Engineering Structures 198 (2019) 109333

Contents lists available at ScienceDirect

Engineering Structures
journal homepage: www.elsevier.com/locate/engstruct

Concrete beams with different arrangements of GFRP flexural and shear T

reinforcement
⁎
M. Krall , M.A. Polak
Department of Civil and Environmental Engineering, University of Waterloo, 200 University Ave. W., Waterloo, ON N2L 3G1, Canada

A R T I C LE I N FO A B S T R A C T

Keywords: This paper presents experimental research on the flexural and shear behaviour of concrete beams reinforced with
GFRP Glass Fiber (Fibre) Reinforced Polymer glass fibre reinforced polymer (GFRP) composite reinforcement. Three-point bending tests on shear critical
Deep beam concrete beams reinforced with GFRP longitudinal bars and GFRP stirrups are presented. The test variables were
Shear the size and spacing of closed loop stirrups as well as the size, amount, and arrangement of longitudinal re-
Arch action
inforcement. Twelve beams were tested with four different stirrup arrangements and three different longitudinal
FRP stirrups
Stirrup contribution
reinforcement arrangements. Longitudinal bars were instrumented at mid-span and stirrups were instrumented
Concrete at the straight and bent portions. The beams had a shear-span-to-effective-depth ratio of 2.5 and behaved like
Concrete contribution deep beams. The beams without stirrups failed in shear-tension while the beams with stirrups failed in shear-
Confinement compression with no ruptured stirrups. Longitudinal reinforcement strains showed evidence of a tied arch
Bend strength mechanism. The analysis found diminishing increases in shear strength with increasing shear reinforcement ratio
when failure occurred as a result of crushing of confined concrete.

1. Introduction GFRP bent bars manufactured with thermoset resin must be shaped
before curing. Bent shapes, in general, are initially formed using pul-
Glass Fibre Reinforced Polymer (GFRP) bars are used as internal trusion to align and coat the fibers. Some manufacturers may shape the
reinforcement for concrete structures and are geometrically similar in bars before curing, while others may partially cure the bars before
size and shape to conventional steel reinforcement. Unlike steel, GFRP shaping. The bending process may range from by-hand lay-up methods
is electromagnetically neutral; it does not participate in electrochemical to automated processes. Curing of bent bars takes place in a separate
reactions, it is non-conductive, and non-magnetic. Owing to these curing chamber. The final bent bar product may, to different degrees,
properties, GFRP bars are used in structural designs requiring specific be influenced by human factors. The strength of bent bars is less than
performance in terms of functionality or durability that cannot be the strength of straight bars in the straight and bent portions [1–3].
achieved using steel. In the presented research programme, 12 concrete beams were
Glass fibre reinforced polymer bars are a composite of long glass tested in three-point bending. The beams were of similar size, with
fibres and a polymer matrix that binds the fibres together. Straight constant depth of flexural reinforcement, d, and a constant a/d ratio of
GFRP bars are manufactured in an industrial process known as pul- 2.5, which classifies them as “semi-deep” members. Nine beams were
trusion. GFRP bars have anisotropic behaviour; they are strong and stiff reinforced in flexure and shear with GFRP bars and stirrups and another
in the longitudinal direction, which is aligned with the long fibres, and three control specimens without stirrups.
weaker and less stiff in other directions. Tensile and compressive be-
haviours, along the same axis, are also known to be different. As a re- 1.1. Research objectives
sult, design with GFRP bars focuses on utilizing their behaviour and
strength in tension, along the strong longitudinal axis, while often ne- The original objective of this work was to investigate the influence
glecting their contributions when loaded in any other way. that longitudinal reinforcement arrangement has on the bend strength
Bent GFRP bars, which are needed to achieve hooks or stirrups, are of GFRP stirrups. It was postulated that moving the centroid of the
not manufactured in the same pultrusion process as straight bars. There longitudinal reinforcement geometry away from the bend, up along the
is no industry consensus on a uniform manufacturing process and a beam height, would prevent failure at the bend, or rather encourage
manufacturer may hold their process to be a trade secret. However, failure in the straight portion of the stirrup. However, no stirrups

⁎
Corresponding author.

https://doi.org/10.1016/j.engstruct.2019.109333
Received 18 July 2018; Received in revised form 22 April 2019; Accepted 21 June 2019
0141-0296/ © 2019 Elsevier Ltd. All rights reserved.
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

