later section of this paper.

A C 1 SP-155 95 0 6 6 2 9 4 9 O523208 T T 4 =
--``,`,-`-`,,`,,`,`,,`---
Specimen Preparation
The primary test specimens for phase I study were 100 x 100 x 360 mm
prisms. These prisms were cast using Plexiglas molds and compacted using a table
vibrator. After casting, the specimens were left in their molds for 24 hours and
covered with polyethylene sheets. After 24 hours, they were demolded and cured
for 27 days in a moist room maintained at 100 percent relative humidity.
Companion 150 x 300 mm cylinders were made to obtain the 28 day compressive
strength. These cylinders were made using plastic molds and were compacted
using a vibrating table. The curing scheme was the same as the one used for
flexural specimens.
For phase II study, the fiber specimens were obtained from commercial
suppliers. Fibers of the same cross-sectional area were selected for physical testing
to insure valid physical test results. The faces of the clamps that grip the fibers
were resurfaced with very fine crocus cloth to insure fibers would not slip when
tested. Strain related properties of the fibers were not considered. Inaccuracies in
the measurement of short gauge lengths and the penetration of the forces into the
clamping region make determination of effective gauge length difficult. Therefore
only the breaking strengths of the fibers were determined. At least 10 specimens of
each type of fiber were tested for breaking strength. The remaining fibers were
evaluated by SEM. The samples were either 50 mm or 19 mm long.
TEST METHODS
The fresh concrete was tested for workability and air content. Workability
was measured using both the standard and inverted slump cone test, ASTM C143
and C995 respectively. The air content was measured using the pressure method,
ASTM C23 1. The 150 x 300 mm cylinders were tested in compression at 28 days
to obtain the compressive strength.
The accelerated aging test was conducted by storing the specimens in lime-
saturated water maintained at 50°C. Flexural strength and flexural toughness were
used to measure the contribution fibers provide to the concrete matrix. For each
group, twenty-four 100 x 100 x 360 mm prisms were made and cured for 28 days
before placing them in hot water bath.
Three samples were tested in flexure after O, 4, 8, 16, 32, and 52 weeks of
accelerated aging, The tests were conducted using ASTM C1018 procedure, in
which the beams were loaded at mid third points over a simply supported span of
300 mm. The deflection was measured at mid span. The load-deflection curves
were used to obtain the modulus of rupture and toughness index. Maximum load
and uncracked section properties were used for the computation of modulus of
rupture. Toughness index values Is, Ilo and 40were computed using ASTM C1018
I

A C 1 SP-155 95 0 6 6 2 9 4 9 0 5 2 3 2 0 9 930 =
Testing of FRC 121
definitions [8]. The ratios I& and 13&, were used to estimate the ability of the
fiber reinforced concrete specimens to sustain loads at large deflection.
ASTM D4355 method was used to test the fibers. This method subjects the
fibers to a light spectrum with an intensity that matches winter sunlight. The
exposure cycle consisted of 90 min. light only and 30 min. light plus water spray.
The intensity of light was controlled by monitoring the irradiance level at a
specified wavelength, Table 1. Approximately 20 specimens of each fiber type
were exposed in one xenon arc weatherometer at the same time to insure equal
weathering conditions. Tensile strength tests were conducted after O, 150, 300,
and 500 hours of exposure and the percent tensile strength was determined.
Scanning Electron Microscopic (SEM) photos were also taken at magnificationsof
400X and 1200X. For smaller diameter short fibers, only photos were taken
because tension tests could not be conducted.
TEST RESULTS AND DISCUSSION
Accelerated Aging Test

When fibers were added, the standard slump decreased from 180 mm to
about 5 mm. However, the inverted slump cone times were only 5, 3, and 11
seconds for nylon, polypropylene, and polyester fibers respectively, indicating
excellent workability under vibration. The addition of fibers increased the air
content slightly. The 28 day compressive strengths were approximately 20 m a .
Fibers at 4.75 kg/m3reduced the compressive strength by about 10 percent.
The load deflection curves obtained at various stages of aging are shown in
Fig. 3, 4, and 5 for nylon, polypropylene, and polyester fibers respectively. The
toughness indices are presented in Table 2. The modulus of rupture and the ratios
of 1 3 0 / I ~ ,are presented in Fig. 6 and 7 respectively. A careful study of Table 2 and
the figures lead to the following observations:
'Polyester fibers provide a higher modulus of rupture before aging. This

could be due to the higher modulus of elasticity of polyester fibers as compared to
the other two fibers.
'The post-peak resistance decreases with aging for polyester fibers. For the
nylon and polypropylene fibers, the post-peak performance improves in certain
instances.
'The toughness index IS is about the same for all three fiber types and
different aging periods. This is in agreement with earlier results that indicate that Is
is not sensitive enough to distinguish fiber contributions [i].
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-I155 95 0662949 0 5 2 3 2 3 0 6 5 2
'Nylon and polypropylene fibers had higher 110and 130 values than polyester
fibers. Decreases in the 130 values for polyester fibers were observed with aging.
This difference can be observed by noting the variation in i3O/I10 values, Fig. 7.
'Visual examination of the aged specimens also showed considerable

disintegration of polyester fibers.
W Light Exposure Test

The performance of nylon and two fibrillated polypropylene fibers
commonly used in concrete is presented in Fig. 8. This figure shows the percentage
strength retained after O, 150, 300, and 500 hours of exposure. This figure shows
after 500 hours exposure that one of the type of polypropylene fibers totally
disintegrated while the other type retained 63 percent of the original strength. The
difference in performance could be attributed to polymer composition. Nylon fibers
retained 95 percent of their strength.
The literature points out that "solar irradiation of PP and PE leads to the
embrittlement of the polymers and the formation of oxidation products" [IO]. For
polyamides photooxidation is initiated by impurities in the polymer.
SEM photos for the various fibers are shown in Fig. 9 to 15. For the
polypropylene fibrillated fibers both 400X and 1200X magnification are shown
because these fibers are more coarse and it is difficult to see the entire fiber under
1200X magnification. Fig. 9 to 15 confirm the disintegration process for the
polypropylene fibers and the results shown in Fig. 8. All polypropylene fibers
become brittle after exposure to W light. The fibers with smaller diameters
disintegrate more rapidly due to increased exposed surface area as shown in Fig.
-
15.
CONCLUSIONS
Based on the results presented in this paper and observations made during
the experimental investigation, the following conclusions can be drawn.
'At a fiber content of 4.75 kg/m3, all three fiber types (nylon, polypropylene
and polyester) provide post-peak resistance.
'Nylon and polypropylene fibers are durable in the alkaline concrete

environment. Post-peak resistance decreases for polyester fibers after aging.
--``,`,-`-`,,`,,`,`,,`---
'If the structure is totally protected from UV light, then both nylon and
polypropylene fibers are expected to be durable. But if the fibers are exposed to

Testing of FRC 123
UV light exposure, polypropylene fibers will start to deteriorate. Since the

exposure will occur in critical sections such as maximum stress locations due to
cracking, the consequence could be a total loss of post-peak resistance. Once the
fibers in the cracked location disintegrate, the composite action of that particular
section will be that of plain concrete. Hence, if polypropylene fibers are used to
provide post-crack resistance, they must be protected from UV light. It might be
possible to surface coat the fibers to provide such protection.
REFERENCES
1. Balaguru, P., and Shah, S. P., Fiber Reinforced Cement Composites, McGraw-
Hill, 1992, 535 pages.
2. Bentur, A., and Mndess, S., Fiber Reinforced Cementitious Composites,

Elsevier Applied Science, London, 1990.
3. “Chart on Sunlight, Ultraviolet Light & Accelerated Weathering”, Q-Panel,

Clevland, Ohio.
4. Litherland, K. L., Oakley, D. R., and Proctor, B. A., “The Use of Accelerated
--``,`,-`-`,,`,,`,`,,`---
Aging Procedures to Predict Long-term Strength of GFRC Composites”,
Journal of Cement and Concrete Research, Vol. 11, 1981, pp. 455-466.
5. Daniel, J. I., “Glass Fiber Reinforced Concrete”, Fiber Reinforced Concrete,

Report No. 2493D and 26 14D, construction Technology Laboratories, Inc.,
Skokie, Illinois, 1988, pp. 5.1-5.30.
6. AATCC - Technical Manual, Vol. 68, American Association of Textile

Chemists and Colorists, 1993.
7. Society of Automotive Engineers, S A E 51885, “Accelerated Exposure of

Automotive Interior Trim Components Using a Controlled Irradiance Water
Cooled Xenon-Arc Apparatus”, Materials, Vol. 1, Section I l ,217, SAE Inc.,
1992.
8. ASTh4, Annual Book of ASTM Standards, “Standard Test Methods for

Deterioration of Geotextiles from Exposure to UV Light and Water - Zenon-
Arc Apparatus”, 1992, Race St., Philadelphia.
9. ASTM, Annual Book of ASTM Standards, Concrete and Mineral Aggregates,

Sec. 4, Vol. 04 02, 1993, Race St., Philadelphia.
10. Encyclopedia of Polymer Science & Engineering, 2nd edition, Volume 4, 1986.

A C 1 SP-155 95 m 0662949 0523232 425 m
--``,`,-`-`,,`,,`,`,,`---
Test Method & Imdiince at Temp & RH

Description specified wavelength eC) (W) Cycle
(w/m2 @I um)
SAE J1885 0.55@ 340 89"@ 50% RH 60 min iight
Automotive Interior 3 8 O @ 95% RH 30 min dark
AATCC 16 E 1.1 a 4 2 0 63"@ 30% RH Continuous iight
Colorfastnessto Light
ASTM D4355-92 0.35 @ 340 65' @ 3û??RH 90 min light
"Deterioration of 30 min light +
Geotextiles..." water spray
TABLE 2 - TOUGHNESS INDICES AT VARIOUS STAGES OF ACCELERATED AGING
Age Weeks
OPP - polypropylene
+PY - polyester

A C 1 SP-155 75 I0662949 0523213 361 W
Testing of FRC 125
Fig. 1-Modulus of rupture versus accelerated aging time (3)
4c - U.K.weathering
ni e Accelerated nging at
i O
2:
Wl
- 30- % *
O
A
50°C
60°C
a ** 80°C
3r $*a
D
$ 20-
0 4
h
O
0& B WO$*O,o
g
CI
10-
s
Fig. 2-Comparison of accelerated and natural aging (4)
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 m 0662949 0 5 2 3 2 3 4 2 T B
12
I :h: ACCELERATED AGE:
z
2¿
2
O 4
I4
I I I I
O 1 2 3 4
DEFLECTION, m m
Fig. &Load deflection behavior of beams subiected to accelerated aging: nylon fibers
12 - ACCELERATED AGE:16 WEEKS
Y /
- - - - - -b
I
o 1 2 3 4
DEFLECTION, m m
Fig. +Load deflection behavior of beams subieaed to accelerated aging: polypropylene fibers
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0662949 0523235 134 I
Testing of FRC 127
CELERATED AGE :O WEEKS

--``,`,-`-`,,`,,`,`,,`---
I'
4
Fig. 5-load deflection behavior of beams subjected to accelerated aging: polyester fibers
6
r 52
ACCELERATED AGE, UEEKS:52
32
o
H 4
w
e:
E
0.0
2 3
cri
O
VI
P
c
l
302
O
E
1
NYLON
Fig. LModulus of rupture of aged specimens

A C 1 SP-155 75 0662749 O523236 O70 m
Fig. 7-Ratio of @,,, for aged specimens
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 9 5 I0 6 6 2 9 4 9 0 5 2 3 2 3 7 T O 7 m
Testing of FRC 129
--``,`,-`-`,,`,,`,`,,`---
NYLON POLYPROPYLENE POLYPROPYLENE

TYPE 1 TYPE 2
Fig. "Tensile strength versus UV light exposure time

A C 1 SP-155 95 = 0662949 0523238 943 =
O hrs exposure (1 200X Magnification)
500 hrs exposure (1 200X Magnification)
--``,`,-`-`,,`,,`,`,,`---
Fig. 9-SEM photograph, nylon fibers, 120h magnification

A C 1 SP-155 9 5 IO662747 0 5 2 3 2 3 9 B B T
Testing of FRC 131
O hrs exposure (400X Magnification)
500 hrs exposure (400X Magnification)

--``,`,-`-`,,`,,`,`,,`---
Fig. IO-SEM photograph, polypropylene fiber type 1, 400x magnification

A C 1 SP-I155 75 I0662747 0 5 2 3 2 2 0 5 T L
0 hrs exposure ( 1200X Magnification)
--``,`,-`-`,,`,,`,`,,`---
Fig. 11-SEM photograph, polypropylene fiber type 1, 120ûx magnification

A C 1 SP-155 9 5 m 0 6 6 2 7 4 9 0 5 2 3 2 2 3 438 D
Testing of FRC 133
--``,`,-`-`,,`,,`,`,,`---
Fig. 12-SEM photograph, polypropylene fiber type 2, 400x magnification

A C 1 SP-I155 95 0662949 0523222 374

--``,`,-`-`,,`,,`,`,,`---
Fig. 13-SEM photograph, polypropylene fiber type 2, 12OOx magnification

A C 1 SP-1-55 95 0 6 6 2 9 4 9 0523223 200 M
Testing of FRC 135
Fig. 14-SEM photograph, polypropylene fiber iype 3, 12OOx magnification

--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0 6 6 2 9 4 9 0 5 2 3 2 2 4 147 W
O hrs exposure ( 1200X Magnification)
Fig. I L S E M photograph, polypropylene fiber, single filament, 1200x magnification.

Note: Sample had totally disintegrated at 500 hrs; therefore, photomicrograph at 400 hrs wos
--``,`,-`-`,,`,,`,`,,`---
taken
Restrained Shrinkage Tests on Fiber
Reinforced Cementitious Composites
by N. Banthia, M. Azzabi, and M. Pigeon
Svnopsis: The usehlness of fiber reinforcement in improving the cracking

resistance of cement-based materials under restrained shrinkage conditions is
indisputable. In fact, in many instances, this may be the sole reason of adding fibers
to concrete. In spite of this general recognition, there is no universally accepted
technique of demonstrating or quantifying the effectiveness of fibers under the
conditions of restrained shrinkage. This paper describes a newly developed technique
where prismatic specimens with a linear restraint along the longitudinal axis are
subjected to a drying environment such that conditions of uni-axial tension are
generated. The specimen cracks under these conditions and if fiber reinforcement is
present, the influence of fibers on the cracking pattern can be established. Results
with seven types of fibers are presented. Based on the observations of the crack
patterns, a "fiber efficiency factor" is proposed which appears to be an appropriate
basis for characterizing the fibers.
Kevwords: Carbon; composite materials; cracking (fracturing); £iber

reinforced concretes; fibers; polypropylene fibers; restraints; shrinkage;
steels
--``,`,-`-`,,`,,`,`,,`---

Not for Resale
A C 1 SP-155 75 II 0662749 0 5 2 3 2 2 6 T I T H
138 Banthia, Azzabi, and Pigeon
--``,`,-`-`,,`,,`,`,,`---
AC1 member Nemkumar Banthia is an associate professor of civil engineering

at the University of British Columbia, Vancouver, Canada.
Maha Azzabi is a doctoral student in the Dept. of Civil Engineering, Laval

University, Quebec, Canada.
AC1 fellow Michel Pigeon is a professor of civil engineering at Laval

University, Quebec, Canada.
INTRODUCTION
In the plastic state, cement paste undergoes a volumetric contraction as high

as 1% of the absolute volume of dry cement. If restrained, this contraction can cause
tensile strains far in excess of those needed to cause cracking in young pastes with
poorly developed strengths. To avoid cracking in the plastic state, it is crucial that
the materials be not subjected to an environment where shrinkage induced strain may
exceed the strain capacity of the material. In spite of every effort, plastic shrinkage
induced cracking still remains a real concern, particularly in large surface area
placements like slabs on grade, thin surface repairs, etc.
The possibility of cracking in cement-based materials subjected to a drying

environment depends, among other things, upon the severity of the environment, the
extent of imposed restraint, length of exposure, cement content of the mix, age at
first exposure, the area over which drying occurs and the presence or absence of
reinforcing elements. In this regard, reinforcement of cement-based materials with
short, randomly distributed fibers is found to be very effective (i, 2, 3). The general
observation has been that instead of a sigle wide crack which is usually formed in
plain cement-based materials under restrained shrinkage, multiple cracking occurs in
fiber reinforced composites with each of the several cracks substantially reduced in
width. With multiple cracking, substantial amounts of energy are consumed with a
corresponding increase in the strain capacity of the material. Such a behavior is
desirable both fiom mechanical and durability considerations.
Fibers used for reinforcing cementitious materials may, in general, be divided

in two broad categories: macro and micro. Macro-fibers are the conventional large
fibers of steel with lengths between 25 and 60mm and transverse dimensions
between 0.3 and 3 mm. These fibers are most commoniy used in cast-in-place
concrete and in shotcrete at nominal volume fiactions (about 1%). Beyond such
volume fiactions, mixing and handling become difñcult and extensive fiber baiiing
can occur. At such low volume fiactions, however, oniy insignificant increases in
the compressive or the tensile strengths may be expected and the real advantage of
A C 1 SP-155 95 O662747 0 5 2 3 2 2 7 9 5 b
Testing of FRC 139
adding nominal volume fiactions of macro-fibers is in the improved toughness after

matrix cracking (4,5,6) when fibers bridge cracks and undergo energy intensive pull-
out processes. Micro-fibers, on the other hand, are very fine fibers with their lengths
less than 20 mm and their transverse dimension less than 25 mm. Comparatively,
while macro-fibers have their specific surface areas less than 10cm2/g, specific
surface areas for micro-fibers are in excess of 200 cm’/g. Micro-fibers can be added
to cementitious mixes at very high volume fractions and depending upon fiber
characteristics, notable improvements in both strength and toughness may be
expected (7). At high volume fiactions, it is believed that these fibers provide
reinforcing mechanisms at the micro-level, stabiie and promote distributed
microcracking and alter the intrinsic behavior of the brittle cement matrix itseif(8,9).
Given the distinctly different reinforcing mechanisms of macro and micro-fibers, it
was undertaken to investigate the relative effectiveness of these fibers under
restrained shrinkage conditions.
A standard method of conducting restrained shrinkage tests does not exist.

The various test techniques developed so far include: (a) The Ring Test (1, 10, 11,
12, 13J, (b) The Doubly Restrained Plate Specimens (14, 15), and (c) Uni-axial
Tests (2,3).
It is generally recognized that uniaxial tensile tests are desirable fiom the
point of view of obtaining fimdamental material information and also in order to
understand the mechanisms of fiber reinforcement when under drying shrinkage
conditions. Based on this recognition, it was undertaken to develop a rational test
technique capable of subjecting cement-based fiber reinforced composites
EXPERIMENTAL
Restrained Shnnkacte Test Set-UD
The test Set-up designed to conduct restraint shrinkage tests is shown in

Figure 1. It consists of a specimen mould (40 mm x 40 mm x 500 mm) with two
triple-bar anchors at its ends. The other ends of the anchors are rigidly connected to
vertical posts which are, in turn, securely attached to the 50 mrn thick base plate.
The mold itself is mounted on two frictionless rollers that are ftee to slide dong two
horizontal slide rails spanning the two vertical posts. A travehg microscope (5Ox)
capable of moving both in the longitudinal and the transverse directions with respect
to the specimen is mounted on a third slide rail for crack observations. The
microscope is equipped with a vernier such that crack width and length
measurements to the nearest 0.01 mm can be carried out. The entire assembly is
mounted on a trolley which can be wheeled in and out of the d y n g chamber.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0662747 0 5 2 3 2 2 8 8 9 2
In an actual test, the freshly mixed cement mixture was poured in the mould
and the whole assembly was transferred to the drying chamber. Later, when
possible, the specimen was demoulded and further drying occurred in the chamber.
In this state, the only restraint in the specimen came fiom the end-anchors in the
longitudinal direction. AU crack observations and measurements were canied out in
the drying chamber itself on the most severely cracked surface for up to twenty-four
hours.
Test Procram
The equipment described above was used for studying cracking in fiber
reinforced cement paste and mortar under one-dimensionai restraint when subjected
to a severe dtying environment. M e r several unsuccessful attempts at inducing
cracking in milder environments, a particularly severe environment with a
temperature of 50°C and a relative humidity of less than 50% was chosen. Seven
fiber types belonging to two major categories macro(1arge) and micro(fine) were
investigated. The details of these fibers are given in Table 1 where .the extremely fine
s i e of micro-fibers as compared to the macro-fibers can be noted. At equal volume
fractions, one could expect two to three orders of magnitude more micro-fibers in
--``,`,-`-`,,`,,`,`,,`---
the composite than macro-fibers. Consequently, it was anticipated that the micro-
fibers would provide reinforcing mechanisms distinctly different from those of the
macro-fibers.
Two matrices (cement paste and cement mortar) were investigated, although
only the results obtained with the mortar matrix are presented in this paper. Results
with the paste matrix may be found elsewhere (16). The mortar matrix proportions
were (cement:water:silica-fÙme:sand) 1.0:0.4:0.2:2.0,and appropriate quantities of
superplasticizer were added to obtain a satisfactory workability. The use of silica
fume was considered essential for an effèctive dispersion of high specific surface area
micro-fibers (17) and for a better fiber-matrix bond. A routine mortar mixer was
used. The following fiber volume fractions were investigated (see Table 1 for
notation):
Fiber F 1: 0.5,0.75 1.0 and 2%

Fiber F2 and Fiber F3 : 0.5, 0.75 and 1.0%
Fiber C 1: 0.25, 0.50, 0.75, 1.0%
Fiber C2: 0.15, 0.25, 0.50%
Fiber P: 0.25, 0.50, 1.0, 1.25%
Fiber S : 1.0, 2.0,3.0,3.5%
Crack obseivationswere continued for up to about 24 hours after casting. In

each test, the following were noted:

