Materials and Structures/Mat6riaux et Constructions,Vol.

36, May2003, pp275-280

RILEM TC 162-TDF: 'TEST AND DESIGN METHODS FOR STEEL FIBRE

REINFORCED CONCRETE'

Round-robin analysis of the RILEM TC 162-TDF

uni-axial tensile test: Part 2

Fibre distribution

Prepared by B.L G. Barr and M.K.Lee, with contributionsfrom B. Barragdn, D. Dupont,

R. Gettu,J.F. Olesen, H. Stang and L. Vandewalle

TC MEMBF_~_SHIP:Chalri~: L. Vandewall~
B.t
H.]
Net
S.S
H,:
Netherlands.
.... il !i! i i !!i!{i!i!!!ii !i!i!ii i!il!!!ii i!!il!!i i84
ili84184184

ABSTRACT reported elsewhere and will not be repeated here [3, 4].
This paper, in particular, reports on a study carried out
Plain and steel fibre reinforced concrete cylinders during the course of the round robin test programme
were tested uni-axially in tension under closed-looped executed on the uni-axial tensile test.
conditions in a round robin test programme. During the In the round robin testing programme for the uni-
test, three deformation readings around the notched axial tension test, it was found that noticeable rotations
plane were measured. It was found that during the occurred during the course of the tests. This is charac-
course of the test, rotations were occurring and thus a terised by variations between the individual extensome-
fibre counting exercise was initiated. This was to investi- ter or linear variable differential transducer (LVDT)
gate the influence of fibre distribution within the cylin- readings from their mean measurements. Following this
der specimens on the observed rotations of the speci- observation, a study was carried out to investigate
mens. The results of the investigation could not detect whether fibre distribution could have an influence in
any strong correlation between the fibre distribution and such behaviour.
the observed rotations and a more likely source for the Compaction during the fabrication of SFRC speci-
variations observed is the boundary conditions of the mens has been found to influence the distribution and
tests. Significant variations were found in the number of orientation of the fibres. Souroushian and Lee [5] con-
fibres with coefficients of variation in the range of 30% cluded from their study involving nineteen beams that
to 50%. Additionally, it was found that toughness is the vibration of SFRC tend to orientate the fibres
approximately linearly related to the number of fibres in towards horizontal planes. Toutanji and Bayasi [6] car-
the fracture surface up to a certain limiting value where ried out a study using SFRC beams and found that the
the toughness appears to plateau in relation to the num- influence of testing direction (in relation to the casting
ber of fibres. direction) increases with increasing workability as it
encourages fibre settlement. More recently, Barrag~in et
al. [7] conducted a limited study on 150x300mm cylin-
1. INTRODUCTION ders and found that there was preferential orientation of
fibres in the horizontal direction for specimens com-
Recently, the RILEM TC 162-TDF presented two pacted via table-vibration and tamping. Some fibre seg-
recommendations on the testing of steel fibre reinforced regation was also found to exist for the specimens com-
concrete specimens [1, 2]. Round robin test programmes pacted via table-vibration. These findings provided the
involving both beam and cylinder specimens were car- motivation to carry out a fibre counting exercise to
ried out under a Brite-EuRam III project framework. investigate the influence of fibre distribution in the rota-
Further details of the round robin test programme are tions observed during the tests. Additionally, the influ-

1359-5997/03 @RILEM 2 75
Materials and Structures/Mat6riaux et Constructions, Vol. 36, May 2003

ence of the number of fibres on the Table 1 - Average total number of fibres with coefficients of variation, V(%)
material properties have been investi-
Concrete grade Inner sectors Outer sectors Grand
gated. Total Total
(Fibre dosage, Total
kg/m3) 1 3 5 2 4 6