Nomenclature PE experimental beam strength measured as the peak load

recorded during the test (kN)
Av cross-sectional area of one stirrup leg (mm2) Pi applied load at data point i (kN)
AF cross-sectional area of a longitudinal bar (mm2) rbend bend radius of GFRP stirrup reinforcement (mm)
a shear span length (mm) s spacing of shear stirrups (mm)
a/d shear-span-to-depth ratio V applied shear load (kN)
b beam width (mm) VC concrete contribution to shear strength (kN)
d effective depth to reinforcement measured from extreme VC,i concrete contribution to shear strength at data point i (kN)
compression fibre (mm) VE experimental beam shear strength measured as the peak
E modulus of elasticity of the straight portion of GFRP re- shear load recorded during the test = 0.5PE (kN)
inforcement (GPa) VS stirrup contribution to shear strength (kN)
fS normalized stirrup stress = VS/(n·A) (MPa) VS,i stirrup contribution to shear strength at data point i (kN)
fu_bend strength of the bent portion of the GFRP stirrups (MPa) Δmid mid-span displacement (mm)
fu_straight strength of the straight portion of the longitudinal and ε strain, reinforcement strain
stirrup reinforcement (MPa) εavg,i average stirrup strain at data point i
h beam height (mm) εFu rupture strain of GFRP reinforcement
i a point in the index of data from the recorded in- εmax maximum strain recorded during the test
strumentation measurements εmid strain recorded by the mid-span strain gauge
l beam length (mm) ρf flexural reinforcement ratio
n the number of engaged stirrup legs in the shear span ρv shear reinforcement ratio = 2Av/(b·s)
P applied load (kN)

ruptured during the tests. Thus, the presented study is on the effec- and are discussed.
tiveness of GFRP, brittle-type, stirrup reinforcement on load carrying Ahmed et al. [10] tested three beams that were reinforced in flexure
capacity of concrete beams reinforced in flexure and shear with GFRP with 3 layers of steel strand and in shear with GFRP stirrups at different
bars and stirrups. The behaviour of GFRP shear reinforcement at the spacings. Their analysis found that the effective stirrup stress was very
bends and the contribution of GFRP longitudinal bars to dowel action close to the straight portion strength, 95–105 percent, when the beams
through different arrangements of bars were studied. failed in shear by rupture of a stirrup bend [10]. In another paper
Ahmed et al. [11] describe tests on three beams that were identical to
the first set, but which were reinforced with CFRP stirrups. In those
1.2. FRP bend behaviour motivating the research programme
tests, two of the beams failed in shear by rupture of a stirrup bend, but
with an effective stirrup stress of 72 percent or less than the straight
Bent FRP bars are known to rupture at the bends when subjected to
portion strength. Ahmed et al. did not characterize the discrepancy, but
common loading arrangements. When the bar ruptures at the bend, the
total shear was higher for the beams with CFRP stirrups, which have
stress carried in the straight portion of the bent bar will be a fraction of
both a higher modulus and strength when compared to GFRP.
the straight portion’s strength; this reduced load carrying capacity is
Bentz et al. [12] examined a number of beams of various sizes and
termed bend strength and characterizes the material. This phenomenon
reinforcement arrangements to study the size and strain effects on GFRP
has been observed by several researchers over the last two and a half
reinforced concrete beams. The authors observed a shear failure by
decades [1,2,4–12]. The bend strength depends primarily on the bend-
rupture of the stirrup bend on a beam that had one layer of flexural
radius-to-bar-diameter ratio [13], though three other factors, described
reinforcement. That beam had a counterpart with four layers of flexural
below, also reduce the strength of FRP bars at bends.
reinforcement and it was observed to fail in shear by rupture of the
The first is due to manufacturing defects, namely fibre kinking [3]
stirrup straight portion above the top layer of flexural reinforcement. In
and variation in cross-sectional shape [14]. These defects are manu-
their analysis, Bentz et al. found that the shear stress on the beam with
facturer dependant and their impact exacerbates the other two factors.
4 layers of flexural reinforcement was highest at the top of the cross-
The second factor is due to transverse loading across the bend, for
section and the stress went to zero quickly over the depth of the flexural
example when a shear crack opens through a stirrup bend or in a pull-
reinforcement layers. However, the shear stress on the beam with 1
out type test. This loading will produce bond stresses parallel to the bar
layer of flexural reinforcement was highest at the stirrup bend near the
surface and transverse bearing stresses. The combination of these
depth of the single layer of longitudinal reinforcement. The authors
stresses creates a multiaxial state of internal stress along the bar in
concluded, “…that it would be prudent to design FRP beams with
which stresses do not act in the direction of the fibres. If the strength
stirrups with multiple layers of flexural reinforcement to protect the
along a weaker axis is exceeded, then the bar will fail [1–3,7,9–12,14].
weaker bend location of the stirrups” [12]. The Ahmed et al. [10,11]
Ahmed et al. [3] note a retained strength of 40–60 percent, concluding
studies also had multiple layers of longitudinal reinforcement.
45 percent in their own investigation, while Shehata et al. [1] found a
retained strength of 50 percent.
The third factor is due to transverse loading from dowel action, 2. Experimental programme
which involves transverse loading caused by the movement of concrete
crack faces. This phenomenon is also present in straight FRP bars The research programme consisted of 12 beams tested in three-point
crossing inclined shear cracks. Transverse movement of the crack faces bending under displacement controlled loading. The specimens were
will load a bar in shear and create a multiaxial state of stress internal to concrete beams with rectangular cross-sections and were reinforced in
the bar within the vicinity of the crack faces. Similarly, if the strength flexure and shear with GFRP bars. The shear reinforcement was pre-
along a weaker axis is exceeded, then the bar will fail [4,5,7,15]. formed closed loop stirrups.
Shehata et al. found this factor less significant than the previous and The specimens were designed using the CSA S806-12 design stan-
noted a retained strength as low as 65 percent [1]. dard [16], which uses an equivalent stress block approach to flexural
Of the several studies on FRP stirrups, two include stirrup failures at design and a shear design approach that sums the separate concrete and
both the bend and straight portions of their studied beam specimens stirrup contributions. The predicted shear strengths were checked