A C 1 SP-155 95 = 0662947 0 5 2 3 2 2 9 727
Testing of FRC 141
- total number of cracks, n
- maximum observed crack width, w,,

n
- total cumulative crack width, w,,definedas w, = wi

i=¡
n
- total cumulative crack length, L,, deñned as L, = 1 Li

i=l
Data Analysis
When a prismatic element with a longitudinal restraint cracks under tensile

stresses generated as a result of drying, the presence of fibers is expected to influence
not only the lengths of the resulting cracks but also their widths. When no fibers are
present, the element would crack under minimal tensile strains and once the ñrst
crack is created, the element will essentially lose the effect of end-restraints.With the
element free to shrink now, the aiready created crack could only get wider and no
further cracking is possible elsewhere in the specimen. Iffibers are present, however,
the matrix would stiü crack at about the same strains as when unreinforced but the
cracks would be refrained fiom growing wider if sufficient fiber bridging action can
take place. If the fibers are long enough and if fiber breakage does not occur at the
crack, one may expect the stressesto be transferred back to the matrix over a certain
--``,`,-`-`,,`,,`,`,,`---
transfer length and with sufficient buiid-up of stresses in the matrix, the matrix will
develop another crack parallel to the first one (18). This multiple cracking can go on
until the entire element develops cracks which are approximately equidistant and run
perpendicular the direction of tensile stresses.Multiple cracking can occur, however,
only if the fibers at a section can cany the stresses higher than that carried by the
matrix at cracking i.e., if the critical fiber volume fiaction is exceeded, if the fibers
are long enough and also if fiber breakage across the crack does not occur.
Based on the above discussion, it is clear that fibers influence the restrained
shrinkage behavior in two ways: first, they distribute cracking more evenly over the
entire length resulting in closely spaced cracks, and second, they reduce crack widths
through effective crack bridging. It appears logical, therefore, that any attempt at
quantifj4ng fiber effectiveness must consider both these mechanisms. In this study, a
non-dimensional parameter called thefiber eficiency factor was adopted in order to
quantimng the effectiveness of a given fiber. It was defined as,
fiber eficiency factor = L, /w (1)
where, as defined before L, is the cumulative crack length and w, is the cumulative
crack width over a given surface of observation. Clearly, an unreidorced matrix that
usuaiiy failed with a single wide crack, had a low eficiency factor.
A C I sp-155 95 0 6 6 2 9 4 9 0 5 2 3 2 3 0 440
RESULTS
General Observations
Specimens without the fibers, as expected, cracked with one wide crack
spanning the entire width of the specimen. In the case of the macro-fiber reinforced
composites, however, the specimens developed a number of closely spaced cracks
and each crack stayed substantially narrower than the sigle wide crack observed in
the unreinforced specimens. In the micro-fiber reinforced composites, on the other
hand, the specimens usually cracked with one crack in the middle of the specimen,
and this crack stayed very narrow. In Figures 2 and 3, the number of cracks, n, as a
function of fiber volume fraction for the macro and micro-fiber composites,
respectively, are plotted.
Effectiveness of Macro-Fibers
Detailed results are given in Table 2. The maximum observed crack widths,
,w are plotted as a íùnction of fiber volume fraction in Figure 4 for the macro-fiber
composites. Note a consistent decrease in w,, with an increase in the fiber volume
fiaction. W e specimens without fibers always developed one wide crack in the
middle, fiber reinforced specimens developed several well distributed cracks much
narrower in widths; at large fiber volume fiactions, as many as 15 cracks were
observed. Even with those many cracks, the cumulative crack widths, w,,for the
fiber reinforced specimens stayed considerably smaller than the width of a Sigle
--``,`,-`-`,,`,,`,`,,`---
,I
crack in the unreinforced specimen. Even more interestingly, as the number of cracks
I
grew larger with the fiber volume fiaction, the cumulative crack width, w,,in fact,
decreased.
I In Figure 5, thejìber eflciency factors (Eqn. 1) are plotted as a function of
fiber volume fraction. It may be noticed that a fiber deformed along the entire length
(Fiber F3) is more effective than those deformed oniy at the ends (Fibers FI and F2).
Also, if the geometry is the same, a longer fiber (Fiber F2) is more effective than a
shorter fiber (Fiber F1) in reducing the crack widths, in causing more multiple
cracking and in the overall reinforcing efficiency.

A C 1 SP-L55 7 5 E 0 6 6 2 9 4 9 0523233 3 8 7 E
Testing of FRC 143
Effectiveness of Micro-Fibers
Detailed results for micro-fiber reinforced composites are given in Table 3 .

The lack of multiple cracking in the case of micro-fiber reinforced composite was
discussed previously with referenceto Figure 3. In Figure 6, the maximum observed
crack widths, IV-, are plotted as a function of fiber volume fraction. The fiber
efJiciencyfactors (Eqn. 1) are shown in Figure 7. The general effectiveness of micro-
fibers, similar to that of macro-fibers (Figure 4), in reducing the crack widths may be
noticed from Figure 6. Although the composites with micro-fibers underwent less
multiple cracking (lower LE),the cumulative crack width (wc) values were also
smaller. The combined effect of these two phenomena was that thejiber efJiciency
factors (Figure 7) for the micro-fibers were in the same range as the macro-fibers.
This supports the premise behind thejber eficzeency factor concept.
Comparatively, carbon fibers with a moderate modulus were far superior in

reducing the crack widths, in uniformly distributing the cracks and in the overaii
efficiency. High modulus steel micro-fibers, on the other hand, were the least
effective of all. It is probable that this is not related to the modulus but to the
strength of the fiber-matrix interfacialbond. Possibly, as a result of their coarser s i e
and very uneven surface, steel fibers developed a very strong interfacial bond with
the surrounding matrix and hence fractured across a crack. While polypropylene and
carbon fibers both had relatively smoother surfacetextures, and in spite of the much
finer s i e of polypropylene fiber compared to carbon, the very low modulus of
polypropylene led to a relatively inferior behavior. Finally, among the two lengths of
carbon fiber tested, the longer 10 mm fiber appears to be far more effective than the
shorter 3 mm fiber.
Based on the observations here, it is difficult to draw definite conclusions

about the influence of a particular micro-fiber property on the restrained shrinkage
behavior of its composite. Clearly, fiber diameter, elastic modulus, tensile strength,
surface characteristics,interfacialbond strength, aspect ratio, percent elongation, etc.
are some of the relevant characteristics. Given the interdependent nature of these
variables and our limited understanding of the related mechanisms, a precise
interpretation of the test data is not possible at this stage. Some analyticaltreatment
--``,`,-`-`,,`,,`,`,,`---
ofthe data obtained here is presented elsewhere (16).
CONCLUDING REMARKS
The paper describes a simple test technique developed to conduct restrained

shrinkage tests on fiber reinforced cement-based composites. Tests were conducted
on macro and micro-fiber reinforced composites and both were found to have
significantly improved resistance to cracking in a drying environment. These fibers,

A C 1 SP-155 95 = Obb2ïYï 0523232 213
however, depicted very different cracking characteristicsas a result of their dissimilar

reinforcing mechanisms.
Afiber efJiciency factor,dehed as the ratio of cumulative crack length to

cumulative crack width, is proposed which appears to be an appropriate and
justifiable basis of characterizing and comparing the performance of different fibers
which otherwise may have distinctly different reinforcing mechanisms.
ACKNOWLEDGEMENTS
The authors wish to thank the Natural Sciences and Engineering Research
Council of Canada (NSERC) for continued financial support.
REFERENCES
1. Gnybowski, M. and Shah, S.P., Shrinkage Cracking of Fiber Reinforced

Concrete, AC1 MaterialsJournal, 87(2), 1990, pp. 138-148.
2. Ong, K.C.G. and Paramsivam, P., Cracking of Steel Fiber Reinforced

Mortar due to Restrained Shrinkage, in Fiber Reinforced Cements and
Concretes - Recent Developments (Eds. R.N. Swamy and B. Barr), Elsevier
Applied Science, London andNew York, 1989,pp. 179-187.
3. Paillere, A.M., Buil, M. and Serrano, J.J., Effect of Fiber Addition on the
Autogeneous Shrinkage of Silica Fume Concrete, AC1 Materials Journal,
86(2), March-April 1989, pp. 139-144.
4. Bentur, A. and Mindess, S., Fiber Reinforced Cementitious Composites,

Elsevier Applied Science, London, 1990.
5. Balagmu, P.N. and Shah, S.P., Fiber Reinforced Cement Composites,

McGraw-HU, Inc., New York, 1992.
6. Banthia, N. and Troîtier, J.-F., Concrete Reinforced with Deformed Steel

Fibers, Part 2: ToughnessCharacterization,AC1 Mat. Journal (in press).
I 7. Banthia, N. and Sheng, J., Micro-Reinforced Cementitious Materiais,

Materials Research Society Fall Meeting Proc., Vol. 21 1 (Eds. S. Mindess
and J. Skainy), Boston, Nov. 1990, pp. 25-32.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 9 5 111 0 6 6 2 9 4 9 0 5 2 3 2 3 3 L 5 T =
Testing of FRC 145
8. ûuyang, C. and Shah, S.P., Toughening of High Strength Cement Matrix

Reinforced by Discontinuous Short Fibers, Cement and Concrete Research,
1992 (to appear).
9. Shah, S.P. and Ouyang, C., Mechanical Behavior of Fiber Reinforced

Cement-Based Composites, Journal of American Ceramic Society, 74(1i).
1991, pp. 2727-38,2947-53.
10. Krenchel, H. and Shah, S.P., Restrained Shrinkage Tests with PP-Fiber
Reinforced Concrete, Fiber Reinforced Concrete Properties and
Applications, SP-105, American Concrete Institute, Detroit, 1987, pp. 141-
158.
11. Malmberg, B. and Skarendahí, A., Method of Studying the Cracking of

--``,`,-`-`,,`,,`,`,,`---
Fiber Reinforced Concrete under Restrained Shrinkage, Testing and Test

Methods of Fiber Cement Composites, RILEM Symposium Proceedings,
The ConstructionPress, Lancaster, 1978, pp. 173-179.
12. Dalk P.A., Influence of Fiber Reinforcement on Plastic shrinkage and

Cracking, in Brittle Matrix Composites-I (Eds. Brandt and Marshall),
Elsevier Applied Science, London, 1986, pp. 435-441.
13. Kovler, K., Sikular, J. and Bentur, A, Restrained Shrinkage Tests of Fiber
Reinforced Concrete Ring Specimens: Effect of Core Thermal Expansion,
Materials and Structures,RILEM, 26, 1993, pp. 23 1-237.
14. Khajuria, A and Balagum, P. Plastic Shrinkage Characteristics of Fiber

Reinforced Cement Composites, in Fiber Reinforced Cement and Concrete
(Ed. R.N. Swamy), Proceedings of the Fourth REEM International
Symposium, E & FN Spon, 1992, pp. 82-90
15. Kraai, P.P.,A Proposed Test to Determine the Cracking Potential due to
Drying Shrinkage of Concrete, Concrete Construction, 1985, pp. 775, 778.
16. Banthia, N., Maha, A. and Pigeon, M., Restrained Shrinkage Cracking in
Fiber Reinforced Cementitious Composites, Materials and Structures
WEM), 26, 1993, pp. 405-413.
17. Banthia, N., Pitch-Based Carbon Fiber Reinforced Cements: Structure,

Performance, Applications and Research Needs, The Canadian Journal of
Civil Engineering, 19(1), 1992, pp. 26-38.
18. Avenston, A., Cooper, G.A. and Kelly, A., Single and Multiple Fracture, in
The Properties of Fiber Composites, Proc. Conference of National Physical
Laboratories, IPC, Science and Technology Press, LJK, 1971, pp, 15-24.

A C 1 SP-155 95 0 6 6 2 9 4 9 0523234 096
8
e4
m
L
o
n
.-
L
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-L.55 9 5 I0 6 6 2 9 4 9 0 5 2 3 2 3 5 T22
--``,`,-`-`,,`,,`,`,,`--- Testing of FRC 147
TABLE 2 - RESULTS OF RESTRAINED SHRINKAGE TESTS ON MACRO-FIBER REINFORCED

CEMENT COMPOSITES
I Mortar I
~
0.00 2.85 2.85 40.0 14.03 1

O. 50 1.28 2.58 95.0 36.82 4
F1 0.75 1.05 1.83 108.0 59.01 5
1.00 0.87 1.96 141.0 71.73 6
2.00 O 40 130 119.0 91.53 13
0.50 1.50 1.93 120.0 64.10 4
F2 0.75 0.73 2.10 225.0 107.14 8
1.o0 0.27 1.17 191.0 163.20 11
0.50 0.85 1.93 115.0 59.58 4
F3 0.75 0.40 1.33 195.0 146.6 8
1.00 0.18 0.48 90.o 187.5 5
1
Maximum Observed Crack Width in mm
Total Cumulative Crack Width in mm
3
Total Cumulative Crack Length in mm
* Number of Cracks
LJw,: Fiber EfficiencyFactor

A C 1 SP-155 9 5 0662749 0523236 9 6 9
TABLE 3 - RESULTS OF RESTRAINED SHRINKAGE TESTS ON MACRO-FIBER REINFORCED

CEMENT COMPOSITES
I Mortar I
Fibre Vf, % W1- d L3c LJWC n*
0.00 2.85 2.85 40.0 14.03 1
- 0.25 1.90 1.90 40.0 21.09 1

c1 0.50 1.15 1.15 40.0 34.78 1
0.75 0.58 0.68 52.0 76.47 1
1.o0 o. 12 0.12 26.0 216.6 1
0.15 0.88 0.88 40.0 45.45 1
c2 0.25 0.36 0.36 43.0 119.44 1
0.50 110 cracks
I 1.00 I 2.75 I 2.75 I 40.0 I 14.54 I 1 I
S I 2.00 I 1.67 I 1.67 I 40.0 I 23.95 I 1 I
3.00 0.39 0.39 40.0 97.43 1
3.50 0.20 0.20 21.0 105.0 1
0.25 2.10 2.10 40.0 19.05 1
0.50 1.40 1.40 45.0 31.03 1
P 0.75 0.85 0.95 50.0 52.60 2
1.o0 0.50 0.70 52.5 75.00 3
1.25 O. 16 0.22 39.5 179.54 3
MaWnum Observed Crack Width in mm

1
Total Cumulative Crack Width in mm
Total Cumulative Crack Length in mm
* Number of Cracks
LJw,: Fiber Efficiency Factor
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 I0 6 6 2 9 4 9 0 5 2 3 2 3 7 ô T 5 I
Testing of FRC 149
40x40 x500
i
In
i
ímm)
Fig. 1-Restrained shrinkage test set-up
Mortar (Macro-Fibers) 0 F1
' F2
+ F3
--``,`,-`-`,,`,,`,`,,`---
0.0 0.5 1 .o 1.5 2.0 2.5

fiber Volume Fraction, Vf, %
Fig. 2-Number of shrinkage induced cracks, n, in macro-fiber reinforced mortars. In these

composites extensive multiple crocking occurred as indicated by large number of cracks formed

A C 1 SP-155 95 = Ob62949 O523238 731
Maar (Micro-Fibers) 0 c1
--``,`,-`-`,,`,,`,`,,`---
c2
+ s
" P
T T +
O I I I
O 1 2 3 4
Fiber Volume Fraction, V€,%
Fig. &Number of shrinkage induced cracks, n, in micro-fiber reinforced mortars. Notice that
composites usually cracked with a single crack without any multiple cracking
g 3.0
Mortar (M acro-Fibers)
s 0 F1
8z F2
24
2.0; + F3
j2g3 1.0-
O
ti I
0.0 I I I
0.0 0.5 1 .o 1.5 2.0

Fiber Volume Fraction, Vf, %
Fig. 4-Maximum crack widths (w,,,,,~)observed in cement composites reinforced with macro-
fibers under restrained shrinkage conditions
Mortar(MaaeFibers) o
cl I
F2
+ F3
Q
iz 0.0 I I l I
0.0 0.5 1 .o 1.5 2.0 2.5

Fig. &Fiber efficiency factors (LJwJ plotted for various macro-fibers (see Table 1 for fiber
details)
A C 1 SP-15.5 9 5 I0662947 0523237 678
Testing of FRC 151
3.0
Mortar (Micro-Fibers) o c1
c2
2.0
i'I \\ + s
1.o
0.0 I I I
0.0 1 .o 2.0 3.0 4.0

Fiber Volume Fraction, Vi, %
Fig. ó-Maximum crack widths (w ) observed in cement composites reinforced with micro-
fibers under restrained shrinkage comndxitions
--``,`,-`-`,,`,,`,`,,`---
Mortar (Micro-Fibers) 0 c1
3s
I=-
+ s
c2
L 2oo.ol M P
0.0 1 .o 2.0 3.0 4.0

Fig. 7-Fiber efficiency factors (LJwJ plotted as function of fiber volume fraction for various
micro-fibers

A C 1 SP-155 9 5 IOb62949 0 5 2 3 2 4 0 3 ï T m
SP 155-8
Direct Tensile Strength Testing at

6 Hours of Fiber Reinforced
Concrete Mortar Fractions
by P. P. Kraai and G. L. Vondran
Synopsis: The prime consideration in minimizing concrete cracks in

the field is to maximize the early (6-hour) tensile strength development
iii order to resist the volume reduction due to rapid water loss.
This paper describes a test method, which siindates field

conditions, for measiring direct tensile strength soon after initial set at
6 hours. The prototype direct tensile test described preseiits aii effort
to quanti@ results as a measure of crack resistance.
In this investigation three different types of concrete mortar

fractions were evaluated: i ) plain, 2) polypropylene fiber mixed in the
batch, and 3) the same fiber but roughened by intergrinding with
cement for better mechanical bond.
Results of tensile testing indicate that the process of iiitergrindiiig

fibers with cement improves the tensile strength of similar mortar
reinforced with smooth fibers by 63%. Comparing the grouiid fiber
results to a plain (no fiber) mortar mixture shows almost three tiines
higher direct tensile strength.
Based on this exploratory work on early tensile strength testing,

it appears to be a viable method to arrive at quantifiable values, which
will lead to a better understanding of the concrete cracking
plienoinenon and its control.
Keywords: Cements; comminution; cracking (fracturing); fiber reinforced

concretes; mortars !material); polypropylene fibers; shrinkage; tensile
strength
--``,`,-`-`,,`,,`,`,,`---

153
Not for Resale
A C 1 SP-I155 95 I0662949 0 5 2 3 2 4 3 22b M
154 Kraai and Vondran
Paul Y. Kraai was a Professor at San Jose State ünivcrsity where he

taught concrete materials and laboratory procedures. He was formerly
with Kaiser Sand and Gravel aiid Kaiser Cement as director of
laboratories and technical services. Kraai devoted his entire career to
field applications of cement and concrete technology.
Gary L. Vondran is President of VonTech International Corporation

in Los Altos, California. He has patented Intergrouiid Fiber Cement in
the U.S.A. and foreign countries. Vondran was fonnerly associated
with Novocoii Int'l and was Director of R & D for Fibennesh Co. He
is an active member of AC1 544 & 506 and ASTM (2.09.42 011 FRC.
INTRODUCTION
Most concrete shrinkage tests are conducted on specimens that

have been laboratory-cured for a minimum of 24 hours due to the
--``,`,-`-`,,`,,`,`,,`---
need to work with hardened concrete specimens. Many researchers'

conclusions drawn from these 1-day-old laboratory specimens
consider this age as the zero starting point of shrinkage. Much of the
early-age volume change has been ignored iti favor of looking at long
tenn shrinkage. By ignoring the fii-st-day volume changes of concrete,
laboratory tests do not reflect what really happens in the field. For
example, slab-on-grade coiicrete initially fills the form i 00%. Within
an hour or two consolidation is accomplished and evaporation starts.
During the first 6 hours after water is added, concrete is subject to n
wide range of conditions and variables that are critical with respect to
initiation, propagation and growth of early cracks.
The initial concrete slab thickness loss is as much as 0.4% and
length loss is about 0.3% within 4 hours after concrete is placed (1).
To evaluate the early cracking problem in the field, better laboratory
tests are needed that represent what really happens during these early
I
hours.