NSC Avg 36.0

2. DEFINITION OF FIBRE (25) v(%) (30.7) i
C O U N T I N G EXERCISE NSC Avg 64.2
i

l
(50) V(%) (41.5)
The round robin programme was
divided into two phases. During the NSC Avg 7.71 8.76 9.00 25.5 22.9 16.1 22.4 61.4 85.5
(75) V(%) (87.6) (88.0) (75.2) (62.9) (41.6) (64.4) (71.5) (50.3) (48.7)
first and second phase of the round
robin test programme, fibre counting HSC Avg 2.15 2.00 1.95 6.10 7.50 6.50 5.60 19.6 25.9
took place for all the concrete cylinders (25)
v (%) (102) (74.3) (88.8) (57.3) (62.9) (50.5) (63.0) (43.0) (33.6)
involved. The cylinders were broken Note: indicates that detailed count upas not carried out.
into two and the total number of fibres
and the number of effective fibres were Table 2- Average number of effective fibres with coefficients of variation, V(%)
counted. The effective fibres were Concrete grade Inner sectors Outer sectors
defined as fibres with straight ends. In (Fibredosage, Total Total Grand
1 3 5 2 4 6 Total
other words, they have been deformed kg/m3)
and the original end hook has been NSC Avg , , , , , , , , 36.0
straightened out. Counting took place (25) q(%) (30.7)
on both halves of the specimens and NSC Avg , , , , , , , , 64.2
the numbers totalled. (50) v(%) (41.5)
Additionally, a more comprehen- NSC Avg
4.41 4.06 4.82 13.3 13.2 8.2 12.4 33.8 45.2
sive fibre count took place for two (75) v(%) (92.1) (89.0) (55.9) (50.3) (39.9) (74.3) (72.6) (47.8) (34.5)
types of material. These were the nor-
HSC Avg 0.65 0.50 0.65 1.80 3.50 3.25 2.85 9.60 11.4
mal strength concrete (NSC or (25) V(%) (125) (165) (144) (68.9) (86.2) (76.6) (89.1) (68.0) (60.4)
C25/30) cylinders with 75 kg/m 3 of
fibres and the high strength concrete Note: * indicates that detailed count was not carried out.
(HSC or C70/85) cylinders with 25
kg/m3 of fibres. The letter "C" indicates a characteristic compressive strengths in N/mm 2 measured from cylin-
compressive strength and the numerals indicate the ders and cubes respectively.
The fractured surfaces were divided into 3 distinct
sectors adjacent to their corresponding extensometers.
The sectors were divided further into 2 areas. This divi-
Extensometer 1 sion of the fractured surface area is illustrated in Fig. 1.
Therefore, there are 6 divided areas where fibre
counting took place. The sectors were defined in such a
way that they correspond to a particular extensometer
(or LVDT) location. This procedure enabled a check to
be carried out for a possible correlation between the
number of fibres in a sector and the corresponding
extensometer (or LVDT) reading.

3. RESULTS

Tables 1 and 2 list the average number of fibres and

the coefficients of variation, V, for the total number of
fibres and the number of effective fibres.
Fig. 2 shows a typical plot of the variation of individ-
ual LVDT readings from their mean values for a typical
Extensometer 3 Extensometer 2 concrete cylinder. It can be observed that rotation seems
to be developed in the pre-cracked phase. After a local
Note: peak value, the rotation decreases slightly and then set-
tles to an approximately constant value in the remaining
Notched area
post-cracked phase.
9 Position ofextensometer To investigate the rotations in the post-cracked
phase, the ratio of individual LVDT readings (at pre-
Fig. 1 - A broken half o f the cylinder divided into 6 areas. scribed deformations) were plotted against the ratio of

276
TO 162-TDF

0.4

<>

0.2

/-
0
3 -

2- 0
0
-0.2
I- 0 O O0 0

0
- - - Mean LVDT reading

-0.4 i , i

i = i
0.5 1 1.5
0 0.5 1 1.5
Mean LVDT (ram)
Ratio of total number of fibres (3+4)/(1+2)
Fig. 2 - Typical plot showing the variation between individual
LVDT readings. Fig. 4 - Ratio of LVDT2/LVDT1 against ratio of total number of
fibres in areas (3+4) and (1+2) for C25/30 cylinders (with 75kg/m3
of fibres) at a deformation, 8, of 200 btm.
6