2
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

against the Nehdi et al. [17] shear strength model, specifically their are indicated in Table 1. In Table 1, b and h are the beam width and
‘optimized equations’. A set of ‘design equations’ from Nehdi et al. [17] height as showing in Fig. 1, l is the total length of the specimen, ρf is the
that included slightly more conservative coefficients, were found to be flexural reinforcement ratio, and ρv is the shear reinforcement ratio,
good predictors of beam strength when compared to 7 other models in a where ρv = 2Av/(b∙s), Av is the area of one stirrup leg, b is the width of
review by Machial et al. [18]. the beam, and s is the stirrup spacing. Additional details may be found
The beams were designed to be shear critical based on the design in Krall [19].
equations from CSA S806-12 standard [16]. The beam dimensions were
constrained to keep strengths below 500 kN, a limit set by the capacity 2.2. GFRP and concrete material properties
of available testing machines at the time. The constrained dimensions
led to congested reinforcement in the beams with 12 mm longitudinal Beam specimens were reinforced with GFRP longitudinal bars and
bars; however, longitudinal bar anchorage length, which was the pri- stirrups. The bar properties, as specified in the manufacturer’s product
mary concern with reinforcement congestion, was extended as per literature at the time, are shown in Table 2 and the bars are shown in
equation 9–1 of CSA S806-12 [16]. The concrete cover was 20 mm for Fig. 2. The straight and bent bars came from the same manufacturer,
all specimens. but the bent bars are manufactured in a corrugated tube to facilitate
bending.
2.1. Beam design parameters The concrete was a high slump mix with pea-stone type aggregate
having a maximum size of 9.5 mm. A local ready mix supplier provided
All specimens had an effective depth, d, of 270 mm and a shear both the mix design and concrete. The 28 day strength, based on three
span, a, of 675 mm (or a total span of 1350 mm) for a shear-span-to- Ø100 mm × 200 mm concrete test cylinders, was 47.3 MPa [19].
effective-depth ratio, a/d, of 2.5, as seen in Fig. 1. The 12 beams are
divided into 4 groups based on stirrup arrangement, but may also be 2.3. Beam instrumentation
divided into 3 groups based on longitudinal reinforcement arrange-
ment. The beam naming convention outlines these divisions: for ex- Information on applied load, mid-span displacement, longitudinal
ample, “BM16-150”, identifies the specimen as a beam “BM”, the “16” bar strains, and stirrup strains were recorded during the beam tests.
indicates the GFRP manufacturer’s bar diameter in millimeters for the Applied load was measured using a load cell attached to the actuator of
longitudinal reinforcement, and “150” indicates the spacing of stirrups the load frame. Mid-span displacement was measured using an LVDT
in millimetres. Note that beams without stirrups had an “INF”-inite with a 150 mm stroke. The layout of instrumentation is shown on Fig. 3,
stirrup spacing. Fig. 2 shows the cage for BM16-150 and the re- which is typical for all beams.
inforcement used in the experimental programme. Strain gauges were installed on each layer of longitudinal re-
Where stirrup arrangements were concerned, the BMX-INF group inforcement 50 mm away from exact mid-span to avoid potential da-
had no stirrups, the BMX-220 & -150 groups had 12 mm stirrups, and mage during the formation of the mid-span flexural crack. Beams
the BMX-230 group had 20 mm stirrups. Otherwise, where longitudinal BM16-INF, -220, & -150 had additional strain gauges spaced at 150 mm
reinforcement arrangement was concerned, the BM12-X group had intervals along the upper longitudinal bars as shown in Fig. 3. These
three layers of 4–12 mm bars, the BM16-X group had two layers of additional gauges were used to observe arching action. Gauge locations
3–16 mm bars, and the BM25-X group had one layer of 2–25 mm bars. are indicated in Fig. 3 and are shown at the black tape locations in
The BMX-s230 group is a special case where the cross-sectional di- Fig. 2.
mensions were increased to accommodate the thicker stirrups. The ar- The stirrups were instrumented with one gauge at mid-height of the
rangements may be seen in Fig. 1 and the remaining design parameters stirrup leg and another on the opposite leg immediately above the