A C 1 SP-155 95 W 0662749 0523242 162
Testing of FRC 155
It is well kiiowii tliat fibers reduce both the ainoluit o í cracking

and the crack widths when properly used in concrete. Fibers are not a
panacea for structural cracking but are a significant help in reduciiig
early plastic shrinkage cracking tliat is so prevalent, especially in
flatwork concrete.
The laboratory experiments detailed herein describe our intent to
devise a consistent methodoloby for evaluating early tensile
properties, which are particularly critical when evaluating the effects
of adding fibers to concrete.
TENSILE STRENGTH TElSTING AT EARLY AGES
Concrete normally slirinks and cracks under adverse curing

conditions and improperly spaced control joints. The greatest
reduction iii volume as evidenced by cracking occurs between 1 and 4
hours after placement. Tlie amount of cracking due to drying
--``,`,-`-`,,`,,`,`,,`---
slirinkage tends to levei oft affer 4 hours. 'l'his leveling off is

confirmed by Adam Neville (2). Relative cracking results attained
within 24 hours of dain versus fiber reinforced concrete are also
indicative of intrinsic (nonstnictural) cracking properties of concrete
several years old. However, test results vary widely due to variations
in ambient conditions and materials. Reproducibility of test results is
a problem. No satisfactov standard test method exists for measuring
quantifiable values tliat relate directly to cracking. Early direct tensile
tests at 6 hours described in this report may be a reliable measurement
to control early concrete cracking.
Robert Tobin, a structural engineer in Southern California, has
reported: "Tlie potential volume change (and cracking tendency) iii
fresh concrete may be as much as 200 times greater than it is in
hardened concrete." And later in the saine report referenced by Kraai,
he says that "...the order of magnitude is tremendous and we may well
be paying insufficient attention to the really big shrinkage and the
original source of many cracks."(3)

A C 1 SP-1-55 95 0662949 0523243 OT9
Direct tensile strengths are measured as a normal standard of

quality for steel reiiiforciiig bars, wire, rope, fishing twine, plastics,
synthetic fibers and even rock, but not for concrete or fiber reinforced
concrete.
Critical Early-Age ShrinkaEe
Noted Swiss researchers, Kral and Gebauer, confirm that early-
age shrinkage is very important. Their report states, "This (second)
phase of shrinkage (at 2 to 3.5 hours) is thus the most critical stage for
the concrete with regard to slxinkage cracking."( 1) See Fig. I . The 1.5
hour second phase iininediately follows vertical settlement shrinkage.
In many plastic shrinkage cracking tests of slabs, normally the first
crack occurs within minutes after bleeding stops. It is at this very
short time period that surface cracking occurs. This (second) phase
period is when proper curing is difficult to time perfectly to stop
cracks because inicrocracking has already started to occur and the
time is so short. After 4 hours the high percentage of volume change
levels off, as shown in Fig. 1.
Importance 3f Tensile Strengt5 Values
In AC1 Monograph No. 6 Neville (2) states, "Mather pointed out
l
that ideally it should be possible to express the failure criteria under
1 all possible stress combinations by a single stress parameter, such as
strength in uniaxial tension. However, such a solution has not yet
been found." An updated quote from Bryant Mather's correspondence
states: "If we knew how to build a specimen with a known nuinber
and type of bonds, we might have a 'single stress (or strain)
--``,`,-`-`,,`,,`,`,,`---
parameter,' such as strain (or strength) in uniaxial tension that would

be ail the failure criterion you need under all possible stress (load)
i combinations."(4)
I
Practical Uniaxial Tension Test DeveloDment
~
Enhanced tensile strength of concrete is the most important
physical property that fibers contnbute to the brittle concrete, and yet
tensile testing has not become a part of FRC acceptance criteria.
Many previous tensile tests resulted in relatively high standard
deviations within batches with corresponding coefficient of variations

A C 1 SP-155 75 W 0662749 0 5 2 3 2 4 4 T35
Testing of FRC 157
of 7% for laboratory concrete. These excessive variatioiis are dire tü

the coinplex nature of the interaction of concrete inaterials and the
difficulty of tensile stress in a test specimen. Perhaps one reason
direct tensile testing lias not been endorsed in the last two or three
decades is that the concrete industry looks too closely for perfection
and homogeneity iii a heterogeneous product. An deid tensile test
may never be available, but a practical uniaxial tension test is
attainable.
A new direct tensile test method that simulates field conditions
has been developed at San Jose State University for measuring tensile
strength soon after initial set at 6 hours. Its simplicity helps make it a
practical and reliable direct tensile test.
INTERGROUND FIBER CEMENT EXPERIMENTS
An evaluation program was conducted to determine the

performance of Interground Fiber Cement ( P C ) where the only
material variation is surface texture and geometry created by
controlled grinding of the same fiber with dry cement clinker. In this
investigation tluee different types of concrete (usiiig only the mortar
without coarse aggregates) were evaluated: 1 ) plain, 2) polypropylene
fiber mixed in the batch, and 3) the same fiber but roughened by
interg-indiiig with cement in accordance with a new IFC process.(5)
These comparison tests were repeated five times.
The intent of the tensile strength testing procedure was tri
arrive at quantifiable values as early as 6 hours after mixing, and to
better understand what IFC contributes to coiitrolling the concrete
cracking phenomenon.
Direct Tensile Cylinder Molds
For early tensile strength values at 6 hours, the test procedure uses
plastic pipe as molds with duct tape sealing the tension-fracture zone
of the two part mold before casting. The molds are left on the
specimen, and duct tape is removed just prior to applying the tensile
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 IOb62949 0523245 971 M
load, wliicii is sliortiy after the mortar or concrete has artaineci initial
set. The molds also are lined with an expanded metal reiiiforciiig that
is bolted to the side wall for assuring a good mechanical as well as a
friction bond to the concrete to be tested. Eye-bolts are threaded into
a welded nui on top of a threaded rod that runs tlirough the side wall
sections that make up the top and bottom portions of the specimen
inold to form a siinple link at both ends of a cylindrical specimen (see
Fig. 2). I11 this direct tension test, the iiiteiit is to keep the weak link in
the center of the 6.7 inch ( 1 7 cm) length at the duct-tape joint of the
3.2 inch (8. i cm) diameter cylinder. Simplicity of mold construction
to help assure repeatability was a primary objective.
Test Set-up
I n these direct tension tests only the mortar fiaction of concrete
mixtures was employed. Future tests will expand the specimen sizes
larger than 4 inch ( I O cin) diameters so that concrete mixtures with
coarse aggregates can be tested. Figures 3 through 5 illustrate the
--``,`,-`-`,,`,,`,`,,`---
specimens, apparatus, test set-up arid procedures used. It is important
to note that the bottom is capped and the top of the specimen is left
open in order to more closely siinulaie iypicai siab sur%aceexpusure to
drying conditions. After casting, the specimens are placed in a
controlled chamber for the first 6 Iiours before testing. The chamber
is controlled at 80 degrees F. and air is circulated by internal fan.
Mixture Pronort ions

This test program used the mortar fraction based on mix
proportions of typical concrete [517 pounds per cubic yard (307
kg/m3) cement, 1 iiicli (25 min) maximum aggregate, concrete sand
and water for a 4 inch (I O cm) slump]. Polypropylene fibers were
added at the rate of 1% by mass of the portland cement. The fiber
amount expressed as a percentage of mortar volume was 0.56%
Expressing this mortar fraction as concrete with coarse aggregate, the
theoretical fiber content would be 0.33% by volume. The water and
sand requirements for all mixtures were held constant. Two types of
fibers (monofilaments and fibrillated) were used during the
comparison. All materiais are froin the same source. Mixture
proportions are listed in Table I . The first three mixtures are: i )
Control, plain no fibers; 2) Mixed-In Fiber (MIF); and 3 ) Iiiterground
A C 1 'aP-ai55 9 5 IO b 6 2 9 4 9 O523246 808 I
Testing of FRC 159
Fiber Cement (IFCj. Mixtiires ii 4-6 are basically the saine bui
adjusted to equal yield for a more precise volume comparison.
Intergrwnd Fiber Cernent M ateritils
Research on the interground fiber cement process in a laboratory
griJldiJig inill indicated that fibers become iinpregiated with grains of
cement (6). Figures 6 and 7 show some of the roughened
polypropylene fiber surfaces and cement impregnation. In this series
of lab tests, milling was controlled to simulate the actual cement
milling process where fibers more closely resemble the roughened
fiber shown in Fig. 7 microgaph. The same cement was used in all
1FC mixtures. The intergrinding of the 3/4 inch (19 mm) long
polypropylene fibers and cement was controlled at 15 minutes in a lab
ball inill. A sieved IFC sample of 131 fibers indicated an average
fiber length of 1 /2 inch ( 1 3.4 inm) after grinding.
--``,`,-`-`,,`,,`,`,,`---
Each type of polypropylene fiber was roughened by using the

new IFC patented process ( 5 ) . The comparison tests were conducted
on the nonnally smooth fibers typically batched and mixed by ready-
nixed concrete producers. Two &fferei?t fiber types (inonefi!xnent
and fibrillated) and fiber surface geometries (smooth and ground)
were compared to plain control mixtures without fibers in an effort to
quanti@ effectiveness of the new IFC process.
Three dry mortar mixtures (concrete without coarse aggregate)---

plain, MIF and IFC, were weighed out in separate containers and put
aside prior to mixing. All 14 batches were premixed dry for 30
seconds. Then wet mixing was done with a high-speed paddle blade
for 60 seconds. It was observed that all the Interground Fiber Cement
(IFC) mixtures were more workable than its counterpart MIF.
Apparently this consistency differential is due to the lower-aspect
ratios and multigraded smaller (3/4" MIF to 1/2" IFC average) lengths
caused by the interbyinding process.
Specimens were filled in layers. The mortar was consolidated by
rodding, tamping and tapping the mold to make sure it flowed into the
mesh attached on the side walls. Once level, another layer was added
until tlie cylinder was completely filled. The rop was levekd aiid
troweled once. Wet inortar is placed in a separate container to check
the rate of hardening. The test specimens are placed in the
temperature controlled chamber with continuous air circulation. T h e
is noted and 6 hours added as tim-to-test. Setting characteristics are
monitored with a Vicat needle as shown in Fig. 3. The cement paste
inust set sufficiently for testing. In each test series tlie identical
iiiixing and handling procedure is repeated for tlie three separate
batches.
Direct 1,oadioE Procedure
--``,`,-`-`,,`,,`,`,,`---
At the end of the prescribed 6-hour time interval, the specimens

are removed one at a time in the sequence cast, duct tape is removed
and specimens are suspended on the loading apparatus. A bucket is
liuing on the bottom eye-bolt (Fig. 4). Load is applied by opening a
valve wliicli allows dry sand to flow into the bucket at a rate of 1 1 to
15 pounds ( 5 to 7 kg) per minute.
All 28 specimens tested at the 6-lioiir age failed in tension at the

reliter sem. of t!ie meld. Tt is interesting to note that for a!l cases,
fracture occurred closer to or progressing into the top portion of the
specimens. This pheiiomenon is shown in Fig. 5. Also, the distance
from the center seam fracture appeared to be related to strength, ie.,
higher strengths failed closer to the center seain. The bucket of sand
and broken bottom half were weighed and recorded. The diameter of
I
I the cylinders was 2.64 inches (6.7 cin) at the center point seam for the
first three test series, which employed a washer insert to reduce
diameter. For the last two test series the diameter was 3.19 inches
(8.1 cm). Tensile strength is then expressed as the weight at time of
failure divided by the cylinder's cross-sectional area. Results are
tabulated in Table 2. The overall results are shown graphically in Fig.
8 as an average of two tests.
It would have been preferable to perform the tensile tests iii a
mechanical tensile testing machine, but such a machine with accuracy
in the low load ranges involved was not available at the time the tests
were conducted. In later-age experiments it was found that the plastic
pipe form with metal lath was inadequate to handle the higher loads
(greater than 1O0 pounds). Further later-age experiments employing
A C 1 SP-155 95 m 0 6 6 2 9 4 9 0 5 2 3 2 4 8 640
Testing of FRC 161
steel pipe with annular riiigs weided at the center fiactiire seairi
provided a more rigid mold. The steel pipe inold is shown in Figres
9 and 10. Not eiiough later-age experiments using the steel molds
were coiidiicted to provide statistically reliable data for this report.
A summary of the five series of direct tensile tests results is
shown and expressed in terms of stren@h improvement percentages in
Table 3.1, which illustrates the substantial tensile strength increases
attributed to the Intergroiuid Fiber Cement process.
--``,`,-`-`,,`,,`,`,,`---
CONCLUSIONS
1) The direct tensile test at 6 hours is a viable and quantifiable

method to determine cracking potential of fiber reinforced
concrete at early ages.
2) For later ages at loads over i 00 pounds a mechanical testing

machine shou!d be lised eiqhying steel pipe mo!rls d ! i welded
annular rings at the fracture seam.
3) The 6-hour tensile strengths for IFC mixtures [7 psi (48

kPa)] are 1 .6 times better than those of MIF [4.3 psi (30 kPa)]
and nearly hp!e plain mixtures [2.5 psi ( 17 kPa)] .
4) Expressed as a percentage of the average ó-liour tensile

strength, interground 1/2 inch ( I 3 mi)fibers outperfonned the
same iuiground (smooth) 3/4 inch (19 mm) long fiber by 173%,
and compared to plain mixtures without fibers by 281"/0.
5) The fiber cement bonding characteristics due to eiilianced

fiber roughness are increased hi all cases.
6) At equal fiber volumes results indicate that the tensile

strength increases are due to improved bonding.

A C 1 SP-155 95 0662949 0523249 517
FUTURE RESEARCH REFINEN1 ENTS
These 6-hour tensile test results relate to and help study early plastic
shrinkage cracking of slabs exposed to evaporative conditions.
Plastic shrinkage has already been established as acceptance criteria
for fiber reinforced concrete by one code ageiicy (7). Their criteria
requires tests nui side by side of plain and fiber reinforced slabs, and
comparing the square area of cracks, requiring a ininimuin 40%
reduction with FRC. Further study of the relationship of tensile strain
capacity at 6 hours with early cracking inay provide answers on how
fibers contribute quantifiably to the crack-resistance performance of
hydraulic cement composite inaterials. Future research will definitely
be needed in this area. Obviously more sophisticated improvements
will be made based on use and experience. This brief discussion will
lay the groundwork for stimulating renewed interest in new and more
effective tensile tests. Determining the necessary tensile strength to
ensure a crack-free concrete in the field was Paul Kraai's goal. With
his passing that goal lives on iii the form of a new FRC research tool,
the Kraai tensile test.
I
References
1. Kral, S. & Gebauer, J. "Shrinkage and Cracking of Concrete at Early Ages";
"Advances in Concrete Slab Technology" (Eds. E. K. Dihr and J. G. Munday),
Pergarnoti Press, Oxford & New York, 1980, pp. 412-420.
2. Neville, A. M. "Hardened Concrete: Physical and Mechanical Aspects", AC'I
Momgraph No. 6, ACI, Detroit, 19ï1, pp. 37-56.
3. Kraai, P. P. "Concrete Drying Shrinkage Facts and Fallacies", A C ï SZ'-76-3
"Designing for Crack & Shrinkage in Concrete Structures", AC1 Detroit,
1982, pp. 25-51, Ref 5, p. 36.
4. Mather, Bryant, Faxed Letter to Vondran, U.S. Department of the Army,
Vicksburg, April 20, 1994, pp. 1-3.
1 5. Vondran, G. L. "Interground Fiber Cement", United States Patent Number
5,298,071 issued March 29, 1994, pp. 1-6.
I 6. Vondran, G. L. "Interground Fiber Cement, A New Process", Proceedings
for the National Science Foundation-Universityof Sheffield Workshop on
"Fibre Reinforced Cement and Concrete" (Eds. R.N. Swamy & V.
Ramakrkhnan), July 28-30, 1994, pp 18-37.
7. ICBO AC32, "Acceptance Criteria for Concrete with Synthetic Fibers",
International Council of Building Officials, Whittier, CA, July 1993, pp. 1-13.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 9 5 IOb62949 0 5 2 3 2 5 0 2 3 9 I
Testing of FRC 163
TABLE 1 - MIXTURE PROPORTIONS FOR DIRECT TENSILE TESTS

Batch Weights in Grams for 2 Tensile Specimens
Materials Mix # 1._ _ 2 - _ 3 - - 4 _ - 5 - _ - 4
Cement 1075 1075 1064 1078 1078 1078
--- --- Il ---
--``,`,-`-`,,`,,`,`,,`---
Mixed-In 3/4" Fiber --- Il

Interground 1/2" Fiber --- --- 11 --- ___ Il
Sand 2575 2575 2575 2575 2543 2543
Water - _ I 660 -_600-- - 6 6 L --660--660_--_660-_-
Water/Cement .6 1 61 L A 1 h 1 . 6 1
TABLE 2 - LOADING TEST RESULTS OF TWO SPECIMENS AND AVERAGE psi
I'EST Mix # Description Loads íPounds) AverageBsi
I I Control 6.0 5.0 1.o
2 Mixed-In 314" Fiber 20.5 18.5 3.6
3 Intaground 112" Fiber 23.0 44.0 6.3
II 1 Control 15.0 10.5 2.4
2 Mixed-In 314" Fiber 22.5 23.0 4.2
3 Interground 1/2" Fiber- 29.5 45.0 6.8
III 1 Control 12.5 17.0 2.8
2 Mixed-In 3/4" Fiber 22.0 29.0 4.8
3 Interground 1/2" Fiber 42.0 49.0 8.5
IAbove data are based on 2.64" diameter and below are 3.19" diameter averages. 1
IV 1 Con~rol 22.0 27.c 3.1
3 Interground 13 mm Fiber 51.5 45.0 - 6.2
V 4 Control 21.0 27.5 3.1
5 Mixed-In 19 mrn Fiber 36.0 34.5 4.5
6 Intermound 13 mrnFiber 44.5 66.0 72
SI Unit Conversions: 1 pound = .4536 kg ; 1 inch = 25.4 mm; 1 psi = 6.895 kPa
TABLE 3 - SUMMARY OF CONTROL, MIXED-IN AND INTERGROUND FIBER

Tensile Strenyth in Pounds Per Square Inch
Test FiberType C&ol Mixed-In 3/4" Fiber Intepround 5/211 Fiber f
I Monofilament 1.0 3.6 6.3

II Monofilament 2.4 4.2 6.8
III Fibrillated 2.8 4.8 8.5
IV Monofilament 3.1 -- 6.2
V Monofilament 3.1 4 . 5 -
Averaeepsi &Pa) 2.5 í-17.1) 4.3 (29.6ì 7.0 (48. I).
TABLE 3.1 - SUMMARY PERCENT STRENGTH IMPROVEMENT

Colitrol Mixed -111 3/4" Fiber Iiiteraound 1/2 Fiber 11 '
Comparison No.1 100 YO 173 Oo/ 281 O/o

Comparison No.2 -------- 100 Yo 163 Yo
A C 1 SP-355 95 0662949 0523253 3 7 5
Cement Content : 300 k dd

w/c
YC
O 1 2 3 4 6 6 7 8
TIME AFTER WATER ADDITION, hr
Fig. 1-Vertical and horizontal shrinkage tests by Kral and Gebauer (3) show high-volume
change in second phase where most cracking occurs. After 4 hn shrinkage percent and cracking
level off
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 I0 6 6 2 9 4 9 0523252 001 =
Testing of FRC 165
--``,`,-`-`,,`,,`,`,,`---
Fig. 2-Direct tensile specimen molds
Fig. 3 - C u r i n g chamber and Vicat setting test

A C 1 SP-155 9 5 Ob62949 0523253 T 4 8 H
--``,`,-`-`,,`,,`,`,,`---
Fig. &Test apparatus and set-up
Fig. &Fractured tensile specimens after testing

A C 1 SP-155 95 Ob62747 0 5 2 3 2 5 4 7 8 4
Testing of FRC 167
Fig. ó-Micrograph of cernent grains imbedded in fibers after grinding in lab ball mill. Note
plain fiber cylinder shape
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 9 5 I0bb2949 0523255 B L O W
Fibers before After milling
Fig. 7-Electron micrographs showing fibers before and after intergrinding in cement plant
production run
TENSILE STRENGTH at 6 HOURS

Concrete Mortar Fraction---P. Kraai
7
I
7
4.3
2
1
U
PLAIN Mixed-In Fiber Interground
Fig. U o m p a r a t i v e results of mixed-in and interground fibers tensile strengths (average for 5
tests of 2 cylinders each)
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-255 95 fa 0662749 0523256 757 I
Testing of FRC 169
--``,`,-`-`,,`,,`,`,,`---
Fig. 9-Steel pipe test specimen prior to fracture at 28-day uge
Fig. lû-Steel pipe mold of fractured specimen at 28 days

A C 1 SP-1.55 '75 Ob62747 0 5 2 3 2 5 7 6 7 3
SP 155-9
Impact Tests on Cement-Based

Fiber Reinforced Composites
by N. Banthia, K. Chokri, and J. F. Trottier
Svno~is: This paper describes the construction of two simple impact machines--
one small with a capacity of 100 Joules and the other large with a capacity of loo0
Joules--designed to conduct impact tests on fiber reinforced mortars and concretes
in the uni-axial tensile mode. During a test, the applied load, accelerations and
velocities are measured such that with a proper analysis scheme, the raw data can
be analyzed to obtain fundamental material properties under impact loading.
Carbon, steel and polypropylene micro-fiber reinforced mortars and steel fiber
reinforced concrete were tested and it was demonstrated that the proposed
technique is a simple and rational method of obtaining meaningful material
properties. In general, fiber reinforced composites were found to be more impact
--``,`,-`-`,,`,,`,`,,`---
resistant than their unreinforced counterparts and the improvements were
proportional to the fiber volume fraction. In addition, both the unreinforced matrix
as well as fiber reinforced composites were found to b e stress-rate sensitive, but the
extent of sensitivity observed was smaller than usually reported in the literature for
cement-based materials under uni-axial tensile loading.
Keywords: Cements; composite materials; fiber reinforced concretes;

fracture properties; impact tests; mortars (material); strength

A C 1 SP-155 95 m 0 6 6 2 9 4 9 0523258 52T
172 Banthia, Chokri, and Trottier
AC1 member Nemkumar Banthia is an associate professor of civil engineering at

the University of British Columbia, Canada. He is a member of AC1 Committees
544, Fiber Reinforced Concrete; 549, Ferrocement and Other Thin Reinforced
Products; and 440, Fiber Reinforced Plastic Reinforcement.
Kouki Chokri was a graduate student at Laval University, Quebec, Canada where
this work was carried out. He investigated strain-rate effects in micro-fiber
reinforced cement composites.
Jean-FrancoisTrottier was a graduate student at Laval University, Ste-Foy, Quebec,

Canada when this work was carried out. He is now an assistant professor of civil
engineering at the Technical University of Nova Scotia in Halifax, Canada. He has
--``,`,-`-`,,`,,`,`,,`---
done extensive investigation of static and impact behaviour of steel fiber reinforced
concrete.
INTRODUCIïON
The o-Eresponse of any concrete in tension is governed by the properties

of the constituents and the interfaces, and these, in turn, are affected by the rate
at which the load is applied. By definition, loading may be considered to be static
only when it does not change with time. Static loads, however, are never
encountered in practice; all loads fluctuate with time. In most structures, however,
the loading characteristics do not change at a rate high enough to warrant a special
consideration, and quasi-static properties determined in routine lab tests are
normally sufficient. In some other structures such as, airport runways, piles, offshore
facilities, defense shelters and structures in seismic areas, however, loads may be
applied at rates high enough to warrant a special consideration of the sensitivity of
the materials to loading rate (1).
Several technique have been developed in the past to test concrete under
impact. The most commonly used is the Drop Weight Test (2-5) in which a mass is
raised to a predetermined height, and then allowed to drop directly on a concrete
specimen. Similar in principle are the SwingingPendulum Machines (6-8)--as in the
conventional Charpy or Izod impact machines used by the metallurgists--where a
swinging pendulum is allowed to strike a specimen in its path thereby transferring
momentum and causing high stress-rates. Other significant impact tests include the
Split Hopkinson Pressure Bar Test (9,lO) in which the specimen is sandwiched
between two elastic bars and high stress-rates are generated by propagating a pulse
through one of the elastic bars using a drop weight or a similar device. In most
modern impact test systems, sufficient instrumentation is provided such that along
with loads and deformations, additional specimen responses such as accelerations,
velocities, etc., are also measured; these are needed for a proper analysis of the
data later. Using the various test techniques, plain as well as fiber reinforced
concrete have been shown to be stress-rate sensitive under all modes of loading
(11-15).