6000

5000

9. 9 +
g.
4000

0 + +r 9 +
0
3000 ~+
O
6 + .+ ,9 ~=IO00pJ
0 0 O0 0
9 ,9 0 0

+. ,"
2000 + ++ " o
+ o o,, o

1000
-I i i i

0 0.5 1 1.5 - ~- X
x ~.
5=-I00~an
Ratio of total number of fibres (5+6)/(1+2) 0
50 1O0 150 200
Fig. 3 - Ratio of LVDT3/LVDT1 against ratio of total number of Total n u m b e r of fibres
fibres in areas (5+6) and (1+2) for C25/30 cylinders (with 75
kg/m 3 of fibres) at a deformation, 8, of 200 btm.
Fig. 5 - Toughness of NSC cylinders (with 25 kg/m 3 and 75
kg/m3 of Dramix 65/60 BN fibres) at various deformations, 8
(dashed lines show trendline if extreme values are excluded).
the total number of fibres in the sectors. Figs. 3 and 4
show some typical plots obtained. Other combinations tion values and plotted against the total number of fibres
of the ratio between individual LVDT readings and their as shown in Figs. 5 to 7. The curves of toughness against
corresponding ratio of the total number of fibres in the the number of effective fibres are similar.
sectors reveal similar trends. In addition, from the G-8 curves, G-w curves have
Additionally, an investigation was carried out to also been plotted. The crack opening, w, is evaluated by
study the influence of the number of fibres on the subtracting the elastic deformation of the material within
toughness of the material. The toughness of the material the gauge length but located outside the crack. This is
is simply defined as the area under the stress-deforma- achieved by assuming the material unloads from the peak
tion (G-8) curves. The areas under the G-5 curves for load with the same slope as the tangent at the initial
the cylinders tested were calculated at several deforma- loading path, as illustrated in Fig. 8. Subsequently, the

277
Materials and Structures/Mat6riaux et Constructions, Vol. 3,6, May 2003

1800 Initial slope of

LIo
1600 8=2000llm

~ 1400
g. 1200

~ 1000
~ X ~-IO00gm
800 *9 0 .. I 0 / Unloading

600 "" 0 0~' 0

,~ 400
0 ~500pm Average deformation, 5

2O0

0
0 50 100 150
TOtal n u m b e r offibres

Fig. 6 - Toughness of NSC cylinders (with 50 kg/m 3 of Dramix

80/60 BN fibres) at various deformations, 8 (dashed lines show
trendline if extreme values are excluded).

2500

+ +
2000 y

6=2000pl Crack width, w

i Fig. 8 - Procedure for calculating c - w curves from a-8 curves.
"~1500

3000 ,
1000 + + ++ + i o - o Total number of fibres
~=IC~OHrr X <>
2500 -t x Effective fibres
++ ++ +0 o + 0 0
X
500
oo 0 0 X 0 X 2000 -1

, . . ..., . , ,
~:~ml
~lOttm,
50gin l.
1500
t xx ~
xo
10 20 30 40 50
XO0 X
Total number of fibres
x~X "O~X 0 0
Fig. 7 - Toughness of HSC cylinders (with 25 kg/m 3 of Dramix 1000 ~ x o
80/60 BP fibres) at various deformations, & x~ ~ooo

XO
500 1

stress-deformation curve (o-8) and the stress-crack

width curve (o-w) can be plotted. The o-w curves were ol
used to calculate the value of Wf. In the context of the 0 20 40 60 80
work carried out here, Wf is defined as the area beneath N u m b e r o f fibres
the o-w curve between crack widths of 0 I.tm to 1400 Fig. 9 - W f against the number of fibres for NSC cylinders with
tim. The values of Wf have been plotted against the 25 kg/m 3 of Dramix 65/60 BN fibres.
number of fibres (both total and effective fibres) as
shown in Figs. 9 to 12.