Fig. 1. Specimen cross-sections and typical layout.

3
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

Fig. 2. Photos of BM16-150 cage, longitudinal bars, and stirrups.

Table 1
As-designed parameters.
Parameter BM12-INF BM16-INF BM25-INF BM12-220 BM16-220 BM25-220 BM12-150 BM16-150 BM25-150 BM12-s230 BM16-s230 BM25-s230

b (mm) 200 200 200 200 200 200 200 200 200 230 230 230
h (mm) 350 345 330 350 345 330 350 345 330 365 360 345
l (mm) 1810 1930 2470 1810 1930 2470 1810 1930 2470 1810 2110 2310
ρf (%) 2.51 2.23 1.82 2.51 2.23 1.82 2.51 2.23 1.82 2.18 1.94 1.58
ρv (%) – – – 0.51 0.51 0.51 0.75 0.75 0.75 1.19 1.19 1.19

Table 2 percent of the final maximum load. Other entries in italics, which are in
GFRP bar properties. the strain results, also indicate values wherein the true maximum
Property Longitudinal bars Stirrups
cannot be known because of premature gauge failures or because
failure occurred in the un-instrumented span.
12 mm 16 mm 25 mm 12 mm 20 mm Table 3 shows that beams of the same shear reinforcement ratio
have similar peak loads when compared to the other ratios and that
fu_straight (MPa) 1000 1000 1000 1000 900
E (GPA) 60 64 60 50 50
peak load increased with increasing shear reinforcement ratio.
AF or Av (mm2) 113 201 491 113a 314a For the group of beams without stirrups, BMX-INF, both peak load
εFu (%) – 2.61b – – – and displacement at peak load increased with increasing flexural re-
fu_bend (MPa) – – – 700 550 inforcement ratio while the maximum longitudinal strain increased
rbend (mm) – – – 42 70
with decreasing flexural reinforcement ratio. The behaviour of the
a
Cross-section calculated from reported diameter.
BMX-INF series can be attributed to the shear resisting mechanisms of
b
Ultimate strain only reported for size 16 mm bars. (i) aggregate interlock & shear friction, and (ii) dowel action of the
longitudinal reinforcement. Both mechanisms increase shear strength
flexural tension zone bend, as shown in Figs. 2 and 3. All beam groups with increasing longitudinal reinforcement ratio. The slight increase in
had strain gauges on the stirrups centered within the south shear span, beam height, which increases with reinforcement ratio, may also have
but the BM16-X group of beams also had strain gauges on the stirrups at contributed to increases in strength.
mid-span and immediately above the bearing plate. Further in- The beams with shear reinforcement showed deep beam behaviour
strumentation details may be found in Appendix E of Krall [19]. and arching action was the dominant shear resisting mechanism, which
is influenced mainly by the a/d ratio [18].
Fig. 4 shows the load-deflection results for each group of beams on
3. Experimental results separate plots with identical scales and are organized by shear re-
inforcement ratio. The BM25-s230 beam peaked off the region of the
All beam specimens showed concrete controlled shear failures. plot at the values shown in Table 3. Individual plots for each beam may
Beams without stirrups, BMX-INF, failed in shear tension. Beams with be found in Krall [19].
stirrups, BMX-150, -220, -s230, failed in shear-compression showing Fig. 4 shows that beams with higher longitudinal reinforcement
deep beam behaviour with no stirrups rupturing. ratios, plotted with light grey lines, had stiffer post cracking responses
Table 3 presents the primary results from each beam: the peak/ than those with lower ratios. The plots also show that beams with
failure load recorded during the test, PE, the displacement at mid-span, stirrups have some degree of post peak deformability and that those
Δmid, and the maximum reinforcement strains, εmax. The longitudinal with lower longitudinal reinforcement ratios showed more post-peak
and shear reinforcement ratios are included for reference. deflection. The increase in deformability is likely related to bar surface
A power failure occurred while testing BM16-220 and the results for area, which is a proxy for the concrete-bar bond, resulting from the
this beam are shown in italics. Before dropping to a state of no load, the specific way longitudinal bar arrangements were defined for this re-
data acquisition system recorded a small 12 kN jump to a load level 43 search programme. Beam group BM25-X was reinforced with 2–25 mm