A C 1 SP-155 95 IObb2747 0 5 2 3 2 5 9 466
Testing of FRC 173
In spite of the general recognition that cement-based materiais demonstrate

sensitivity to stress-rate, our understanding of this phenomenon is, at best, only
qualitative. In the quantitative sense, there is very little agreement in the literature,
and based on the published data, concrete can be classified as a material highly
sensitive to stress-rate to one with no sensitivity at all. At the root of this confusion
is the absence of a standardized test technique for testing cement based materials
under impact. The data reported in the literature are obtained using different
impact machines with different specimen gripping techniques, different energy loss
mechanisms and different ways of generating high stress-rates, ali of which have a
strong influence on the results.
Our understanding of the impact resistance of cement-based materials in

general, and their fiber reinforced counterparts in particular, is very limited. The
primary reason for this lack of understanding is the absence of a standardized
testing technique and a rational analysis scheme. Continued research is needed first
to fully understand the influence of the various test parameters o n the resulting
impact data, and then to develop a simple testing technique capable of generating
reproducible and meaningful material characteristics.
EXPERIMENTAL
Two separate impact machines were designed. One with a smaller capacity
(u 100 Joules) was suitable for testing smaller specimens made of fiber reinforced
mortars, and the other with a larger capacity (u loo0 Joules) was designed for
testing fiber reinforced concrete specimens.
The Concept
The concept behind the two impact machines may be described with respect
to the schematic drawn for the 1000-Joule machine shown in Figure 1. A close-up
of the specimen gripping and loading arrangement is also shown in Figure 1 (inset).
In principie, the specimen bridges two supports, A and B, with support B mounted
o n rollers (called the trolley) and support A being fuced. Support B is struck by the
swinging pendulum on impact points located on either side of the specimen and in
the same plane as the specimen. Support A being f i e d , this causes tensile loading
in the specimen.
The impact hammer carries two dynamic load cells mounted on either side
that record the contact load vs. time pulse during the impact. Under an impact, the
specimen fractures and the trolley travels toward the shock absorbers. O n its way,
the trolley passes through two photocell assemblies where its post-event velocity is
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-L55 75 0662947 0523260 L ô ô
recorded. The other instrumentation provided includes accelerometers (f 500 g,

9.98 mV/g) mounted both on the trolley and the hammer and a photocell assembly
t o record the hammer velocity during approach. The photocell assembly that
records hammer velocity also supplies the triggering pulse €or the data acquisition
system which, once triggered, records data at a sampling rate of 1.2 MHz. The
other details of the two machines are given in Table 1.
The objective of the impact tests was to obtain strength and fracture energy
values for various fiber reinforced cementitious composites as a function o€ the
applied stress-rate. Considering the free body diagram of the troZZey (part B) itself,
the horizontal force equilibrium (ignoring damping) may be written as:
The hammer load Pt(t) in the above equation is the sum of the loads
recorded by the two load cells mounted on either side o€ the hammer and Ps(t) is
the specimen load. Pj(t) is the inertial load (2) given rise to by the accelerations
in the system. If at(t) is the trolley acceleration and mt and m, the masses of the
trolley and the specimen, respectively, Eqn. 1 can be written as:
a,(t) and 4 in the above Eqn. are the specimen stress and the cross-sectional area,
respectively. Note that it has been tacitly assumed that the loading is perfectly
aligned along the specimen centroidal axis without any eccentricity. This was
independently verified by the data from the two load cells as described later. With
trolley accelerations at(t) recorded by the accelerometer, Eqn. 2 can be solved for
a,(t), the peak value o€which could then be taken as the tensile strength of the
composite under impact.
The fracture energy consumed by the specimen was determined using the
principle of conservation o€energy. Using the impulse-momentum principle (6,7,8),
the energy lost by the pendulum (E,,) during its contact with the trolley can b e
written as:
where,
mp=mass of the pendulum

I ~ ( t ) d=contact
t impulse
vi =initial velocity of pendulum

--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0662949 0523261 014 I
Testing of FRC 175
=i/2gh
g =earth’s gravitational acceleration
h =height of hammer drop
If one can ignore the frictional and other losses of energy, the energy lost by the
pendulum can b e regarded as the sum of the energies consumed by the specimen
during fracture (Es) and that gained by the trolley as post-fracture kinetic energy
(EJ. In other words,
Eh - Et +Es (4)
Further, if vt is the post-fracture velocity of the trolley as recorded by the base-

mounted photocell assemblies, then,
(5)
Using Eqns. 3, 4 and 5 and solving for E,,

r
With all quantities on the right hand side known, the fracture energy consumed by
--``,`,-`-`,,`,,`,`,,`---
the specimen can be calculated.
Preparatory testing carried out to confirm the validity of some of the

assumptions made in the above analysis is described elsewhere (16,17). First of all,
it was verified that the applied load was truly non-eccentric in nature. This was
done by comparing the signals from the two load cells mounted on the hammer and
striking on opposite sides of the specimen, and ensuring that there was no apparent
lag between the two. A lack of lag meant that the load was applied evenly on both
sides of the specimen and eccentricity, if at all, was not large. Further tests were
carried out with no specimen in the system and the recorded tup loads were
compared with the calculated inertial loads. With reference to Eqn. 1 (with n o
specimen in the system; P,(t) = O), P,(t) and Pi(t) matched within reasonable limits.
When a specimen was placed in the system (Figures 2 and 3), however, it was
observed that the tup load vs. time trace had two distinct peaks, with the first peak
fully accounted for by inertia and the second peak representing the true specimen
loading with a negligible inertial load. It may be postulated that in the early part of
an impact, the applied load is utilized only in accelerating the system with no load
applied t o the specimen yet. It is only after the trolley gains enough momentum
that the specimen experiences the tensile load which manirests itself as the second
peak in the recorded impulse. For both 100-Joule and 1000-Joule machines, the
pendulum was allowed to strike the trolley at difrerent velocities and each time the
first peak was found to be entirely inertial with only the second peak representing

A C 1 SP-155 95 I0662949 0 5 2 3 2 6 2 T50 =
the true specimen loading. After the first peak, the accelerations died down and the
inertial forces could be ignored.
It was also found that because of hard steel-to-steel impact, the inertial
loads were very high and strong vibrations persisted during the test. Consequently,
in order to reduce the trolley accelerations and machine vibrations, the rate of
transfer of momentum between the pendulum and trolley was reduced by
introducing a rubber pad (stiffness = 2.83 MN/m) at the pendulum-trolley contact
points. The use of a rubber pad reduced the inertial load, delayed the occurrence
of both the peaks and reduced the rate of transfer of momentum (16). Indeed, the
rubber pad would absorb some energy from the hammer at first contact. However,
since the rubber pad is elastically unloaded at the end of the test, the energy
absorbed by the pad should not appear in the final equation of energy balance
(Eqn. 6).
RESULTS
The 1WJoule Impact Machine
Some load-time pulses obtained for mortars reinforced with O, 1, 2 and

3% of carbon, steel and polypropylene fibers using the 100-Joule impact machine
are shown in Figures 4a, b and c, respectively. These tests were conducted on 28-
day old specimens with a hammer approach velocity of 1 m/s at which the hammer
had an incident energy of 21.25 Jouies. The approximate stress-rate was 8 x lo3
MPa/s. Companion static tests were carried out in a floor mounted Instron at a rate
of 0.02 MPa/s which led to an impacthtatic stress rate ratio of 0.4 x lo6. Notice in
the load-time curves in Figure 4 that the first peak load is the same regardless of
the specimen type. As described earlier, the first peak in these curves represents
pure inertial load stemming from system acceleration with no stressing load applied
on the specimen yet. It is only after the first peak that the specimen experiences
ascending stresses which lead to the second peak in the curve at which presumably
the specimen fractures.
Notice an increases in the area under the curve ( p d t } due to the

presence of fibers. With reference to Equation 3, an increase in P d t meant that
the pendulum lost a greater amount of energy during impact in the case of a fiber
reinforced concrete specimen. In addition to the pendulum losing a greater amount
of energy, there was also a corresponding decrease in the post-fracture trolley
velocity which, on the whole, led to a significant increase in the fracture energy
values for fiber reinforced specimens when compared with their plain counterparts.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 111 0 6 6 2 9 4 9 0523263 9 9 7 I
Testing of FRC 177
The 1ûOO-Joule Impact Machine
In Figure 5, some load-time pulses for concretes with different strengths

are given. In Figures 6a, b, c and d, load-time pulses for steel fiber reinforced
concrete with various fiber geometries are shown. These impact tests were also
performed twenty eight days after casting with a hammer approach velocity of2.94
mís. The 67 kg hammer at this velocity had an incident energy of 291 Joules which
was more than adequate for fracturing even the toughest composite. This approach
velocity produced an average stress-rate of 3.5 x lo3 MPa/s and an average strain
rate of 0.08 sec-'. Companion static tests were carried out using the same setup as
for the impact tests but the hammer was pushed slowly against the trolley using a
hydraulic jack. This produced an approximate stress rate of 0.04 MPa/s and gave an
impact to static stress-rate ratio of 87500. Notice in Figure 5 that an increase in the
impact strength of concrete occurred with an increase in the static compressive
strength. The increased area under the curve beyond first peak for both higher
strength (Figure 5) and fiber reinforced concrete (Figure 6) indicates that there is
an increase in the fracture energy absorption under impact when the static
compressive strength of concrete is increased and when fibers are incorporated in
the mix.
AhWLYSIS AND DISCUSSION
The data obtained from the above impact machines is analyzed in

Table 2. Note that since the static tests were not conducted in a closed-loop test
system, unstable fractures occurred in most cases and consequently, only the
strength values, and not the fracture energy values, are reported for these tests.
Tensile Strengths Under Impact
Based on the results in Table 2, it may be noted that mortars reinforced

with micro-fibers were stronger than their unreinforced counterparts both under
impact and static conditions. Also, strengths under impact conditions were greater
than those under static conditions giving impact/static strength ratios greater than
unity (Table 2). Note also that the stress-rate sensitivity of micro-fiber reinforced
mortars increases with an increase in the fiber volume fraction. In the case of steel
fiber reinforced concrete reinforced with a nominal volume fraction of steel macro-
fibers, the strengths are not altered due to fibers neither under static nor under
impact conditions.
If cement-based materials can be assumed to behave in a linearly elastic

manner, then the strength (o,)can be expressed as a function of stress-rate ((2) by
using the principles of linear elastic fracture mechanics (18):
ho, - -N+1
1
h
1
BU + - In (Oy-, - (7)
N+1
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-I155 95 O662949 0 5 2 3 2 6 4 8 2 3 W
where, N is a material constant (slope of the stress intensity factor, KI, vs. crack
velocity, V, plot on a logarithmic scale), subscripts i and f refer to initial and final
conditions, respectively, and B is given by (13,lS):
2K;iN(N+1)
B -
A (N-2)
whereA and Y are constants. In Table 2, the values of constant N obtained on the
basis of Eqn. (7) are given. Notice that N varies between wide limits and is, in
general, higher than normally reported, particularly for the plain unreinforced
matrices. In other words, the sensitivity of concrete to stress-rate as observed in
these tests is less pronounced than that observed by others (15, 19, 20). This is
surprising given that concrete is expected to be far more sensitive t o stress-rate in
uni-axial tension than in any other mode. The following are the probable causes:
i) Estimated Stress Rate: The imposed stress-rates under impact were calculated
from the slopes of the contact load vs. time pulses in rising part of the "second
hump" (see Figures 3, 4, 5 and 6) in the load-time plots and then averaged. This
gives only an approximate stress-rate.
iil Eccentricity in Loading: The load eccentricity in rapidly applied impact loads
(where the specimen did not get much time to align itself in the direction of the
load) was greater than that in slow static tests. This may have led t o decreased
apparent strengths under impact loading and hence lower impact/static strength
ratios.
iii) Lack of a Linear ResDonse: As pointed out by Mindess (13), the assumption of
a linear elastic fracture response assumed in Eqn. (7) is not entirely valid. Concrete
is not ideally brittle and the o-Eresponse for both concrete and its fiber reinforced
composites is far from linear. The values of constant N therefore may not be
expected to capture the true nature of stress-rate sensitivity in these materials.
iv) Stress-Rate Vs. Strain Rate: Assuming the material is linearly elastic, the
imposed stress-rate ( u ) in a test can be related t o the imposed strain-rate (i)
through a simple Equation:
ir-E& (9)
where E is the elastic modulus. Eqn. 9 implies that for a given applied stress-rate
a stiffer materials would be subjected to lower strain-rates. Which means that if the
failure is governed by a limiting strain rather than a limiting stress value, the data
must be normalized and different materials must be compared only on an equal
strain-rate basis. This is, however, not attempted here given the lack of appropriate
--``,`,-`-`,,`,,`,`,,`---
values of dynamic moduli for the materials tested.

A C 1 SP-1.55 95 B O662949 0 5 2 3 2 6 5 7 6 T I
Testing of FRC 179
Fracture Enerrria Uoder Imnact
Higher fracture energy values recorded for plain concrete with a higher
compressive strength (Figure 5 and Table 2) is, in general, against the prevalent
belief that high strength concretes are more brittle than normal strength concretes.
O n the other hand, the observed increases in the fracture energy absorption due
to fiber reinforcement (Figures 4 and 6; Table 2) are well anticipated and in tune
with most published data. When fracture energy values are normalized by dividing
them by the area of the specimen, the 1000-Joule machine appears t o give much
higher specific fracture energy values. This indicates that the normalized fracture
energy values are specimen size dependent and also dependent o n the machine
capacity. There is also doubt over this way of normalizing the fracture energy
values.
CONCLUSIONS
1. The paper proposes a simple technique of testing concrete and its fiber
reinforced composites under impact. It is demonstrated that with the
proposed technique, meaningful material properties under impact can
be obtained, which can then be useful in designing structures subjected
to impact loads.
2. Impact data for plain and fiber reinforced cement-based materials

indicates that these materiais are sensitive to stress-rate. In general, they
--``,`,-`-`,,`,,`,`,,`---
are both stronger as well as tougher under impact loading as compared
to static loading.
3. Fiber reinforcement is significantly effective in improving the fracture

energy absorption under impact. The improvements are, however, fiber
type and geometry dependent.
ACKNOWLEDGMENTS
The continued support of the Natural Sciences and Engineering

Research Council of Canada is gratefully acknowledged.
REFERENCES
1. Struck, W. and Voggenreiter, W., Examples of Impact and Impulsive

Loading in the Field of Civil Engineering, Materiais and Structures
(RILEM), 8(44), 1975.

A C 1 SP-15-51 9 5 = 0662949 O523266 bTb
2. Banthia, N., Mindess, S., Bentur, A and Pigeon, M. Impact Testing of

Concrete Using a Drop Weight Impact Machine, Experimental Mech.,
1989, 29(2), pp. 63-69.
3. Naaman, A. and Gopalaratnam, V., Impact Properties of Steel Fiber

Reinforced Concrete in Bending, The International Journal of Cement
Composites and Lightweight Concrete, 3( i), 1983, pp. 2-12.
4. Suaris, W. and Shah, S.P., Properties of Concrete and Fiber Reinforced

Concrete Subjected to Impact Loading, American Society of Civil
Engineers, Journal of the Structural Division, Vol. 109, SE', July, 1983, pp.
1717-1741.
5. Gokoz, U. and Naaman, AE., EfCect of Strain Rate on the Pull-out

Behaviour of Fibers in Mortar, International Journal of Cement Composites
and Light Weight Concrete, Vol. 3, No. 3, 1983, pp. 187-202.
6. Hibbert, A.P. Impact Resistance of Fiber Concrete, Ph.D. Thesis, University

of Surrey (UK), 1979.
7. Gopalaratnam, V., Shah, S.P and John, R. . A Modified Instrumented

Charpy Tests for Cement Based Composites, Experimental Mechanics,
24(2), 1984, pp. 102-111.
8. Banthia , N. and Ohama, Y., Dynamic Tensile Fracture of Carbon Fiber

Reinforced Cements, Proc. Int. Conf. on Recent Developments in Fiber
Reinforced Cements and Concretes, Cardiff, UK, 1989, pp. 251-260.
9. Zielinski, A.J., Fracture of Concrete and Mortar Under Uniaxial Loading,

Ph.D. Thesis, Delft University of Technology (The Netherlands), 1982.
10. Malvern, L.E., Tang, T., Jenkins, D.A and Gong, J.C., Dynamic
Compressive Strength of Cernentitious Materials, In Mindess, S. and Shah,
S.P. (eds.), Cement-Based Composites: Strain Rate Effects on Fracture,
Materials Research Society Symposia Proceedings, Vol. 64, Materials
Research Society, Pittsburgh, pp. 119-138.
11. Sierakowski, R.L., Dynamic Effect in Concrete Materials, in Shah, S.P.

(ed.), Application of Fracture Mechanics to Cementitious Composites,
Martinus Nijhoff Publishers, Dordrecht, 1985, pp. 535-557.
12. Reinhardt, H.W., Strain Rate Effects on the Tensile Strength of Concrete
as Predicted by Thermodynamic and Fracture Mechanics Models. In
Mindess, S. and Shah, S.P. (eds.), Cement-Based Composites: Strain Rate
Effects on Fracture, Materials Research Society Symposia Proceedings, Vol.
64, Materials Research Society, Pittsburgh, 1986, pp. 1-14.
13. Mindess, S., Rate of Loading Effects on the Fracture of Cementitious

Materiais, In Shah, S.P. (ed.), Application of Fracture Mechanics to
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 IPI 0 6 6 2 9 4 9 0.523267 5 3 2 =
Testing of FRC 181
Cementitious Composites,Martinus Nijhoff Publishers, Dordrecht, 1985, pp.

617-636.
14. Gopalaratnam, V. and Shah, S.P. (1986). Properties of Steel Fiber

Reinforced Concrete Subjected to Impact Loading, Journal of the
American Concrete Institute, Vol. 83, No. 1, January, pp. 117-126.
15. Banthia, N., Mindess, S. and Bentur, A. Impact Behavior of Concrete

Beams. Materials and Structures (Paris), Vol. 20, No. 119, 1987, pp. 293-
302.
16. Banthia, N. et al, Fiber Reinforced Cement-Based Composites under

Tensile Impact, Advanced Cement Based Materials, 1, 1994, pp. 131-141.
17. Banthia, N. and Trottier, J.-F., Impact Resistance of Concrete and Fiber
Reinforced Concrete under Uni-axial Tension, ACI, SP on Impact and
Impulsive Loading, submitted.
18. Nadeau, J.S., Bennet, R. and Fuller, E.R. Jr., An Ekplanation of the Rate-
of-Loading and Duration-of-Load Effects in Wood in Terms of Fracture
Mechanics, J. of Mat. Science, 17, 1982, pp. 2831-2840.
19. Ross, C.A., Fracture of Concrete at High Strain-Rates, in Toughening

Mechanisms in Quasi-Brittle Materials (Ed. S.P. Shah), NATO AS1 Series,
Vol. 195, Kluwer Academic Publishers, 1991, pp. 577-596.
20. Zielinski, A., and Reinhardt, H.W., Stress-Strain Behavior of Concrete and
Mortar at High Rates of Tensile Loading, Cement and Concrete Research
12, 1982, pp. 309-319.
TABLE 1 - DETAILS OF IMPACT MACHINES
Small 100 J Impact Large loo0 J Impact

Machine Machine
Hammer Mass 42.5 kg 67 kg
Allow, Drop Height 0.3 m 1.55 m
Machine Capacity =lOOJ ~1000J
Load Cell Capacity 22.3 kN 222.4 kN
Trolley Mass 4.524 kg 20.53 kg
Photocell Spacing 38 mm 37 mm
Accelerometers & 500 g, 9.98 mV/g 2 500 g, 9.98 mV/g
Specimens Geometq Dog-bone Specimens, Contoured Specimens,
--``,`,-`-`,,`,,`,`,,`---
78 mm long, 25.4 mm x 400 mm long, 75 mm x
25.4 rnrn critical section 75 mm critical section
A C 1 SP-155 95 I0662947 0 5 2 3 2 6 8 479 M

--``,`,-`-`,,`,,`,`,,`---
Il
Y
c>
O 0
52
A C 1 SP-155 75 0 6 6 2 9 4 9 0 5 2 3 2 b î 305
Testing of FRC 183
e
a
U
E
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 I0 6 6 2 9 4 9 0523270 O27 =
O 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Time, ms
--``,`,-`-`,,`,,`,`,,`---
O 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Time, mc
Fig. 2-Load-time plots with and without specimen for small 100-Joule impact machine
ao I I I 1 I I I I I
70
60
50
5
n- 40
Id
O
30
20
10
n
O 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5
Time, ms
Copyright American Concrete Institute Fig. 3-Load-time plots with and without specimen for large 1000-Joule impact machine
A C 1 SP-155 9 5 I0662949 0 5 2 3 2 7 3 T 6 3
Testing of FRC 185
Time, ms
Fig. 40-Impact load-time plots for mortars reinforced with 1, 2, and 3 percent by volume of
carbon fiber
Fig. 4b-Impact load-time plots for mortars reinforced with 1, 2, and 3 percent by volume of
steel fiber
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-15.5 95 = 0662949 0523272 9 T T
Time, ms
Fig. 4-Impact load-time plots for mortars reinforced with 1, 2, and 3 percent by volume of
polypropylene fiber
1 O0 I I l 4 I I I
High Strength Concrete (85 mPa)

__-- Mid-Strength Concrete (52 mPa)
-.-<- Normal Strength Concrete (40 mPa)
I I
O 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0 4.5 5.0
Time, ms
Fig. %Impact load-time plots for concretes with various strengths (high strength: 85 MPa;
midstrength: 52 MPa; normal strength: 40 MPa)
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-L55 95 = 0662949 0523273 836
Testing of FRC 187
1 O0 I
90 -
ao -
70 -
-
5 60 _ - - -Plain
U- 50 - Plain + Fiber FI
m
9 40-
30 -
20 .
10 -
O 1 .o 2.0 3.0 4.0 5.0 6.0
O
Time, ms
Fig. 60-Impact load-time plots for steel fiber reinforced concrete with 0.51 percent by volume of
hooked-end fiber F1 (the plain matrix had static compressive strength of 52 MPa, which increased
to 56 MPa with fiber addition)
100 r I I I
Time, ms
Fig. ób-Impact load-time plots for steel fiber reinforced concrete with 051 by volume of
crimped fiber F2 with circular section (the plain matrix had static compressive strength of 52 MPa,
which remained unchanged with fiber addition)
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 = 0662949 0523274 772
1O0
90 -
80 -
70 -
- - - - - Piain
5 60
Plain + Fiber F3
x- 50 -
40-
30 -
20
--``,`,-`-`,,`,,`,`,,`---
10 -
O
O 1 .o 2.0 3.0 4.0 5.0 6.0
Time, rns
Fig. óc-Impact load-time plots for steel fiber reinforced concrete with 0.51 percent by volume of
crimped fiber F3 with crescent section (the plain matrix had static compressive strength of 52
MPa, increased to 53 MPa with fiber addition)
100 I I I I 1
---- Plain
Plain + Fiber F4
u-
m
O
-I
I 1 4
O 1.o 2.0 3.0 4.0 5.0 6.0

Time, ms
Fig. ód-Impact load-time plots for steel fiber reinforced concrete with 0.51 percent by volume of
twin-cone fiber F4 (the plain matrix had static compressive strength of 52 MPa, which increased
to 53 MPa with fiber addition)

A C 1 SP-155 95 W Obb2949 0 5 2 3 2 7 5 607 W
SP 155-10
Measurement of the Pull-Out Force

at Different Rates of Loading
by A. Patios and S. P. Shah
SvnoDsis: With the objective of understanding the reinforcing

mechanisms of fibers in steel fiber reinforced concrete, the adherence
between the fiber and the matrix is studied by conducting pull-out test
of fibers from a cementitious matrix. In this paper the effect of factors
such as loading rate, inclination of fibers and number of fibers have
been investigated. An innovative measurement system is developed for
high rates. It was experimentally obtained that by increasing the rate
of loading, both pull-out resistance and slip at peak were increased.
Peak pull-out force presents a higher rate sensitivity for a higher
number of fibers. The lower the number of fibers the higher the slip at
peak rate sensitivity. Regardless of the number of fibers a higher rate
sensitivity for inclined fibers was observed.
Kewords: Fibers; impact tests; loading rate; pullout tests; slippage; steels
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 = 0662949 O523276 545
190 Pacios and Shah
Antonia Pacios is a research architect working at the I.E.T.C.C., Spain.