278
TO 162-TDF

500 -
test specimen in the pre-peak regime is somehow sub-
jected to some eccentricity due to the loading arrange-
o Total number of fibres ment. Cracks do not form simultaneously around the
400 -
x Effective fibres notch-tip leading to the rotations observed. This actually
200 -
reflects well on the test as it means that the boundary
X
x o conditions have sufficient degrees of freedom to accom-
0 0 0 -
modate any eccentricities and settle to a position of equi-
x ,o librium.
x 0
800 - x
x 0
0 Figs. 3 and 4 do not show any obvious trend between
o,0. o the ratio of individual LVDT readings and the ratio of
600 -
X oo the number of fibres in the corresponding sectors.
x xo o Hence, the difference in the number of fibres is not a
400 -
factor in causing the difference between individual
LVDT readings. A more likely possibility is the bound-
200 9
XO ary conditions of the test set-up since it is extremely dif-
ficult to get a perfectly even surface. Furthermore, the
0 rotational stiffness and the uniformity of gluing may be
50 1O0 150 influencing factors.
Number of fibres Figs. 5 to 7 show the change in material toughness
with respect to the total number fibres at different stages
Fig. 10 - W f against the number of fibres for NSC cylinders with
of deformation. In general, the figures indicate that a
50 kg/m 3 of Dramix 80/60 BN fibres.

4000 1600
o Total number of fibres o Total number of fibres
3500 1400
x Effective fibres x Effective fibres x
X X

3000 1200

x o
2500 1000
x x o x o
X X X O O O
~mE~2000 ,~X O0
oo o o 800
=E x o
x xo o
oo • o
1500 600 • o
• o x x , '< 80 <>
o X 0
1000 O 400 x~ oo o<>
Y 0

500 200

o 0
o 50 100 150 200 0 10 20 30 40 50
Number of fibres Number of fibres

Fig. 11 - W f against the number of fibres for NSC cylinders with Fig. 12 - W f against the number of fibres for HSC cylinders with
75 kg/m 3 of Dramix 65/60 BN fibres. 25 kg/m 3 of Dramix 80/60 BP fibres.

4. DISCUSSION higher fibre count gives greater material toughness. This

observation is to be expected since the steel fibres pro-
From Tables 1 and 2, it can be observed that there is vide resistance to crack opening. Figs. 5 and 6 show an
significant variability in the number of fibres at the frac- interesting trend in terms of toughness against the num-
ture surface. The coefficient of variation for the total ber of fibres and warrants further study. The trendlines
number of fibres ranges from approximately 30% to for the toughness against the number of fibres intercepts
50%, whereas the coefficient of variation for the num- the y-axis at some point. Since at this point x=0, this
ber of effective fibres ranges from 35% to 60%. It is physically implies to no fibres in the cross section, hence
apparent from the table that the average number of fibres it is expected that this corresponds to the toughness of
increases with increasing fibre dosage. the concrete matrix. Therefore, the trendlines are not
From Fig. 2, the rotation is observed to develop in expected to intercept the y-axis at significantly large y-
the pre-peak regime. Thereafter, the rotation decreases axis values as the toughness of the concrete matrix is sig-
and approaches a constant value. This implies that the nificantly lower than the steel fibre reinforced compos-