4
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

Fig. 3. Instrumentation layout for BM16-150.

Table 3 cracks formed on the sides of the beams following peak load. Fig. 5
Primary results. shows the failed shear span of BM12-INF.
Beam ρv, % ρf, % PE, kN Δmid Longitudinal Stirrup Stirrup
Beams with stirrups failed in shear-compression. The primary me-
at PE, εmax, με bend straight chanism was crushing of web concrete adjacent to the load plate. In
mm εmax, με εmax, με general, specimens did not display any abrupt behaviours as testing
approached and passed peak load. As shown in Fig. 4, displacement
BM12-INF – 2.51 163.1 4.3 2523 – –
controlled loading was continued as load gradually diminished until the
BM16-INF – 2.23 150.2 4.2 2744 – –
BM25-INF – 1.82 125.1 3.2 2919 – – authors were satisfied the beams would not pick-up more load. How-
ever, BM16-150 displayed an abrupt failure after some post peak de-
BM12-220 0.51 2.51 382.4 12.0 7448 1237 4952
BM16-220a 0.51 2.23 309.3 8.4 5863 256 4742 formation and BM12-s230 displayed similar behaviour immediately
BM25-220 0.51 1.82 360.1 24.0 8783 11,484 10,558 within the vicinity of the second lower peak. These two failures were
BM12-150 0.75 2.51 405.2 15.6 8561 11,425 14,256
accompanied by loud noise, movement, and the ejection of concrete
BM16-150 0.75 2.23 416.5 14.4 10,393 8429 5904 pieces. Fig. 5 shows the failed shear spans of BM12-s230, BM25-220,
BM25-150 0.75 1.82 415.8 22.1 9982 5546 11,837 BM16-150, which were selected to include a result from each long-
BM12-s230 1.19 2.18 466.9 12.5 9909 877 4081 itudinal group and stirrup group.
BM16-s230 1.19 1.94 450.1 22.0 9059 1289 3652 Cracking in beams with stirrups over the course of each test was as
BM25-s230 1.19 1.58 444.0 35.6 11,977 2388 10,035 expected: the mid-span flexural crack formed first followed by others
near the stirrup locations. The flexural cracks transitioned into flexural-
Note: Italicised entries should be considered with caution for the reasons out-
shear cracks that were then followed by shear cracking. Ultimately, the
lined in the following paragraphs.
a
BM16-220 experienced one unintended 43% load cycle before failure as a
critical shear cracks ran from load plate to bearing plate.
result of a power failure.
3.2. Longitudinal reinforcement strains
bars and had both a lower reinforcement ratio and less surface area
than beam groups BM16-X and BM12-X, which had incrementally The longitudinal reinforcement strains at mid-span were typical for
more. This implies that the 25 mm diameter bars were more prone to GFRP reinforced beams. These results are shown in Fig. 6 for each
slip and showed less tension stiffening of cracked concrete compared to group of beams on separate plots with identical scales and are orga-
the other arrangements. nized by longitudinal reinforcement ratio.
Fig. 4 also shows that a number of beams had a two peak load- In general, the mid-span strain response of the longitudinal re-
deflection response. The results presented in this paper consider the inforcement was stiff until cracking. The response softened and shortly
highest peak and the mid-span deflection at that peak. thereafter recovered a slight amount. This behaviour was observed on
When load-deflection curves were plotted by longitudinal bar ar- all beams and is best seen at about 1750 micro-strain in the BM25-X
rangement, which is omitted from this paper, they showed that beams group of Fig. 6. Beams with the same flexural reinforcement ratio
within each group had a very similar pre-cracking response, similar showed very similar strain responses. Beams BM12-220, BM16-150,
cracking loads, and similar post cracking stiffnesses regardless of the BM16-s230, and BM25-s230 experienced longitudinal strain gauge
shear reinforcement ratio. failures before peak load by about 5 kN, 10 kN, 50 kN, and 30 kN, re-
spectively.
Beams BM16-INF, -220, & -150 were instrumented with additional
3.1. Cracking and failure modes strain gauges along the longitudinal bars to observe strain development
along the length of the shear span over the course of the test, see Fig. 7.
Beams without stirrups failed in shear-tension with critical shear Fig. 7 is an arch action plot that shows strain on the y-axis and strain
cracks extending from load plate to bearing plate for BM12-INF and gauge location along the span on the x-axis, as commonly done by other
BM16-INF and from load plate to quarter span for BM25-INF. Splitting researchers [20–22]. Each curve on the plot represents a different load