She obtained her Bachelor's degree and her Ph.D. at the Polytecnic
University of Madrid. She spent 2-1/2years at Northwestern University
where she worked on pull-out of fibers. Her research interests include
fiber reinforced construction materials and high performance concrete.
AC1 member Surendra P. Shah is a Walter P. Murphy Professor of Civil

Engineering a t Northwestern University, Evanston, Illinois and Director
of the National Science Foundation Center for Science and Technology
of Advanced Cement-Based Materials. He has been involved with
research on constitutive relationships, failure and fracture of concrete,
nondestructive testing, impact and impulsive loading and fiber
reinforced concrete. He is a member of several AC1 Committees, has
published more than 300 papers, has edited 12 books and coauthored
a book on fiber reinforced cement-based composites. He has received
the AC1 Anderson Award, ASTM Thompson Award, and the RILEM
--``,`,-`-`,,`,,`,`,,`---
Gold Medal Award. He has been awarded the Alexander von Humboldt
Senior Visiting Scientist Award t o Germany as well as the NATO
Visiting Senior Scientist t o France.
INTRODUCTION
Characteristics of the fiber-matrix interface are very important

for fiber reinforced composites. To evaluate the behavior of the fiber-
matrix interface, a test of fibers being pulled-out from a matrix is often
.
used In this area, some studies have been previously conducted t o
evaluate the influence of fiber type, matrix type and fiber embedded
length on the interfacial behavior. [I -31. Extensive studies have been
conducted t o understand and evaluate the loading rate effect in
tension, flexure and compression of concrete structures, but some
conflicting results have been reported on t h e rate effect on the
response of the fiber-matrix interface. For example Gokoz and Naaman
[41 found the pull-out behavior of straight fibers t o be insensitive t o
pull-out slip rate, whereas Gray and Johnston 151 reported different
results.
When short and randomly distributed fibers are used, they

usually have an inclination t o the matrix crack. When the matrix crack
i is widened, it applies a pull-out force on fibers. As shown in Fig. 1, the
pull-out of an inclined fiber is a complex mechanism that involves
debonding of fiber-matrix interface, slipping between fiber and matrix

A C 1 SP-155 95 0662749 0 5 2 3 2 7 7 481
Testing of FRC 191
on the debonded interface, bending of fiber at the surface of the matrix

crack and spalling of the matrix due to the bending of fibers. [6-81.
Since the way these mechanisms are modified by the slip rate is
unknown, it is necessary to conduct a systematic study on the rate
effect on pull-out behavior of inclined fibers.
Experiments and theoretical models often treat a single fiber

being pulled out from a matrix. However, effects of orientation and
inclination may be influenced by the number of fibers pulled out.
In this paper an experimental study conducted t o evaluate the

rate effect of fiber pull-out behavior is described. An innovative
measurement system was developed for the Charpy pendulum. This
measurement system allows one t o use t w o independent methods t o
evaluate the pull-out load. Quasi-static tests were conducted t o
compare the results with those obtained from the Charpy tests. Fibers
with inclination angles of O, 14, 27 and 37 degrees were pulled-out
from a cementitious matrix. Two groups of fibers were also tested.
The study described here is further detailed in Ref. 9.
FIBER PULL-OUT EXPERIMENT
For the specimen preparation, a brass mold was used, similar t o

the one developed by Li et al. 131. A series of six specimens, which
assures homogeneity, are cast horizontally. Specimens are cut by a
bandsaw and then ground t o eliminate any fiber damage produced in
the preparation process. A specimen consists of t w o parts separated
by a brass plate. Brass plates with 16 holes were used t o provide
alignment of fibers during construction and t o give different inclination
of the fibers. Two rows of fibers connected both parts of the
specimen. Different fiber inclinations were obtained by passing the
fibers through different holes (Fig. 2) For example, t o achieve a certain
--``,`,-`-`,,`,,`,`,,`---
inclination, a fiber can pass plate 1 in hole A, then meet plate 2 in hole
B'. In order t o keep a symmetric geometry, the second row of fibers
were inclined with the same angle but in the opposite direction. During
testing this guide plate was used t o separate the pullout section of the
specimen from the anchored end and provide for the transfer of the
load. Both ends of fibers are embedded in the matrix. However, part of
the fiber within the long embedment side was periodically deformed t o
assure that the fiber was pulled-out from the end with the short
embedment length.
The pull-out of fibers from the cementitious matrix with quasi-

static rates was conducted using a closed loop MTS machine. Two
A C 1 SP-I155 75 = Obb2949 O523278 318 D
192 Pacios and Shah
extensometers, with a range of 0.075 in. (1.9 mm) and a gage length
of 0.5 in. (12.7 mm) were mounted on both sides of the specimen t o
measure slip displacement. The average reading from the t w o
extensometers was used as the feedback signal t o control loading rate.
Once the debonding process was completed the rest of the fiber was
pulled out from the matrix a t a constant rate under stroke control with
the same speed as that observed during the first part of the test. A
scheme of this configuration can be found in Fig. 3.
Pull-out of fibers at high slip rate was performed using a

modified Charpy pendulum, which is discussed next.
CHARPY MODIFICATION
A conventional Charpy pendulum, Tinius Olsen Model 64, was

modified and instrumented to facilitate the test of fiber pull-out from
concrete at high rates. The design of the support and striker was based
on the following requirements: a) the same Set-up should also be used
in the MTS test machine for the quasi-static tests, b) the impact force
should be applied t o the center of the specimen so that only axial pull-
out force is generated on fibers, and c l there should be a certain gap
between the bottom of the specimen and the Charpy machine so that
no friction forces will be developed when the specimen moves under
impact. A low blow fixture and a hydraulic shock-absorber previously
designed were used. [IO].
Two main modifications have been made in order t o satisfy

these requirements: a support device and a U-shaped hammer head
were used as shown in Fig. 4. The support device was instrumented
with t w o dynamic ring load cells (with a capacity of 1,000 Ib), which
are named load cells A and B. The U-shaped head was instrumented
with another load cell with a capacity of 5,000 Ib (referred to here as
load C). At the same time t w o non-contact gages, multi-purpose
variable impedance gages (MULTI-VIT, Kaman KD-2300-2S1, which are
indicated as sensors F a n d R were used to measure the slip between
fibers and the matrix. These t w o sensors had a measurement range of
0.1 in. (2.54 mm). The gage length for the slip measurement was 0.5
in. An accelerometer was mounted at the bottom of the specimen to
measure the acceleration of the specimen during the impact. The
obtained acceleration, after being filtered (5-1.0 kHz) to eliminate
parasitic noises, can be used t o evaluate the inertial force generated
during impact. A detail of the test set up for the impact test is also
shown in Fig. 5.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0662949 0523279 254
Testing of FRC 193
Two oscilloscopes, with a total of six channels, were used t o

store all signals from instrumentation. A 4-channel digital oscilloscope
(Nicolet 4094) associated with t w o 2-channel differential amplifiers
was used t o store signals measured by three load cells (load cells A ,B,
--``,`,-`-`,,`,,`,`,,`---
and C in Fig. 4.) and the accelerometer (sensor 4).The amplifiers had
a sampling rate of 500 ns per point. A Tektronix 2232 oscilloscope
was used t o store the slip signals measured by the t w o MULTI-VIT’S
(sensors R and 13.Load cell A was used to trigger the whole system
as shown in Fig. 4.
The above mentioned arrangement of sensors allows one t o

evaluate the force contributing to pull-out of fibers by t w o different
methods. Results obtained by the t w o methods are used t o check each
other t o assure accuracy of the measurement in the present study. The
pull-out load can be directly measured from load cells A and 8 mounted
on both arms of the support. This pull-out load, PABl is simply the sum
of forces measured by load cells A and BI
PAB=FA+FB
where FA and FE are the force measured from load cells A and B,
respectively.
On the other hand, the pull-out load can alternatively be

measured from load cell C.However, since the impact hammer directly
hits load cell C, some acceleration will be generated on the specimen
during impact. This acceleration will result in the inertial force, which
has been reported in many previous studies. (1 1, I 21. As a result, the
force obtained from load cell C is the sum of the pull-out load and the
inertial force. This inertial force can be evaluated from the acceleration
measured by accelerometer D mounted on the specimen (see Fig. 5).
The inertial force is the product of the acceleration and the mass of the
U-shaped head and the part of the specimen that moves. This leads t o
where F;. is the inertial force, a, is the acceleration of the specimen

measured, and rn is the mass of the U-shaped head and the part of the
specimen which moves. This total mass was approximately equal t o
3.97 Ib. After determining the inertial force, the pull-out load can be
obtained by subtracting the value of Fi from the impact force
measured by load cell C.This results in
Pc = Fc - Fi (3)
where F, represents the pull-out load measured using load cell C.It is
A C 1 SP-155 95 0 6 6 2 9 4 9 0523280 T7b W
194 Pacioc and Shah
noted that since magnitudes of Fi are often an order of magnitude

higher than values of P,vaiues of <.should be measured very carefully,
particularly for high impact rates. Since F;- is proportional t o
acceleration as shown in Eq. 2, a decrease in F;. can be achieved by
reducing the acceleration. To decrease the acceleration of the
specimen during impact, a rubber pad of 0.13 in. was placed at the
contact point. To verify that the values measured by load cell A and
B are not affected by inertial oscillations, slip values are compared t o
the double integration of the acceleration. The results are not
significant enough t o take into consideration.
Pull-out forces measured by load cells A, B and C, and the

inertial force calculated based on the measured acceleration from
accelerometer D for t w o specimens with 16 fibers pulled-out a t t w o
--``,`,-`-`,,`,,`,`,,`---
different slip rates are given in Fig. 6. Based on the measurements,

pull-out loads can be evaluated by the t w o previously mentioned
methods. The pull-out loads for these t w o specimens are indicated in
Fig. 7. Values of PABand Pc obtained from both methods are quite
consistent, specially during the prepeak stage.
The experimental technique was checked for specimens with

steel and nylon fibers with very good results. In this paper only the
results for steel fibers pull-out are reported.
Pull-out slip rates were designed to vary from 0.000005 t o 53.5

in./s (from 0.00013 t o 1360 mm/s) as shown in Table 1. Quasi-static
slip rates, ranging from 0.000005 t o 0.00025 in./s (from 0.00013 t o
0.0064 mm/s), were obtained using a closed loop MTS test machine,
whereas higher slip rates, varying from 6.5t o 53.5 in./s (I65t o 1360
mm/s) were obtained using the modified Charpy machine. Two groups
of specimens, series A with 16 and series B with 8 fibers, respectively,
were tested. Since the diameter of the fiber was identical and the
geometry of specimens was the same, both fiber spacing and fiber t o
matrix volume ratio were different in the t w o series. The fiber spacing
was 0.25in. (6.35mm) for series A and 0.50 in:(l2.7 mm) for series
B. Embedded length of 0.5 in. was used for all specimens. Specimens
were cured under water for 56 days, dried at room conditions for 24
hours and then tested. Properties of the cementitious matrix and the
steel fiber used are listed in Table 2.
Specimens with four fiber inclinations of O, 14, 27 and 37

degrees, named AO, A 1, A 2 and A3, respectively, were tested for the
Testing of FRC 195
different loading rates. A minimum of three specimens were tested for

each item listed.
EXPERIMENTAL RESULTS AND DISCUSSIONS

--``,`,-`-`,,`,,`,`,,`---
The observed pull-out load versus slip relationship for a set of

specimens tested at the quasi-static slip rates is'shown in Fig. 8. The
results are given for different inclinations of fibers for the specimens
with 16 fibers as well with 8 fibers.
Initially, the pull-out load increases with the slip almost linearly.
Nonlinearity in the pull-out load and slip curves, which is often
regarded as an indication of propagation of an interfacial crack
(debonding), is usually observed before the peak load. After the pull-
out load reaches the maximum value it decreases slowly with the
increasing slip. Softening type load-slip curves are observed from the
pull-out test. One can observe similar slopes prior t o the peak load for
both series of specimens. However, both the peak pull-out load and the
pull-out load after the peak increase with fiber inclination. The effect
of fiber inclination on the peak pull-out load was theoretically predicted
by Ouyang e t al. [ I 31 using energy considerations.
Since the slip value prior to peak load is relatively small, the
effect of frictional stress on pull-out of inclined metal fibers may not be
dominant before the peak load. The influence of frictional stress
becomes ever more important after the peak' load because more
slipping occurs. This is particularly evident through the pull-out load
and slip curves of series B specimens shown in Fig. 8.
The peak loads and corresponding slips increase with increasing

slip rate for all inclinations tested, as can be seen in Fig. 9 and Fig. IO.
In these plots O degree inclination and 27 degree inclination are being
shown as a representation of the tests. Similar diagrams are observed
for 14 and 37 degrees inclination. [91. The curves presented are the
plots of the individual specimens tested, selected by their being close
t o the average peak pull-out load of the set.
The experimental values of series A and B are listed in Tables III

and IV respectively, where each specimen of series A contains 16
fibers and that of series B contains 8 fibers. In these Tables PaB
represents the total pull-out force of 16 or 8 fibers for the quasi-static
test, FAand i,the t w o components of the pull-out force for the impact
test and p,, the maximum pull-out force per fiber assuming a non
uniform load since different load and slip values from both sides of the
A C 1 SP-155 95 0662949 0 5 2 3 2 8 2 849 W
196 Paclos and Shah
specimen were observed. Considering p,, the average pull-out of a

single fiber, assuming uniform load on all the fibers, the same trend
and relations are still obtained.
Spalling of the matrix at exit point of the fiber was not observed.
However, some radial matrix cracking at the exit of the fiber from the
matrix were detected.
Effect of Loadina Rate

--``,`,-`-`,,`,,`,`,,`---
Fig. 11 shows a plot of the average peak pull-out force per fiber
versus slip rate. Values from aligned and 37 degrees inclined
specimens were selected. Scatter for series B specimens (8 fibers) is
larger than for series A, but still does not mask the trend. 1141.
In series A specimens, p, increases with the slip rate for all

inclinations, having maximum values for 27 degrees. The ratio pmuxd/
p,,," increases for all inclinations, where pmexdis the maximum peak
pull-out force per fiber a t impact rates of loading and p,," is the
maximim peak pull-out force per fiber at 0.000005 in./s used as the
slip rate reference. The trend follows a gradient from O t o 37 degrees.
Fig. 12 shows the effect of the rate of slip on the ratio s/ipd/!/ip"
where slipdis the slip a t the point of p , for the dynamic rates and slip"
is the slip at the point of p,, for the reference rate of 0.000005 in./s.
Results from specimens with aligned and 27 degrees inclined fibers are
represented.
In series B specimens, the slip value a t the point of maximum

load is more sensitive to slip rate as compared t o the series A
specimens. Although the peak slip value increases the most with
increasing slip rate for straight fibers, when the inclination of the fiber
is increased, a sensitivity to slip rate is still observed. For series A
specimens, modifying the slip rate, the slip at peak load is slightly
modified with the exception of 27 degree inclination. The slip a t peak
rate sensitivity is not as high as that for p,, even though missing
data for the 37 degree inclination make it difficult t o confirm the trend.
Effect of Inclination
Fig. 13 shows that increasing the inclination of the fiber, pm,

increases for all rates for series A specimens. The ratio is not constant
but one can still see the trend through the different slip rates. In series
A C 1 SP-155 95 W 0 6 6 2 7 4 7 0 5 2 3 2 8 3 7 8 5 I
Testing of FRC 197
B specimens, p, varies with increased inclination of the fibers too.

The ratio does not show much modification for 0.000005 and 0.00025
in./s 10-3%) but increases between 35-100% for the impact rates.
When the inclination is modified, slip at the point of maximum

load does not change in the same ratio as p,,. The increment in slip
due t o the inclination is smaller for 37 degrees 'in series B, with the
only exception of 6.5 in./s rate. These values can be seen on Tables
III and IV.
Effect of Number of Fibers
--``,`,-`-`,,`,,`,`,,`---
Although the effect of number of fibers on the debonding force
is not fully understood, a certain trend can be observed from the
experimental results.
The values of p,, for series A specimens are slightly more

sensitive than those for series B to rate increment. The same tendency
is also observed for inclination increment.
Series A specimens present a slip value more sensitive t o change

for 2 7 degree inclination. Missing results at high rates and high
inclinations make it difficult to draw conclusions. For series B, slip
increases greatly with rate and show almost no modification or lower
values on the slip when inclination of the fiber is higher.
It should be noted that the unit values of p,, are higher for
series B specimens with aligned fibers. The rest of the inclinations
present higher values for series A specimens, showing a different
effect for impact rates than quasi-static rates.
CONCLUSIONS
The effect of slip rate on pull-out of fibers from a cementitious

matrix has been studied. An innovative system using six sensors was
developed for the Charpy pendulum. This measurement system allows
one t o use t w o independent methods t o evaluate the pull-out load.
Two series of specimens, with 16 and 8 fibers, have been tested and
four different inclinations of the fibers (O, 14, 27 and 37 degrees).
It was experimentally obtained that by increasing the rate of

loading, both pull-out resistance and slip at peak were increased. Peak
pull-out force presents a higher rate sensitivity for a higher number of
fibers. The lower the number of fibers the higher the slip at peak rate
A C 1 SP-155 95 I0662949 0 5 2 3 2 8 4 611 =
198 Pacios and Shah
sensitivity. Regardless of the number of fibers, slip at peak is not as

rate sensitive as peak pull-out force.
Pull-out resistance of inclined fibers is generally greater than that

of aligned fibers. Average peak pull-out load per fiber is higher for
series B specimens with aligned fibers. The rest of the inclinations
present higher values for series A specimens, showing a different
effect for impact rates than quasi-static rates.
A variety of inclinations should be tested t o clarify the trend that

experimental scatter may mask.
Extended work for fiber embedment length should be performed

for inclined fibers t o understand if it follows the.same relationship as
for aligned fibers.
ACKNOWLEDGEMENTS
--``,`,-`-`,,`,,`,`,,`---
Support of National Science Foundation Center for Science and

Technology of Advanced Cement-Based Materials (ACBM) is gratefully
appreciated.
REFERENCES
(1) Gray, R. J., "Analysis of the Effect of Embedded Fiber Length

on Fiber Debonding and Pull-out from a Elastic Matrix," Journal of
Materials Science, Vol. 19, pp. 1680-1691, 1984.
(2) Li, V. C., Wang, Y., and Backer, S . , "Effect of Inclining

Angle, Bundling, and Surface Treatment on Synthetic Fiber Pull-Out
from a Cement Matrix," Composites, Vol. 21, NO. 2, 1990, pp. 132-
140, 1990.
(3)Li, 2,.Mobasher, B., and Shah, S. P., "Characterization of

Interfacial Properties in Fiber Reinforced Cementitious Composites,"
Journal of the American Ceramic Society, Vol. 74, No. 9, pp. 2156-
2164, 1991,
(4) Gokoz, U.N., and Naaman, A. E., "Effect of Strain-Rate on

the Pull-Out Behavior of Fibers in Mortar," The International Journal of
Cement Composites, Vol. 3, No. 3 , pp. 187-202, 1981.
(5) Gray, R. J., and Johnson, C.D., "The Measurement of Fiber-

A C 1 SP-155 95 I0 6 6 2 9 4 9 0 5 2 3 2 8 5 558 I
Testing of FRC 199
Matrix Interfacial Bond Strength in Steel Fiber-Reinforced Cementitious

Composites," Testing and Test Methods of Fiber Cement Composites,
RILEM Symposium 1978, Edited by R. N. Swamy, The Construction
Press, Lancaster, England, pp. 31 7-328, 1978.
(6) Morton, J., and Groves, G. W., "The Cracking of Composites

Consisting of Discontinuous Ductile fibers in a brittle matrix -- Effect of
Fiber Orientation," Journal of Material Science,Vol. 9, pp. 1439-1445,
1974.
(7) Naaman, A. E., and Shah, S . P., "Pull-Out Mechanism in

Steel Fiber-ReinforcedConcrete," Journal of the Structure Engineering,
Vol. 102, NO. ST8, pp. 1537-1541 , 1976.
(8) Brand, A. M., "On the optimal direction of short metal fibers
in brittle matrix composites," Journal of Materials Science,Vol. 20,
1985, 3835-3841, 1985.
(9) Pacios, A. "Measurement of the Fiber-Matrix Interface at

Impact Loading Rates," P.h D. Polythecnic University of Madrid, 1993,
(in Spanish).
(10) Gopalaratnam, V. S., S . P., and John, R., "A Modified