279
Materials and Structures/Materiaux et Constructions, Vol.36, May2003

ite. If all data points are taken into account and a linear In the investigations carried out, it is observed that
relationship is assumed, it seems that this logical supposi- there does not seem to be a relationship between the
tion is not satisfied by the trendlines for 5=1000 gm and number of fibres in a particular sector and the difference
8=2000 ~tm for the NSC cylinders (the y-intercept at in the corresponding extensometer (or LVDT) readings.
these two deformations for the NSC specimens are Additionally, the results suggests that toughness is
large). By ignoring extreme points (at high number of approximately linearly related to the number of fibres in
fibres visually determined at approximately 90 fibres), the fracture surface up to a certain limiting value where
the dashed lines can be plotted (in Figs. 5 and 6) and a the toughness appears to plateau with respect to the
more satisfactory trendline is obtained. It seems that the number of fibres.
relationship between toughness and the fibre count is
linearly related up to a certain limiting value where the
toughness appear to plateau in relation to the number of ACKNOWLEDGEMENTS
fibres. This tendency is not readily detected for the HSC
specimens as only a low fibre dosage was employed for The work reported in this paper forms part of the
this concrete strength (25 kg/m3). A possible cause for Brite-Euram project "Test and Design Methods for Steel
this observed trend is that with an increase in the num- Fibre Reinforced Concrete", contract n ~ BRPR-CT98-
ber of fibres, there could be reduced efficiency of the 0813. The partners in the project are: N.V.Bekaert S.A.
matrix-fibre bond. This is because, with a significant (Belgium c o - o r d i n a t o r ) , C e n t r e Scientifique et
number of fibres, there is a possibility that there could be Technique de la Construction (Belgium), Katholieke
fibre-fibre interaction or the matrix does not envelop Universiteit Leuven (Belgium), Technical University of
the fibres completely. Denmark (Denmark), Balfour Beatty Rail Ltd (Great
Figs. 9 to 12 show the plots of Wfagainst the number Britain), University of Wales Cardiff (Great Britain),
of fibres counted. Wf is calculated from the rs-w curve. F e r t i g - D e c k e n - U n i o n G m b H (Germany), R u h r -
A linear trend can be observed (within the range consid- University-Bochum (Germany), Technical University of
ered) from these figures, which implies a proportional Braunschweig (Germany), FCC Construccion S.A.
increase in material toughness with the number of fibres, (Spain), Universitat Polyt&nica de Catalunya (Spain).
as expected.

REFERENCES
5. CONCLUSION
[1] RILEM TC 162-TDF, 'Test and design methods for steel fibre
The results obtained from the round robin pro- reinforced concrete. Uni-axial tension test for steel fibre rein-
gramme for the uni-axial tensile test showed that there forced concrete. Recommendations.', Mater. Struct. 34 (235)
were significant differences in the individual extensome- (2001) 3-6.
[2] RILEM TC 162-TDF, 'RILEM TC 162-TDF: Test and design
ter (or LVDT) readings. Therefore, a study was carried methods for steel fibre reinforced concrete. Bending test.
out where the fibre distributions of some of the speci- Recommendations',Mater. Struct. 33 (225) (2000) 3-5.
mens were determined. The purpose of the study was to [3] Barr, B.I.G. and Lee, M.K., 'Definition of round robin test.
determine whether the fibre distribution could have Preparation of specimens. Execution and evaluation of round
caused the differences in the individual LVDT readings. robin testing (Uni-axial tensile test)', Report from Test and
Design Methods for Steel Fibre Reinforced Concrete, EU
In the study, fibres were counted at 6 different areas on Contract- BRPR- CT98 - 813 (2001) 70 p.
the final fracture surface. Additionally, the areas were [4] Barr, B.I.G. and Lee, M.K, 'Round robin analysisof the RILEM
assigned to their corresponding adjacent LVDT. This TC 162-TDF uni-axial tensiletest: Part 1 - Test method evalua-
procedure enabled a check to be carried out to investi- tion', accepted for publication in Mater. Struct. 36 (258) (2003)
gate a possible correlation between the number of fibres 265-274.
[5] Soroushian, P. and Lee, C.-D., 'Distribution and orientation of
in a sector and the corresponding LVDT readings. fibers in steel fiber reinforcedconcrete',A CI MaterialsJournal 81
It was found that there were significant variations in (5) (1990) 433-439.
the number of fibres. The total number of fibres had [6] Toutanji, H. and Bayasi,Z., 'Effectsof manufacturingtechniques
coefficients of variation ranging from 30% to 50%. The on the flexural behavior of steel fiber-reinforced concrete',
Cement and ConcreteResearch 28 (1) (1998) 115-124.
number of effective fibres had a similarly high variation [7] Barrag~n,B.E., Gardner,D., Gettu, R. and Ferreira,L.E.T., 'Study
ranging from 35% to 60%. The average number of of the distribution and orientation of fibers in cast cylinders',
fibres, as expected, was observed to increase with Report for Test and Design Methods for Steel Fibre Reinforced
increasing fibre dosage. Concrete,EU Contract-BRP1K-CT98-813(2000) 8 p.

280
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.