Instrumented Charpy Test for Cement-Based Composites,"
Experimental Mechanics, Vol. 26, pp. 21 7-224, 1986.
(1 1) Suaris, W., and Shah, S . P., "Inertial Effects in the

Instrumented Impact Testing of Cementitious Composites," Cement,
Concrete and Aggregates, ASTM, Vol. 3, No. 2, pp. 77-83, 1981.
(12) Banthia, N., and Trottier, J.F., "Deformed Steel Fiber-

Cementitious Matrix Bond Under Impact," Cement and Concrete
Research, Vol. 21, pp. 158-168, 1991.
(13) Ouyang, C., and Pacios, A., and Shah, S . P., "Pull-Out of
Inclined Fibers from Cement-Based Matrices," accepted for publication,
Journal of Engineering Mechanics, ASCE, 1994.
(14) Pacios, A., Ouyang, C., and Shah, S'. P., "Rate Effect on
Interfacial Response between Fibers and Matrix," accepted for
publication in the Materials and Structures, 1994.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0662949 0523286 4 9 4 =
200 Pacios and Shah
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 75 S O662949 O523287 320 I
Testing of FRC 201
t
2a
-
3
(u
m
-
t
>
hl
a
-
t
>
r
m
-
3,
a
-
t
>
O
m
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 '35 0662747 0523288 267 W
202 Pacios and Shah

--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 I0662949 0 5 2 3 2 8 9 lT3 D
Testing of FRC 203

--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-1.55 95 IObb2949 0523290 915 W
204 Pacios and Shah
Debonding
and slipping
Yielding
of fiber
Fig. l-Mechanisms involved in pull-out of inclined fibers

--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 Ob62949 0 5 2 3 2 9 3 85L
Testing of FRC 205
...
51
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0 6 6 2 9 4 9 0523292 7 9 8 =
--``,`,-`-`,,`,,`,`,,`---
206 Pacios and Shah
f
I

A C 1 SP-155 75 I0662747 0523293 6 2 4 I
Testing of FRC 207
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-I155 95 0662749 0 5 2 3 2 9 4 5 6 0 H
208 Pacioc and Shah
Fig. !i-ûetoils of Chorpy set-up
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-I155 9 5 m 0662947 0 5 2 3 2 7 5 4T7 =
Testing of FRC 209
--``,`,-`-`,,`,,`,`,,`---
1400
1200
1O00
Ki
o 600
400
200
O
O 3 5 8 10 13 15 18
Time (ms)
Fig. &Readout of forces from impact test at two different rates

A C 1 SP-1-55 95 0662949 0 5 2 3 2 9 6 333 M
Pacios and Shah
400
300
200
1 O0
O
O 5 10 15 20 25 30 35 40
Time (ms)
600
500
400
300
200
IO0
O
0.0 2.5 5.0 7.5 10.0 12.5 15.0 17.5 20.0
Time (ms)
Fig. 7-Debonding forces calculated by two methods
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 W Obb2747 0523297 27T W
Testing of FRC 211
0.00 0.01 0.02 0.03 0.04 0.05 0.06

Slip (in)
(a) Specimens with 16 fibers
0.00 0.01 0.02 0.03 0.04 0.00 0.06

Slip (in)
(b) Specimens with 8 fibers
Fig. 8-Effea of inclination on quasi-static pull-out of fibers

--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-255 75 Ob62947 0523278 Lob =
212 Pacios and Shah
O 0.01 0.02 0.03 0.04 0.05 0.06

Slip (in)
(a) Specimens with 16 fibers.
(b) Specimens with 8 fibers.
Fig. %Effect of loading rate on pull-out of aligned fibers
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 II 0662947 0 5 2 3 2 9 9 042 II
Testing of FRC 213
800
700
A
600
0Y
500
400
300
200
1O0
O
O 0.01 0.02 0.03 0.04 0.05 0.06
Slip (in)
(b) Specimens with 16 fibers.
--``,`,-`-`,,`,,`,`,,`---
O 0.01 0.02 0.03 0.04 0.05 0.06

Slip (in)
(b)Specimens with 8 fibers.
Fig. 1 k E f f e d of loading rate on pull-out of 27 deg inclined fibers

A C 1 SP-155 9 5 IObb2949 0523300 674
214 Pacios and Shah
Slip rate (crtús)

1E-5 1E-4 1E-3 1E-2 1E-I 1E+O 1E+1 1E+2
70
300 n
275 t
250 $
225 &
U 200 m
-2 40 175
O
7
ti
? 150 43
7 30
125 y
Q
--``,`,-`-`,,`,,`,`,,`---
Y
1 20 100
75 5
.-E
.i IO 50 .E
i7
H 25 z
O O
1E-6 1E-5 1E-4 1E-3 1E-2 1E-I IE+O 1E+1 1E+2
Slip rate (ink)
Slip rate (crtús)

1E-5 1E-4 1E-3 1E-2 1E-1 IE+O IE+I 1E+2
50
n
E. 200 -
t
175 2
ic
150
TI
c
a
125 0"
c
-
-P 100
3
2
B 20 B
75 $j
a#
..-E 10 50
Q
E
X
25
5 5
O o-
1E-6 IE-5 1E-4 1E-3 1E-2 IE-I IE+O IE+l 1E+2
Slip rate (inls)
Fig. 11-Effect of loading rate on maximum peak pull-out load per fiber
A C 1 SP-1.55 9 5 I0662949 0523303 5 2 0
Testing of FRC 215
4.0 1 111111
- 1~1111111
16 fibers
I 1111111 I 111111 I 1,11111 I 1111111 I
3.5 - t
- + Odegrees
3.0
1 27degrees' ' . I
u> 2.5 - - _.- . _ .

-
.-n -
u< 2.0 - ,-
-
.-n -
u> 1.5 -
- m
1.0 -
-
0.5 -
-
0.0 I I111111 I I111111 I I111111 I I111111 I I I11111 I I IiIiiL
Slip rate (ink)
Slip rate (cm/s)

IE-3 1E-2 1E-1 1E+O 1 E+l 1E+2
8.0
7.0
6.0
5.0
-
u>
.-n
4.0
-
.-o
3.0
2.0
I.o
0.0
1E-4 1E-3 IE-2 1E-I 1 E+O IE + l 1E+2
Slip rate (ink)
Fig. 12-Effect of slip rate on ratio slipd/slip'
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 75 = 0662749 0523302 467 =
216 Pacios and Shah
90.0 400
h
80.0 350 z
à5
LI
70.0
300
60.0 Q
250 :
-
O
50.O
40.0
30.0
20.0
10.0
0.0 O
O 5 10 15 20 25 30 35 40
8 (grades)
70.0
300
h
60.0 tL
250 4
IC
50.O L
aJ
Q
200 u
m
40.0 -
O
30.0
Y
100
20.0 Q
10.0 50
6
.-E
X
2
0.0 O
O 5 IO 15 20 25 30 35 40
--``,`,-`-`,,`,,`,`,,`---
8 (grades)
Fig. 13-Effect of inclination on maximum peak pull-out load per fiber

A C 1 SP-155 95 0662749 0523303 3T3
SP 155-11
Testing Applied to the Evaluation of

Damage to FRC and Other Material
Systems Caused by Large Missile impact to
Building Envelopes During Storm Events
by C. D, Hays and R. F. Zoll0
--``,`,-`-`,,`,,`,`,,`---
SvnoDsis: Recent natural disasters involving high wind
events have demonstrated the fact that building
envelopes, including structural walls and roofs, can lose
structural integrity as a result of penetration by
missile objects. As a result, there is heightened
interest in the testing of components and cladding that
are used as a part of building envelopes of habitable
structures.
A large missile impact test has been designed and is
being evaluated in laboratories around the country. The
test, discussed in this paper, is suitable for laboratory
or field applications and is currently undergoing
scrutiny by the ASTM Task Force of Committee E6,
Performance of Buildings. Adoption of the test by the
South Florida Building Code came in the wake of Hurricane
Andrew in 1992.
The test has been applied to numerous types of wall
systems and building products including a fiber
reinforced cellular concrete panel which is designed to
be used as an alternate to masonry infill construction,
architectural precast, demising walls and security
fencing. Additional tests of the missile impact
resistance of fiber reinforced cellular concrete
involving the use of large caliber ballistics are also
discussed. The high energy impact resistance of fiber
reinforced systems is demonstrated and discussed.
Keywords: Cellular concretes; failure; fiber reinforced concretes; impact

resistance; impact tests; structures

Not for Resale
A C 1 SP-155 75 Obb2747 O523304 23T =
218 Hays and 20110
AC1 member carol D. Hays is a registered Professional

Engineer and President of Engineering Analytics, Inc.,
Miami, Florida. She earned an M.S. degree in Civil
Engineering at the University of Miami, Coral Gables,
Florida. She currently serves as a member of AC1
Committee 544, Fiber Reinforced Concrete.
AC1 member Ronald F. Zollo, P.E. is Professor of Civil
and Architectural Engineering at the University of Miami,
Coral Gables, Florida. Dr. Z o l l o serves on AC1 Committee
544, Fiber Reinforced Concrete, currently serving as
Chairman of the subcommittee on Synthetic Fibers, and AC1
Committee 549, Ferrocement. He is currently a member of
the Dade County Building Code Committee which was formed
to institute changes in the South Florida Building Code.
INTRODUCTION
On August 24, 1992 Hurricane Andrew made landfall in

southern Dade County, Florida and moved on a westward
course across the southern tip of the peninsula. The
storm reached a Category 4 on the Saffir-Simpson scale
with fastest mile wind speeds of approximately 145 mph at
the point of landfall. The resulting damages to
residential and commercial structures exceeded an
estimated value of 20 billion dollars.
A common cause of the loss of structural integrity of
roof and wall systems, often leading to complete
structural collapse, was attributed to penetration of the
building envelope by wind borne debris. Rapid and high
level increase in pressurization of the interior of
buildings can occur as a result of relatively small area
penetrations of the building envelope. (i) This problem
had received little attention as a design consideration
in the United States prior to the recent hurricanes Hugo
and Andrew. Until recently, the assumption of an
enclosed structure and no internal pressurization was
--``,`,-`-`,,`,,`,`,,`---
normally applied. The fact is that the problem of wind

borne debris penetration is widespread, and that this is
a threat posed by Atlantic, Gulf and Pacific spawned
hurricanes as well as by tornadoes that occur throughout
the midwestern and southeastern regions of the United
States.

A C 1 SP-155 9 5 II Ob62949 0523305 L7b I
Testing of FRC 219
MAINTENANCE OF STRUCTURAL INTEGRITY
The concern for the life safety of building inhabitants

dictates that special emphasis be placed on maintaining
an unbreached building envelope during high wind events.
The photographs of Figures 1 and 2 were taken in the
aftermath of Hurricane Andrew (2). They depict,
respectively, the penetration of a wood frame building
envelope by a roof sheathing element and an indication of
the kind of damage that can result from sudden internal
pressurization of building structures.
Research and wind force load calculations performed in
accordance with Chapter 6 of ASCE 7-88, ( 3 ) Minimum
Desiqn Loads for Buildinqs and Other Structures,
indicates that an opening of as little as five percent
--``,`,-`-`,,`,,`,`,,`---
(5%) of the windward wall area shall be considered as

resulting in high level pressurization of the interior of
the building. This internal pressurization adds to
uplift forces on roof elements and horizontal forces on
leeward and side wall elements.
A key recommendation in the final report of the Dade
County Building Code Evaluation Task Force (4) was that
all of the components of exterior walls, such as
sheathing, cladding, glazing, doors and windows, be
designed and constructed to resist penetration of
building envelopes caused by impact loads from wind borne
debris. The only permissible alternative requires that
fenestration products, ie., coverings for wall openings,
be protected by shutter systems which are themselves
capable of resisting penetration by wind borne debris.
Following this recommendation as a guideline, the South
Florida Building Code (SFBC) (5) adopted what has been
termed a large missile impact test standard. This
standard applies to any portion of the building envelope,
eg., shutters, windows, doors, walls or roof, in order to
gauge penetration resistance to wind borne debris. The
large missile impact test standard also includes
requirements for multiple impacts and cyclic fatigue
loading.
THE SFBC LARGE MISSILE PROTOCOL (6)
The SFBC large missile impact test procedure is entitled

Protocol PA 201-94 and is briefly described as follows.
The procedure requires that all parts of a test specimen
be full sized and be constructed using materials and
methods of construction that are the same as that
proposed for the actual application. The large missile
is a solid S4S (surfaced four sides) nominal 2x4 82
A C 1 SP-155 95 111 0662949 0523306 002
220 Hays and Zoll0
surface dry Southern Pine between 7 and 9 feet (2.13 and

2.74 m) long, as shown in the diagram of Figure 3. A
sabot, which is attached to the trailing edge of the 2x4
to facilitate launching the missile, cannot weigh more
than 0.5 pounds (0.27 kg). The combined weight of the
2x4 and sabot must be between 9 and 9.5 pounds (4.08 and
4.31 kg).
In current applications the missile is propelled through

a large missile cannon using compressed air as the means
of propulsion. Figure 4 depicts one such cannon'
mounted on a transportable platform and set up for
laboratory use. According to the standards presently
approved for Dade County, Florida, the cannon must be
capable of producing impact at 5 0 feet per second, fps,
(15.24 m/s).
The cannon assembly has four major components: compressed

air supply, pressure release valve, pressure gauge, and
barrel and support frame, as shown in the schematic of
Figure 5. A system for determining the missile speed
usually consists of light beam triggers and video camera
event markers. The barrel consists of a 4-inch (10.16
cm) inside diameter pipe which must be at least as long
as the missile. The barrel is mounted on a support frame
in a manner that facilitates aiming at specified target
locations on the test specimens. The cannon is fired at
a range such that the distance from the end of the cannon
to the specimen is 9 feet (2.74 m) plus the length of the
missile (also approximately 9 feet (2.74 m)).
The protocol requires that each test specimen receives
two impacts. The first impact is to be within a 5 inch
(12.70 cm) radius of a point having its center on the
midpoint of the test specimen and the second to be within
a 5 inch (12.70 cm) radius of a point having its center
6 inches (15.24 cm) from any supporting member at a
corner of the specimen. Deflections must be considered
for qualifying external protection devices, ie.,
shuttering systems, but not for actual fenestration
devices. The standard requires that any shutter system
which is designed to protect a fenestration device must
not deform to within 1 inch (2.54 cm) of the device.
'The authors are grateful for the cooperation of

American Test Lab of Florida, Inc., located in Pompano
Beach, Florida.
--``,`,-`-`,,`,,`,`,,`---

Testing of FRC 221
OTHER CODES AND STANDARDS
A similar test procedure is, as of this writing, under

review by committee of the Southern Building Code
Congress International under the title IISBCCI Standard
for Windborne Debris Impact Tests - SSTD 11-93". Also,
the American Society For Testing and Materials, ASTM, has
formed task group E06.51.17, Task Group of Committee E6
Performance of Buildings, to consider a test
specification which is similar to the SFBC protocol as
well as a pendulum type impact test system.
APPLICATION OF THE LARGE MISSILE IMPACT TEST
The large missile impact test has now been applied in

Dade County Florida to conventional wood and masonry wall
systems, to several developing and often considered
alternative wall system technologies, and to building
fenestration products and protection devices of great
variety. The performance of portland cement based fiber
reinforced products is of particular interest not merely
as a means of evaluating the performance of particular
applications of FRC but also in demonstrating the well
known ability of FRC systems to absorb impact energies.
FIBER REINFORCED CELLULAR CONCRETE (FRCC) (7)
A relatively new FRC material formulation has been tested

against the standard of the impact test protocol. The
material is a composite of synthetic fibrous
reinforcement and cellular cement, often termed cellular
concrete where an air void system is considered as the
aggregate phase. The resulting product is referred to by
the acronym FRCC for Fiber Reinforced Cellular Concrete.
Presently FRCC is produced in a factory precast
operation2 providing panels which are designed to be used
as an alternate to masonry infill construction,
architectural precast, demising walls and security
fencing. Efforts are currently underway to develop a
panel system which may be used as an alternate to joist
and deck roofing systems.
The FRCC panels which have received Dade County product
control approval are used as part of the building
2FRCC panels are produced and supplied by

Thermoflex, Inc., Miami, Florida.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 I0662949 0.523308 985
222 Hays and Zoll0
envelope in residential construction in South Florida for

exterior walls. The approved panels are typically
produced as 4 by 8 foot (1.22 by 2.44 m) nominal width
and height dimension and 7-5/8 inch (19.37 cm) thickness
and are depicted in the photograph of Figure 6.
A slurry mixture of ASTM Type I portland cement concrete,
polypropylene fibers (1% by solid volume), and a
proprietary formulation of mineral admixtures are batch
cast into steel molds. The batch is placed so as to fill
approximately 50 percent of the mold and a volume
expansion takes place which fills 100 percent of the mold
over a period of approximately four hours. The process
is a llcoldgg process in contrast to the well known
autoclave process for producing cellular concrete which
is inherently more costly and prohibitive of the use of
synthetic fibers. The resulting composite material is
relatively lightweight, having a density between 35 to 40
pcf (560 to 640 kgjcu m), and exhibits considerable
strength and toughness as demonstrated by its performance
against the impact test standard.
IMPACT RESISTANCE OF FRCC
In the development stages of the Dade County impact test

protocol a projectile velocity of 80 fps (24.38 m/s) , or
approximately 55 mph, one half of the ASCE 7-88 fastest
mile design wind speed for the South Florida region, was
specified. Preliminary testing indicated that at this
--``,`,-`-`,,`,,`,`,,`---
impact velocity conventional concrete block construction
could not resist penetration of the missile prompting the
eventual reduction of the missile impact velocity
specification. The photograph of Figure 7 depicts a
concrete block wall which was subjected to missile impact
at the 80 f p s (24.38 m/s) speed.
FRCC wall panels were tested prior to the reduction of
the 80 fps (24.38 m/s) missile impact speed. The panels
exhibited little damage and complete resistance to
penetration. Full scale panel prototypes were 7-112
inches (19.05 cm) thick, 8 feet (2.44 m) in length, 4
feet (1.22 m) in width and were simply supported over the
long span against a rigid wall. The photograph of Figure
8 shows the indentation made by the standard missile with
area dimension at approximately the same size as that of
the dimensional lumber which is the missile. Two missile
impacts within approximately 12 inches (30.48 cm) of each
other are shown. Penetrations extended to a depth of
approximately 314 inches (1.91 cm) A closer view of one .
of the indentations is depicted in the photograph of
Figure 9.

A C 1 SP-155 95 0 6 6 2 9 4 9 O523309 611 .
I
Testing of FRC 223
Inspection of the rear face of the impacted panel

evidenced only little damage in the form of hairline
surface cracks directly behind the impact locations. The
results of these tests demonstrate the well known
toughness or energy absorbing characteristics associated
with fiber reinforced materials and prompted additional
testing.
FRCC RESISTANCE TO BULLET PENETRATION
The energy absorption and penetration resistance

capability of FRCC panels is also demonstrated in tests
conducted in which large caliber hand gun bullets were
fired at the panels at close range. The photographs of
Figures 10 through 13 show the effects of 44 caliber and
9 mm bullets when fired into an 8 inch (20.32 cm) thick
FRCC panel. Core samples were cut perpendicular to the
panel face and through the entire thickness over bullet
entry points. These cores were partially cut and split
open to expose the penetration trajectories. The
photograph of Figure 10 shows the trajectory of a 44
caliber magnum bullet which was fired at the wall at a
range of 20 feet (6.10 m) . The bullet penetrated
approximately 213 through the wall thickness. The
photograph of Figure 11 is a close up view of the same 44
magnum bullet trajectory as in the previous figure. The
Figure 12 depicts a second core sample taken from the
wall at a location over the entry of a 9 mm bullet also
fired at the range of 20 feet (6.10 m) . The 9 mm bullet
penetrated approximately 112 of the wall thickness and is
shown in close up in the photograph of Figure 13.
SUMMARY AND CONCLUSIONS
The criteria for the design of structures should not be

limited to consideration of the main wind force resisting
system in response to wind loading. It has been shown
that the quality and integrity of the building envelope
has a major influence not only on the performance of the
structural envelope but also on the safety of building
inhabitants and the degree of damage to interiors and to
personal property. The impact test adopted by Dade
County is, if nothing more, a beginning in the effort to
include impact and cyclic loading penetration resistance
in the design criteria for any wind threatened
construction.
The interest in this problem is, however, not limited to
Dade County, or to Florida, or to hurricane generated
--``,`,-`-`,,`,,`,`,,`---
wind. The concept, at various levels of intensity

A C 1 SP-355 95 IObb2949 0 5 2 3 3 3 0 5 3 3 W
224 Hays and Zollo
applies also to tornado, severe thunderstorm, Santa Ana

winds, chinook winds and nor' easters. In other words
there should be widespread interest not only by
governmental building administrators but also by design
professionals, product purveyors, property owners and
insurers.
At the forefront of the design implementation is the
design testing and adoption of design standards. What
has been described herein is one test method of the many
possible alternatives which could have been considered
--``,`,-`-`,,`,,`,`,,`---
but one which has gained interest and acceptance based on
relative simplicity and economics as well as its
suitability to the task.
Fiber reinforced concrete has long been known for its
impact and cyclic fatigue loading toughness
characteristics. The large missile impact test well
documents and is aptly applied to the testing of FRC
building wall systems. The resistance to bullet
penetration by FRCC wall panels has also been
demonstrated.
REFERENCES
1. Minor, J.E. and Behr, R.A., "Architectural Glazing

Systems in Hurricanes: Performance, Design Criteria
and Designstg,7th U . S . Nat'l Conf. on Wind Eng.,
Univ. of California, L o s Angeles, June 27-30, 1993.
2. Zollo, R. F., etal., "HURRICANE ANDREW: August 24,
1992 Structural Performance of Buildings in Dade
County Florida" Technical Report #CEN 93-1, Dept.
of Civil and Architectural Engineering, University
of Miami, Coral Gables, FL, March 11, 1993, pp 73.
3. "Minimum Design Loads for Buildings and Other
Structures", ANsI/ASCE 7-88, American Society of
Civil Engineers New York, NY, July 1990.
4. Final Report of the Building Code Evaluation Task
Force, Dade County, FL, December 1992.
5. South Florida Building Code, 1988 Edition,
Metropolitan Dade County, Miami, FL, 1988.
6. PROTOCOL PA 201-94, Dade County Building Code
Compliance Office, Miami, FL.
7. Zollo, R . F . , and Hays, C.D., "A Habitat of Fiber
Reinforced Concrete," Concrete International, ACI,
Vol. 16, No. 6, June 1994, pp 23-26.
A C 1 SP-155 '35 Obb2949 0523311 4 7 T
Testing of FRC 225
Fig. 1-Penetration of building envelope by windborne debris
Fig. 2-Destruction of building envelope by infernal pressurization
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0662949 0 5 2 3 3 3 2 306 m
226 Hays and Zoll0
2 x 4 DIMENSIONAL LUMBER
Fig. &Large missile (2 x 4)
Fig. &Large missile cannon - Laboratory assembly
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 9 5 I0662949 0523313 2 4 2 I
Testing of FRC 227
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-II55 95 = 0bb2949 OS23334 II89 =
--``,`,-`-`,,`,,`,`,,`---
228 Hays and Zoll0
Fig. 6-Typical FRCC panel
Fig. 7-lmpaa test of concrete block wall

A C 1 SP-355 95 R 0 6 6 2 9 4 9 0523335 035 I
Testing of FRC 229
--``,`,-`-`,,`,,`,`,,`---
Fig. 8-Impact test of FRCC panel
Fig. 9-Close view of missile damage to FRCC panel

A C 1 SP-155 7 5 I0662947 0523316 T5L
--``,`,-`-`,,`,,`,`,,`---
230 Hays and Zoll0
Fig. lû-Trajectory of 44 caliber bullet in FRCC panel
Fig. 11-Close view of 44 caliber bullet

A C 1 SP-355 95 II O b 6 2 9 4 9 0523337 9 î ô
Testing of FRC 231
Fig. 12-Trajectory of 9 rnrn bullet in FRCC panel
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 0 6 6 2 9 4 9 0523318 824
232 Hays and Zoll0

--``,`,-`-`,,`,,`,`,,`---
Fig. l w l o s e view of 9 rnrn bullei

A C 1 SP-155 95 I0662749 0523319 760
SP 155-12
Fiber Reinforced Concrete Testing

for Practical Application
by G. Spadeo, R. Cavo, D. Gallo, and R. N. Swarny
SvnoDsis: One of the possible factors inhibiting the wider application of fiber
concretes is the fact that to characterize the engineeringproperties of fiber concrete,
both cubedcylindm and prism have to be cast and tested, in addition to the
determination of workability. This paper shows that with the use of superplasticizer,
the slump test can be used to give guidance on the flowability characteristics of the
fresh fiber concrete. The advantage of the slump test is that it is easy to carry it out
in the field, apart fim its simplicity and convenience. It is further shown that the
--``,`,-`-`,,`,,`,`,,`---
equivalent cube strength obtained from the broken pieces of a flexural test can
adequateiyrepresent the compressivestrength of fiber concretes. It is thus shown that
it is possible to characterize the engineering properties of fiber concretes from oniy
onesetofprisms, say, of lOOxlOOx5OOmmsize. Aparîfrombt crackload, moduius
of rupture and fracture toughness properties, the prisms can be used to give additional
information as appropriate such as shrinkage and expansion, and through puise
velocity, internai microcracking. It is suggested that by rationalizing our approach to
testing, it is possible to reduce not oniy the cost and inconvenience associated with
Werent sizes of test specimens but also to enhance the relevance of some of the
information obtained from such testing.
Kevwords: Compression tests; compressive strength; expansion; fiber

reinforced concretes; fibers; flowability; microcracking; shrinkage; silica
fume; steels; &
e
j

233
Not for Resale
A C 1 SP-155 95 0662947 0523320 482 H
234 Spadea et al
Prof. Giuseppe Spadea received his degree at the University of Pira, Italy.
Presently he is Associate Professor of Civii Engine- wiîh inîerest in design and
analysis of structures at the Department of Structural Engine- University of
Calabria, Italy. His research interests are experimental characterization of structural
materials and modehg of the behavior of reinforced concrete and FRP structures.
Rosaiba Cava is a PhD candidate at the University of Calabria, Italy, where she
earned her degree in structural engineering. Her research interest is in the area of
fiber reinforced concrete.
Donatella Gallo is a PhD candidate at the University of Calabria, Italy, where

she earned her degree in structural engineering. Her research interest is in the area of
fiber reinforced concrete.
AC1 Fellow Prof. R Narayan Swamy is at the Structural Integrity Research

Institute and the Centre for Cement and Concrete, University of Sheffield, England.
His research interests are in concrete materials and concrete structures. He received
the George Stephenson gold Medal of the Institution of Civii Enguieers and the
Henry Adams Diploma from the Institution of StructuralEngineers. He is Editor of
Cement and Concrete Composites and the Book Series on Concrete Technology
and Design. He recently received the CANh4ETíACI Award for Conhibutions to
Durabiìity of Concrete.
INTRODUCTION
It is no longer necessary to extol the economic, technical and performance

advantages of fiber reinforced cements and concretes (FRC). Durmg the last three
decades extensive research and developmentai work, together with a wide range of
practical appiicaîim, have confirmed that fiber cement composites are reliable and
efficient and that they can be cost-effective construction materiais when used
intelligently. In spite of the proven characteristics of FRC, however, the usage of
fibers in practice is reiativeiy low compared to normal and other special concretes.
Part of the blame for this state of affairs lies in current test methods and the h g e
number of tests additionai to cube or cylinder compressive strength thaîneeds to be
undertaken morder to charactexk the material. Indeed for a g k n practicalapplication
one is not often sure what tests need to be canied out to establish the engheering
properties of FRC, although extensive iiterature is available on this topic (16).
One can perhaps illustrate this predicament when an ordinary non-specialist

--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 9.5 0662947 0 5 2 3 3 2 1 319
Testing of FRC 235
practicalapplication such as highway pavements is considered. What type and amount

of fiber is to be used? Since fibers affect adversely the flowabiiity properties of the
fiesh concrete, should one use a superpiasticuer (SP) to enhance workability? What
is the amount of SP to be added? Indeed, since much of the iiteraîure data are
related to mortars rather than concretes, what difference does it make between
workabiiity characteristics of cements and mortars, as against those of concretes?
How do non-metallic fibers affect flowabiiity properties compared to steel fibers?
Can one use the conventional slump test to indicate the flow properties of FRC?
How do we match the cost between cost of testing, cost of materials and acceptable
needed properties in the fresh and hardened states? if one wishes to use silica fume
(SF) to enhance durability against deicing salts, how does the presence and amount
of SF d u e n c e the engineeringproperties in the fresh and hardened states? It can be
readiiy seen that there are several aspects of testing which impinge on the use and
appiications of FRC in practice.
There are also other quations related to the relevance of some of the recammended
test methodologies to characterize fiber concrete. Two examples are quoted here.
One relates to the so calied workability of fiber concrete. Whilst it is universaiiy
recognized that the conventionalsiump test gives a useful measure of the consistency
of concrete (without fiber), which forms oniy one aspect of the complex property of
workabiiity, there is considerable evidence to show that the m-ed versions of
tests for Cesh ñber concreteworkability &o do not give a clear idea of their flowability
properties; indeed they show h@iy variable resuiîs and exhibit no consistentrelations
between the various measured values (7-12). Indeed, the most imporîant Equirement
of fiber reinforced concrete, namely, the uniform distribution of fibers w i t h the
body of concrete, is not met or measured by any of the current test methods on fresh
fiber concrete. The second relates to toughness indices. Here again there is widespread
scepticism on the relevance and usefulness of the various toughness indices for
practical applications, although this aspect is not discussed in this paper.
The aim of this paper is to tackle some of the problems related to testing for FRC
discussed above. In particular, the following aspects are considered.
1. What is the amount of SP required to obtain good flowability properties for

--``,`,-`-`,,`,,`,`,,`---
steel and polyacryloniûiie (non-metallic fibers) when used in concrete without and
with SF?
2. Can the conventional slump test (which is universaiiy recognized as a useful

measure for the consistency of concrete but not a proper measure for workabiiity)
be used as an indicator of the flow properties of FRC for piacement in practical
applications?
3. Can ihe cubdcyiindertest for compre& strength be dtspehsedwith andreplaced

by the quivaient cube strength obtained h m the broken pieces a k a flexurai tesí?
A C 1 SP-155 9 5 IObb2949 0523322 255
236 Spadea et al
4. Can prisms (for example, 10Ox1OOx500 m m size) be used as the sole test
specimens cast to obtain the engineering properties of FRC? To illustrate this,
shrinkage, expansion, pulse velocity and i6rst crack load in flexure are reported.
Since flexural test specimens are normally used in any case to obtain moduius of
rupture and fracture toughness parameters, these characteristics are not reported
here.
This paper thus shows that apart from the test for the flowability properties of
FRC,10Ox10Ox500mm prisms alone need to be cast to characterize the engineering
properties of FRC for practical appiications.
Aithough laboratory research reported on FRC is extensive and voluminous, the

practical use of fibersin concrete is relativeiy insignificant compared to the amount
of test data availablein literature. This anomalous situationis pady do to the Wërent
types of test specimens that need to be used and the Werent types of tests that need
to be canied out to supposediy characterize the engheering properties of FRC.
Indeed the relevance of some of this tests and the usefulness of the infomiation
obtained fomi them have also been questioned.
The aim of this paper is to show that it is possible to characterize the engineering
properties of fiber concretes by using oniy one set of test specimens i.e. prism, say,
of 1OOx1OOx500 mm size. Further, the slump test can be used to give practical
information on the flowability properties of fiber concrete, particularly when
superpiasticizers am incorporated, although one recognizes that by the very nature
of the slump test, it cannot give any directiy measurable quality of workabiliíy.
--``,`,-`-`,,`,,`,`,,`---
Materiais
Rapid hardening portland cement was used for all mixtures. The minimumvahies
of flexural and compressive strength required by Italian Standards for this type of
cement are respectively 6 MPaand 32.5 MPa at 7 days and 7 MPaand 42.5 MPa at
28 days. Saca fume (SF) was used as partiai repiacement of the portland cement on
an equal-weight basis.
The fine aggregate used was river sand with a maximum size of 1.5 mm and a
A C 1 SP-155 95 0662949 0523323 191
Testing of FRC 237
water absorption ot 1.58 per cent. i he coarse aggregate was washed and dried
crushed gravel mostly of rounded shape, with a maximum size of 10 mm and a
water absorption of 0.54 per cent.
Hooked-end steel fibers glued together in bundles were used. These fibers
had a nominal length of 30 mm and a diameter of O. 5 mm (aspect ratio 60), with
a specific gravity of 7.1 g/cm3and an elastic modulus of 210 GPa. The
polyacrylonitrile fibers used in these tests were a high modulus synthetic fiber
with a kidney-shaped cross-section. These fibers are bound together in sheaves
by a special water-soluble sizing agent which then softens during mixing releasing
the singie fibers. Polyacrylonitrilefibers were 24 mm long with a nominal diameter
of 16 pm, a specific gravity of 1.18 g/cm3and an elastic modulus of 13.5 GPa.
A superplasticizer (SP), in the f o m of an aqueous solution, was incorporated

in the concrete mixtures in varying amounts depending on the fiber type and
content and SF. It was added at a rate which would produce a good or reasonable
workability depending on the constituents, and a cohesive, flowable matrix, and
in practice, this amounted to O. 8 to 3.0 per cent by weight of binder.
Mixtures
--``,`,-`-`,,`,,`,`,,`---
Two groups of mixes totalling eight were investigated, as shown in Table 1.

These mixes were proportioned to give a compressive strength of about 45 MPa
at 28 days. For the first group a concrete matrix with mixture proportions by
weight of 1:2.2:1.9:0.45 (cement:íine aggregate:coarse aggregate:water) was
used. The water-binder ratio was kept constant for all the mixes. The fiber
reinforcement was formed by steel fibers with a volume percentage of 0.5 and
1.0 and by polyacrylonitrile fibers with a volume percentage of 0.13 and 0.25.
For the second group the concrete matrix was modified with 5.0 per cent of
the cement replaced by s i k a fume. The two types of fibers, steel and
polyacrylonitrile, were used with a volume percentage of 1.0 and 0.25,
respectively.
Fabrication and Curinp of Test Specimens
From each mixture six 100x100x500 mm prisms and at least three 150
m m cubes were cast in steel and polystyrene molds respectively.
The mixtures were prepared using a conventional mixer with a maximum

A C 1 SP-1.55 75 Obb32947 0 5 2 3 3 2 4 Oil8
238 Spadea et al
capacity of 0.35 m3. The mixing procedure was as follows: dry coarse and fine
aggregate were homogenized together and mixed with their absorbed water to
reach a saturated surface-dry condition; after that cement, silica fume if present,
and miXing water were added. Steel fibers, superplasticizer and polyacrylonitrile
fibers were haüy introduced in the mixture. The fibers were gradually scattered
by hand in the matrix. in ail fiber concrete mixtures, care was taken to ensure
that the fibers suffered as iittle damage as possible consistent with thorough
mixing and uniform distribution of fibers. Based on the experience acquired
during the fabrication of trial mixtures, the polyacrylonitrile fibers were added
right at the end of the mixing process when the matrix had been thoroughly
mixed into a homogeneous and flowable mixture because otherwise balling of
fibers could be observed. The specimens were cast in the molds and compacted
on a vibrating table. For each mixture of the first group two batches were
successively prepared.
Specimens were cured by covering them with a plastic sheet while they were
sídi inside the molds. The specimens were demoulded after 24 hours and then
exposed to two curing regimes:
-
Continuous water curing in uncontrolled internal environment.
-
Seven days water curing followed by exposure to uncontrolled internal
environment.
During the curing regimes the temperature and relative humidity variations
were not recorded.
Testine Details
Flowability - whilst fully recognizing that the conventional slump test has
little direct scientific significance to the property of workability, with a
superpiasticizerthe test can give useful practical information to the mixdesigner.
This test was therefore adopted in this study according to B.S. 1881:Parî 2.
-
Compressive S t r e d Compressive strength was determined both from
cubes and the equivalent cubes of the broken parts of the flexural test prisms.
Three cubes and six equivalent cubes were tested for each mixture.
Free Shrinkage and Expansion Measurements Free shrinkage and expansion --

measurements were made using a mechanical extensometer (Demec) of 100
m m gage length. After 24 hours from casting, the molds were removed and
demec points were carefully fixed at selected positions on ali four faces of each
specimen. The trowelled surface was ground to remove heguiarities so that the
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 m Obb2949 0523325 Tb4 m
Testing of FRC 239
two discs were as far as possible in the same plane. The first readings were taken
two hours after the fWng of the discs. Regular measurements were then
-
subsequentlytaken for the duration of the tests every day up to 28 days, then
twice a week for 8 weeks and then once every week up to about 200 days. The
shrinkagelexpansion value of each specimen was an average of the readings on
the four faces. The development of shrinkage and expansion was calculated
treating the fist reading as the initial reading (zero reading). For each mixture
six specimens were tested.
Pulse Velocitv Measurements-- Ultrasonic pulse velocity technique was used

to monitor the changes in the structure of the concrete which occur with age.
Regular readings of longitudinal pulse velocity were obtained setting transmitting
and receiving transducers in the middle of the two terminal sections of the
specimen. For the water-cured condition, the measurements were made on the
specimens after drying the surfaces. The ulîrasonic technique was also used to
detect the onset of first cracking during the flexural tests. The transducers were
attached to the test specimen near the extreme tensile fiber held in position by a
right wood frame. In both cases the transducers used had a frequency of 55
kHz.
Flexural Test -- AAer the long term exposure, the specimenswere tested under four-
pointbending overaspanof4ûûmm Dunngthetesiaconstantcross-hadrateofO.05
d m i n was kept. Deflections were measured at the midspan section by means of two
LWTs witha m;urimUmtrawl of 20 mm. The specimensm d in water were prepared
for bending test mthe fobwingway: after takangthem out of waterthey were died with
a paper and wex left mthe laboratmy for about three hours before testing. The resulis of
this patt of the study are not reported here except for the fmícrack load
Equiwient Cube Strennth -This property was obtainedby canying out an equivalent
cube test on the broken p& of the prism specimens afìer they had been tested m
bending.This was done by locating square steel plates, equident m section to the cross-
sectionoffheprism,beîweenMe~andtestingnaachinebtransnatthecompressniP:
load@& 1).
TEST RESULTS AND DISCUSSION

--``,`,-`-`,,`,,`,`,,`---
The resutoS of the tesis on fie& concrete flowabiiitypr@ are reportedin Table
1. It shouid be mentioned here that duringthe siump tests, the behavior of themixes was
carefülly monitored in terms of ability to flow,segregation of constituents, ability

A C 1 Sp-155 95 0662949 052332b 9 T O
240 Spadea et al
to remain cohesive and homogeneous, and ability of the fibers to separate and
remain embedded in the cement matrix. The discussion below is based on these
close observations. The results in Table 1on slump measuremenis have many practical
Unpiications. They show that whilst a good flow properties can be obtained with a
relatively high volume of steel fibers, the quantity of superplasticizer needs to be
increased as the fiber volume is increased. From the close test observation, a slump
of about 100 mm is considered to be acceptable to ensure good flowabiiity in the
field to ensure proper compaction.
A non-metallic fiber like the polyacrylonitde fiber used here required tugher
amounts of superplasticizer even for low fiber volumes: even then, the slump values
--``,`,-`-`,,`,,`,`,,`---
obtained were less than those of comparable steel fiber concrete mixtures. Further,
even with high superplasticizer contents, oniy slumps of 30-50mm were obtained
with the poiyacrylonitriie fiber. It was considered that about 3 per cent by weght of
cementitious content is about the maximum acceptable superplasticizer content both
ftom cost point of Mew, and the avoidance of secondary effects such as increased
setting times and segregation.
The use of silica h e , without and with fibers, also aáverseiy affected the
workability properties of the mixtures. The presence of silica fume produces an
increase in the fines content, and hence an increase in water demand. Even with
increased amounts of superplasticizer, slumps of oniy about 80 mm and 25 mm
were obtainable for 1.0 per cent and 0.25 per cent steel and poiyacrylonitriiefibers
respeckly. The silica fume mixtures, on the other hand,were found to be cohesive,
and indicated that slump values do not always provide an accurate indication of the
flow properties of concrete mixture, whether it contained fibers or not.
The results in Table 1 emphasize the need of technical compromises in the use
and appiication of flowable and workable fiber concrete in practice - between cost-
effectiveness, workability required and/or obtainable, acceptable amounts of
superplasticizer without side effects on other properties such as setting times and
segregation, types and amounts of fibers, amount of silica fume and the engineering
properties required for a specific use. There is the further point that these factors
need to be considered even at the mix proporîioning and fabIication stage, as these
would d u e n c e the subsequent períormance of the material in situ (13).
One of the aims of this stuây was to find out if the conventionai slump test can be
accepted as an adequateindicator of flowabiiityof fiber concretemixes when different
types and amounts of fibers and silica fume are used as additionai constituents. The
resuits show that provided a reasonable dump is designed for (say about 50 to 120
mm), and a superpiasticizer is incorporated, the slump test could be a good field
gui& ofworkabilitypropertiesfmpracíicai appiications. The results ofthe dump tests
obtained in this study confirm the conclusions arrived by Popovics on his analysis of
slump test data reported recentiy (14). The authom of this paper umcur with these

A C 1 SP-155 75 0662747 0 5 2 3 3 2 7 837
Testing of FRC 241
tindings and suggest that the slump test, although not scientifically based, can give a
lot of useful information to the practical concrete technologist, and that it is a simple,
convenient and cost-effective method of quality control of fiber concrete in the
field.
Silica fume is nowadays widely used for enhancement of both strength and
durability. However, when fibers and SF are both incorporated, serious problems
of lack of flow properties could arise without a superplasticizer. Even when a
SP is used, the amounts required could be quite high: in the tests reported here
2 to 3 per cent by weight of cementitious content was required to obtain slumps
of 75 to 85 mm for steel fibers and 20 to 30 mm for polyacrylonitrile fibers for
moderate volumes of fibers. The need to proportion fiber concrete mixtures
bearing these factors in mind cannot be overemphasized, and perhaps these
problems highlight why fiber concrete mixtures are still not as widely accepted
--``,`,-`-`,,`,,`,`,,`---
as their benefits from laboratory tests would probably indicate.
ComDressive Strenpth
Table 2 shows the 28 day cube compressive strength (Table 2a) and the
equivalent cube strength results obtained from testing the broken parts after
flexural strength testing (Table 2b). The latter were obtained at the ages of 170
and 220 days as indicated in the Table. One of the aims of this study was to fmd
out if the equivalent cube test can replace conventional cube or cylinder tests
for compressive strength. The greater emphasis of the discussion below is
therefore on this aspect.
The results given in Table 2 show that the addition of steel or polyacrylonitrile
fibers or the incorporation of 5 per cent of SF as a cement replacementmaterial
has no significant effect on compressive strength. This result is as it should be
considering the amounts and roles of fibers and SF used in these tests. The
numericalvalues in Table 2 are ail within acceptable experimental errors associated
with the variability of concrete mixtures.
More importantly, the equivalent cube strengths tested at later ages, are
representative of the 28 day cube strengths, and further, reflect the effects of
curing and ageing on the 28 day result. Both the cube strengths and the equivalent
cube strengths are ail of the same order: the data, although limited, give confidence
that equivalent cube strengths can be relied upon to give realistic values of
compressive strength. There are other published data on piain concrete which
conñrm that the equivalent cube test is a good test for compressivestrength: the
test is also included in the British Standards. The study presented here shows
that this test can be confidently extended to fiber concretes without and with
silica fume.

A C 1 SP-1-55 95 H 0 6 6 2 9 4 9 0523328 773 m
242 Spadea et al
Free Shrinkwe and Exmnsion
Shrinkage (including both plastic and drying shnnkage) and expansion data are
often not quoted for fiber concrete, but in practice, the effects of plastic shnnkage
cracking and of the incorporation of fibers in creating internal voids, for example,
are important facton to be considered by the engineer. This is ilhistrated by some of the
re& obtainedinbiisstudy.
Fig 2 to 4 s d e lhe shnnkageand expanSiondata for lhe eght concrete mixtures

tested. Fig. 2 and 3 emphasize the role of the fiber materiaí and hence i& stiffness on
&rinkage. At i.û%fiberwlume,therewasasmaiibutdisiinctductionmfi-eedmnkage
exhibited by this steel &er mix whereas the p o i y a c r y l d e fibers cleady enhance fiee
&inkage, and this increases with fiber volume. This effect of poiyaayloniûile fibem m
enhancing fiee hidage is fiather confirmed m Fig 4 which shows the shiinkage and
expansion data for dica fume rtiixplres. This is again further dected m the water
absorpttOntes$ whereas steelfibersshowed no adverse &ect when immetsed m water,
polyacxyioniíriiefibersat0.25% again causedhigtierexpmion. Siuinkageand expansion
testingcan also @vesomeidea of the itúìuence of replacmg some of the pordand cement
with silica fume. This is shown m Fig 4. ïî is clear thaiwith adequaîe eady water cwing
the mcorporation of silica finne can help to reduce sigdicandy both shrinkage and
expamion, ineEipective of whether the niixtUres conîained fibers or not. Indeed the data
m Fig 2 to 4 show thai m appiicatlons where shrinkage and expansion are dorriinant
propertiesto be consi- the imA>Iporatiaiof 5% silica fume as cement repiacement
can be far more effecthe than the incorporationof fibers. One might m n question the
use of fibers m such k u m s t a n c e s (15).
Pulse Velocitv Measurements
The advantage of the puise wlocity measurementsis that they can give some idea of
the development of the intemal structm~,Particuiariy with time. Fig. 5 and 6 show that
air drying does create m t e d micro crac^ and botfi types of fiber concretes show a
reduction in puise velocity with time, the greater reduction occurring with the
poiyacryimiûiie ñbers.
The results ofthe dika fume umcrete in Fig. 7,on the ohm hand, confirm lhe positive
roie of îhk mineraiadtrMrhirem umcrete. The differences between the mixturesexposed
towetanddrymgcniditionsaremuchlesscanparedtoportlandcementconcreteniixtirres
shownin Fig. 5 and 6. indeed ifone considered oniy compre* skngth and the
effects of wetting or drynig, the daîa presented here show that the mcorporatonof 5 O h
silica b e as a cement repiacement materiaí can be just as effecthe as the addiibn of
steel or p o i y a u y l d e fibers.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-1.55 95 Ob62747 0 5 2 3 3 2 9 bOT H
Testing of FRC 243
First Crack Load
Table 3 shows the fnrst crack loads determined h m the load-transit time c m and
from the load-displacement curves. The puise velocity technique gives better and
more reliable first crack load values compared to those obtained fimload-deflection
curves.
The use of the puise velocity measurements in concrete has thus two majorbenefits
- firsts, it can give some idea of the development of internal microcraclung with the
progress of dryrng, secondiy, it can give more reliable data of first crack load values
compared to those obtained from load-deflection curves.
CONCLUSIONS
The major conclusions derived from this study are as follows:
1. Because of its simplicity and convenience, the slump test is easy to carny out in
the field. It is shown that provided a reasonable slump is designed for (about 50 to
120 mm), and the mix incorporates a superplasticizer,the slump test can be a good
guide of flow properties for field applications of concretes containing steel and non-
metallic fibers without or witb silica fume.
2. Sincefibers and silica fume both affect adverselythe flow properties of concrete,
the need for carehi mix proportioning of concretes containing these constituents to
produce concretes of adequate workability can be a key to wider application of fiber
concretes.
3. The equivalent cube strength test carried out on the broken pieces &er a
flexural test can represent the cube compressive strength of fiber concretes.
4. In applications where shrinkage and expansion are dominant properties to be

considered in design, the results of this study show that the incorporation of 5%
silica fume as cement replacement can be far more effective than the incorporation
of fibers. One mght even question the use of fibers in such circumstance.
5. The inclusion of 5% silica fume in concrete as repîacement of cement was

more effective in counteracting the effects of wetting or drying than portland cement
concrete mixtures.
6. Indeed if one considered oniy compressive strength, drymg shrhkage and

expansion, the use of 5% silica fume as cement replacement can be as effective as
the incorporation of fibers in concrete.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 = 0662949 0523330 32%
244 Spadea et al
Apart Bom the slump test tor Iresh concrete, the oniy other specimens that
i.
need to be cast to characterize the engineering properties of fiber concrete are
lOOx100x500 mm prisms. 'Tests on these specimens can give inîormation on
shrinkage, expansion, puise velocity, first crack load, fracture toughness, modulus
of rupture and equivalent cube strength.
8. With 100xlOOx500 mm test specimens, the number of different sizes of

specimensto be cast and tested for fiber concrete applications can be reduced. This
could help to widen the scope and practical use of fiber concretes.
REFERENCES
1. RILEM, Testing and Test Metho& ofFiBer Cement Comparites, RILKM hi.
Symp. Sheffield, April 1978, Editor R. N. Swamy,The Conshniction Press, pp. 545.
2. RILEM Committee 49-TFR, "Draft Recommendations Tesling Methods fix -

Fibre Reinforced Cement-Based ComposiW,"Mu~rjm , 17, No.
et C o m t r u c ~ mV.
102, June 1984, pp. 441-456.
3. FRC 86, Developments in Fibre Reinforced Cement and Concrete, REEM

Symposium, V. 1 and 2, F~IiîmR. N. Swamy, R. L. We- and D. R. ûakiey,
RILEM Technical Committee 49-TFR, 1986.
--``,`,-`-`,,`,,`,`,,`---
4. A C I C W e 544, 'nlleasurrxnentsofPmpdes of Fiber Reinforced Concrete,"

(ACT 544.2R), A C I M a t e r i a h J w m ~i! 85, No. 6, Nw.-Dec. 1988, pp. 583-593.
5. Fibre Rei$onzed Cementy and Concretes Recent h l o p m e n t y ,Edited by R

N. Swamy and B. Barr,ElsevierApplied Science, Barking, Essex, London, 1989, pp.
700.
6. Fibre Reinforced Cement and Concrete, Proc. of the Fourth RILEM Int.
Sympasium,Edited by R. N. Swamy, E & FN Spon, London, 1992, pp.1347.
7. El-Refsu, F. E., and Morsy, E. I€, "Some Properiies of Fibre Reinford Concrete
Wiai SuperpIastìcim,''mZC86,DevetOpments in FibreRei$omd Cement andconcret,
Ed. R N. Swamy, R L WagstaíTe and D. R Oakiey, V. 1, Paper 3.2, 1986.
8. Rossi, P., Jhrrouche, N., and de Lanard, E, "Method for Opthking the
Compition of M e i a b F h R e i r r f d Conc~&~,"
Fibre Reidowed Cementr and
Concretes :Recent Developments, Edited by R. N. Swamy and B. Barr,1989, pp.
1-10.

A C 1 SP-155 95 0662949 0523331 2 b ô
Testing of FRC 245
Y. uyan, M.. Yiidirim, H., and bryaman, A. H., "Workability and Durability of
Steel Fiber Reinforced Concrete Cast with Nomial Plasticizers," Fibre Raznforced
Cement andc'oncrete, Edited by R. N. swam^, lW2, pp. 70-81.
10. Swamy, R. N., and Jojagha, A. H., "Workability of Steel Fibre Reinforced
Lightweight Aggregate Concrete," Int. Journal of Cement Composites and
Lightweight Concrete, V. 4, No. 2, May 1982, pp. 103-109.
11. S w a y , R. N., andMangat, P. S., "Compactabilityof SteelFibre Reinforced

Concrete," Concrete, V. 8, No. 5, May 1974, pp. 34-35.
12. Swamy, R. N., and Stravides, H., "Some Properties of High Workability
Steel Fibre Concrete," Proc. RILEM Symp. 1975, Constr.Press Ltd.
13. Swamy, R. N., Ali, S. A. R.,and Theodorakopoulos,D. D., "Engheering

Properties of ConcreteCompositeMaterials IncorporatingFly Ash and Steel Fibres,"
ACI Special Publication SP-79, Detroiî, American Concrete Institute, Editor V. M.
Malliotra, V. 1, 1983, pp, 559-588.
-
14. Popovics, S., "The Slump Test Is Useless or Is it?," Concrete International,
V. 16, No. 9, September 1994, pp. 30-33.
15. Swamy, R. N., "Steel Fibre Concrete for Bridge Deck and Builduig Floor
Applications," Steel Fibre Concrete, Swedish Cemenl and Concrete Research
Institute, 1985, pp. 443-478.
TABLE 1 - DETAILS OF MIXTURES
Average of two batches
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 = 0662949 0523332 1 T 4
246 Spadea et al
3 ".
3
Y
O
vi
9
3 w
m
I æ
c
L3 G
i-
æ
Y
Di E
m
I-
V>
c
V>
Y
r Y
m
k2
Y
æ
U
m
L I- Y 09
Y d
3
Y z
J d
m a
æ
U Y
I I
n 3
N
O
--I
YI
N
2
Y 3
Y
2
d
m m
s 3
3 2
d
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 I0662949 0 5 2 3 3 3 3 030
Testing of FRC 247
E
E
O
52
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 m O662949 0523334 T77 m
--``,`,-`-`,,`,,`,`,,`---
248 Spadea et al
y -1000 T
2 I
m -800
....-.DR1 -air
d; -(jo0
-400
PC -water
-200
1O00
' g o I
.-8
' t
400
ii
a
600 i
Fig. 2-Shrinkage and expansion of mixes without SF (steel fibers)
y -1000 -
d -.-_PC -air
. _ . _PAN1
_ . - air
2
-800 -PAN2 - air
PC -water
3 -600 __ ........... PANI - water
3
;n -400
-200
1O00
b
W
O I
8
s
.e
400
600 i
Fig. L S h r i n k a g e and expansion of mixes without SF (polyacrylonitrile fibers)

A C 1 SP-155 95 I06629Li9 0523335 903
Testing of FRC 249
2
PCSF -air
-800 --
E/
rn
Y
- - - - - . D N S F -air
'a
b, 200 --
Age, days
.-8 --
400
ila
Fig. L s h r i n k o g e and expansion of mixes with SF

--``,`,-`-`,,`,,`,`,,`---
38 4.3
- _ - .PC -air
3'gl
PC -water
- - - - - . D R 1-air ........... DR1 -water
DR2 - air DR2 - water
i 3.5 I I I , I
O 50 1O0 150 200
Age, days
Fig. %Relationship between pulse velocity and age for mixes without SF (steel fibers)

A C 1 SP-L55 95 Obb2949 0523336 8 4 T
250 Spadea et al
4.7
o
jj8 4.3
!
I
2 3.9 I
1 PC
..-...PAN1 -air
-air PC -water
3 I ........... PAN1 - water
PAN2 - air PAN2 - water
3.5
O 50 1O0 150 200
Age, daYs
Fig. &Relationship between pulse velocity and age for mixes without SF (polyacrylonitrile
fibers)
4.7
T
-.-__
O ...............................
8 7:.7:-7 : z y. --y-.'
jj 4.3
2 3.9 _.
- _ -PCSF
_ -air PCSF -water
& DR2SF -air DR2SF -water
...... PAN2SF - air
~
...........
PAN2SF - water
3.5
O 50 100 150 200
Age, daYs
Fig. 7-Relationship between pulse velocity and age for mixes with SF
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 = 0662949 O523337 786
SI (Metric) Tables 251
CONVERSION F A C T O R h I N CH.POUND TO SI (METRIC)*
To convert from to multiply by
Length
inch . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . millimeter(mm) ..................... 25.4Ei

foot ................................. meter(m) . . . . . . . . . . . . . . . . . . . . . . . . . 0.3048B
yard ................................. meter (m) . . . . . . . . . . . . . . . . . . . . . . . . . 0.9144E
mile (statute) ........................ kilometer (km) ....................... 1.609
Area
square inch . . . . . . . . . . . . . . . . . . . square millimeter (mm’) . . . . . . . . . . . . . . . . . 645.1

square foot ............ . . . . . . . . . . square meter (m2) ...................... 0.0929
square yard . . . . . . . . . . . . . . . . . . . . . . square meter (m2) ...................... 0.8361
--``,`,-`-`,,`,,`,`,,`---
Volume (capacity)
ounce . . . . . . . . . . . . . . . . . . . . . . . . . . . milliliters (mL) ...................... 29.57

gallon ................ . . . . . . . . . . cubic meter (m3)$ ...................... 0.003785
cubic inch . . . . . . . . . . . . . . . . . . . . . cubic millimeter (mm3) . . . . . . . . . . . . . . . . 16390
cubic foot . . . . . . . . . . . . . . . . . . . . . . . . . cubic meter (m3) ...................... 0.02832
cubic yard . . . . . . . . . . . . . .. . . . . . . . . . . cubic meter (m3)$ ...................... 0.7646
Force
kilogram-force . . . . . . . . . . . . . . . . . . . . . . newton (N) . . . . . . . . . . . . . . . . . . . . . . . . 9.807

kip-force . . . . . . . . . . . . . . . . . . . . . . . . . . kilonewton(kN)
. . ...................... 4.448
pound-force .......................... newton(N) . . . . . . . . . . . . . . . . . . . . . . . . 4.448
Pressure or stress (force per area)
kilogram-forcelsquare meter . . . . . . . . . . . . . . . pascal (Pa) . . . . . . . . . . . . . . . . . . . . . . . . 9.807

kip-forcelsquare inch (ksi) . . . . . . . . . . . . . megapascal (ma)...................... 6.895
newtonlsquare meter (N/m2) . . . . . . . . . . . . . . pascal (Pa) ........................ 1.000E
pound-force/square foot . . . . . . . . . . . . . . . . . . pascal (Pa) ....................... 47.88
pound-fordsquare inch (psi) . . . . . . . . . . . . kilopascat (Ha) ....................... 6.895
Bending moment or torque
inch-pound-force . . . . . . . . . . . . . . . . . . . newtonmeter (Nam) ..................... 0.1130

foot-pound-force . . . . . . . . . . . . . . . . . . . newton-meter (N-m) ..................... 1.356
meter-kilogram-force . . . . . . . . . . . . . . . . newton-meter (Nm) . . . . . . . . . . . . . . . . . . . . . 9.807

A C 1 SP-155 95 = Obb2949 0523338 b L 2
252 SI (Metric) Tables
To convert from to mulliply by
Mass
ounce-mass (avoirdupois) . . . . . . . . . . . . . . . . . gram(g) . . . . . . . . . . . . . . . . . . . . . . . . 28.34

pound-mass (avoirdupois) . . . . . . . . . . . . . .kilogram (kg). . . . . . . . . . . . . . . . . . 0.4536
ton (metric) . . . . . . . . . . . . . . . . . . . . . megagram (mg) . . . . . . . . . . . . . . . . . 1.000E
ton (short, 2ooo Ibm) . . . . . . . . . . . . . . . kilogram (kg) . . . . . . . . . . . . . . .. 907.2
Mass per volume
pound-masdcubic foot . . . . . . . . kilogram/cubic meter (ks/m3) . . . . . . . . .. . 16.02

pound-masskubic yard . . . . . . . . kilogram/cubic meter (kg/m3) . . . . . . . . .. . . 0.5933
pound-mass/gallon . . . . . . . . . . . kilogram/cubic meter (kg/m3) . . . . . . . . .. 119.8
Temperatureg
degrees Fahrenheit (F) . . . . . . . . . . . . . degrees Celsius (C) . . . . . . . . tC = (tF - 32)/1.8

degrees Celsius (C) . . . . . . . . . . . . . . . degrees Fahrenheit (F) . . . . . . . . tF = 1.8tc + 32
' 'íük selected list gives practical conversion faciors ofunits found in wncrete technoIoa. The reference sources for information
on SI units and more exact conversion factors are ASIU E 380 and E 621. Symbois of metric units are given in parenthesis.
t E Indicates that the factor given is exact.
t One liter (cubic decimeter) equals 0.001 m' or 1000 cm'.
i These equations wmert one temperature reading to another and inciude the necessary scale wrrections. To wnvert a
difference in temperature from Fahrenheit degrees to Celsius degrees, divide by 1.8 only,ie., a change from 70 to 88 F represents
a change of 18 F or 18/1.8 = 10 C deg.
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 111 0 6 6 2 9 4 9 0 5 2 3 3 3 9 5 5 9
Index 253
INDEX
Accelerated aging, 115 Flexure, 115

Aguado, A., 23 Flowability, 233
Aluminum oxide, 93 Fracture mechanics, 93
Arino, A., 93 Fracture properties, 171
Anabi, M., 137
Gallo, D., 233
Bolaguru, P., 115 Gettu, R., 23
Banthia, N., 137, 171 Gopalaratnam, V., 23
Beams (supports), 1, 23
Hays, C., 217
Carbon, 93, 137 High-strength concretes, 23
Cava, R., 233
Cellular concretes, 217 Impact resistance, 217
Cements, 153, 171 Impact tests, 171, 189, 217
Chen, L., 41
Chokri, K., 171 Jamet, D., 23
Comminution, 153 Johnston, C., 1, 41
Composite materials, 93, 137, 171
Compression tests, 233 Kraoi, P., 153
Compressive strength, 233
Concretes, 93 Li, c., 93
Cracking (fracturing), 1, 23, 77, 93, 115, 137, load-deflection curve, 77
153 loading rate, 189
Deflection, 1 Microcracking, 233

Ductility, 23 Mindess, S., 41
Durability, 115 Mobasher, B., 93
Morgan, D., 41
Esters, 115 Mortars (material), 93, 153, 171
Expansion, 233
Nemegeer, D., 77
Failure, 217 Nylon fibers, 115
Fiber reinforced concretes, 1, 23, 41, 77, 93,
115, 137, 153, 171, 217, 233 Pocios, A., 189
Fibers, 1, 23, 41, 77, 93, 137, 189, 233 Pigeon, M., 41, 137
FlexuraI strength, 4 1 Polypropylene fibers, 1, 41, 93, 115, 137, 153
--``,`,-`-`,,`,,`,`,,`---

A C 1 SP-155 95 m 0662949 0523340 270
254 Index
Pullout tests, 189 Structures, 217

Swarny, R., 233
R-curves, 93
Restraints, 137 Tatnall, P., 77
Tensile strength, 115, 153
Scanning electron microscope, 115 Tests, 1, 23, 41, 233
Shah, S., 41, 189 Toughness, 1, 23, 41, 77, 93, 115
--``,`,-`-`,,`,,`,`,,`---
Shrinkage, 137, 153, 233 Trottier, J., 171

Silica fume, 23, 233
Slattum, K., 115 Ultraviolet light, 115
Slippage, 189
Spadea, G., 233 Vondran, G., 153
Steels, 23, 41, 77, 137, 189, 233
Strength, 1, 41, 171 Zollo, R., 217


Aci SP-155 - 1995 PDF

Caricato da

Informazioni sul documento

Titolo originale

Copyright

Formati disponibili

Condividi questo documento

Condividi o incorpora il documento

Opzioni di condivisione

Hai trovato utile questo documento?

Questo contenuto è inappropriato?

Copyright:

Formati disponibili

Aci SP-155 - 1995 PDF

Caricato da

Copyright:

Formati disponibili

Testing of Fiber

D. J. Stevens, et. al.

DISCUSSION of individual papers in this symposium may be submitted in

The Institute is not responsible for the statements or opinions expressed in

Printed in the United States of America

Editorial production Victoria Lunick

Library of Congress catalog card number 95-76808

Copyright American Concrete Institute

Fiber Reinforced Concrete (FRC) is a composite material that possesses

unreinforced concrete. Recent advances in testing techniques, instrumenta-

Dave Stevens, Nemy Banthia,

Copyright American Concrete Institute ...

AC1 Committee 544

James I. Daniel Vellore Gopalaratnam

Shuaib H. Ahmad Carol D. Hays Ernest K. Schrader

AC1 Committee 444

Kirk A. Marchand John R. Hayes, Jr.

Leslie A. Clark Mohsen A. Issa Rajan Sen

Copyright American Concrete Institute

PREFACE ........................................... ...

DEFLECTION MEASUREMENT CONSIDERATIONS IN

TOUGHNESS OF FIBER-REINFORCED HIGH-STRENGTH

COMPARATIVE TOUGHNESS TESTING OF FIBER

MEASURING TOUGHNESS CHARACERISTICS OF SFRC -

EXPERIMENTAL R-CURVES FOR ASSESSMENT OF

TEST METHODS FOR DURABILITY OF POLYMERIC

RESTRAINED SHRINKAGE TESTS ON FIBER REINFORCED

DIRECT TENSILE STRENGTH TESTING AT 6 HOURS OF

Copyright American Concrete Institute

IMPACT TESTS ON CEMENT-BASED FIBER REINFORCED

MEASUREMENT OF THE PULLOUT FORCE AT DIFFERENT

TESTING APPLIED TO THE EVALUATION O F DAMAGE

FIBER REINFORCED CONCRETE TESTING FOR

SI (Metric) TABLES. .................................. 241

INDEX ............................................ 243

Copyright American Concrete Institute

DefIectio n Measurement Considerat ions

Svnopsis: The issue of how the method of determining midspan deflection in

Keywords: Beams (supports); cracking (fracturing); deflection; fiber

Copyright American Concrete Institute

In a 1985state-of-the-artpaper on toughness(2), the writer questioned whether

Since the validity and accuracy of deflection measurements influences the

Copyright American Concrete Institute

Both sponsors selected specific fiber types, lengths and amounts to be

Copyright American Concrete Institute

The discussion is confined to toughness indices (I) and residual strength

Effect of Deflection-Measurine Techniaue

The third case is unstable strain-softening immediately after first crack

Copyright American Concrete Institute

Effect of Unstable Strain-Softening Immediatelv After First Crack

Some fiber type-amount combinations, particularly those using low fiber

Part of Fig. 3c enlarged to highlight the unstable region immediately following

Copyright American Concrete Institute

Since the position of the unstable portion of the load-deflection relationship

Effectiveness of Cl018 Parameters in Distinguishing FRC Performance

While the five single-specimen examples in Fig. 3 and 4 permit differences

Effect of Fiber Geometry and Amount--Three varieties of steel fibers

Copyright American Concrete Institute

Effect ofFiber five and Amount--Steel and polypropylenefibers are compared

Effect ofSmcimen Size--From the limited data available, a significant increase

Copyright American Concrete Institute

While there is no conclusive link between the individual values in terms of

1. Effectively utilizing ASTM Cl018 to evaluate the performance of FRC and