5
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

Fig. 4. Load-deflection plots for all beams grouped by shear reinforcement ratio, ρv.

Fig. 5. Typical cracking on beams with & without stirrups.

level shown as a percentage of peak load. The increments along the x- suggests that the longitudinal bars de-bonded from the concrete
axis of Fig. 7 map to the strain gauge location along the longitudinal forming a tie, which indicates deep beam behaviour.
reinforcement of BM16-220 in millimeters.
Fig. 7 shows that the longitudinal reinforcement strains were
highest near mid-span at the 30 percent load level. Other gauges re- 3.3. Stirrup strains
corded increases in strain as the applied load increased with almost
constant strains in longitudinal reinforcement along the beam at the Generally, for all beams, the gauges on the straight portion of the
highest load levels towards the end of the test. Constant strain in the stirrup were significantly more engaged than at the stirrup bends, ex-
longitudinal reinforcement over the centre half of the shear span cept for one stirrup bend on BM16-150 [19]. Fig. 8 shows a plot of load
against stirrup strain for BM12-150 and shows typical behaviour for the

6
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

Fig. 6. Longitudinal reinforcement strain at mid-span for all beams grouped by longitudinal reinforcement ratio, ρf.

Fig. 7. Arch action plot for BM16-220.

7
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

however, the strains at the bends were consistently smaller than in the
straight portion.
Stirrup-C, closest to mid-span, was engaged first due to the forma-
tion of flexural cracks. The strain gauge on the bend of stirrup-C never
showed significant strain because it was the furthest away from line of
action of the strut; shear crack growth took place elsewhere in the span.
Stirrup-A, closest to the support, was the last stirrup to engage. The
stirrup-A straight gauge always measured larger strains than the bend
gauge due to its location near the centre of the line-of-action of the
strut. The bend gauge on stirrup-A showed rapid increases in strain near
the end of the test when the critical shear crack developed. The stirrup-
C gauges tended to show larger strains than stirrup-A over the test, but
the establishment of the critical shear crack and its rapid growth at the
end of the test cause the stirrup-A strains to grow larger because they
were closer to the strut centreline.
Stirrup-B was closest to quarter span where, generally, the bottle
shaped strut is expected to be largest and where a greater amount of
shear cracking is expected to take place. Indeed, stirrup-B showed the
largest strains over the entire course of the test as it confined the bottle
Fig. 8. Stirrup strain plot for BM12-150.
shaped strut at its widest point. The curves for both stirrup-B gauges
disappear way off the right-hand side of the plot.
The straight portions of stirrups-C, -B, & -A showed similar beha-
viour over the course of the test ahead of peak load as indicated by the
slopes of the strain curves. These strains increased somewhat pro-
portionately as shear cracks developed and coalesced into the critical
shear crack. Otherwise, the rapid coalescence and growth of the critical
shear crack just ahead of peak load more heavily influenced the stirrup-
A bend and stirrup-B straight & bend gauges.
Overall the strain plot tends to suggest that GFRP stirrup bend
strength may not be a critical factor in the failure of deep beams;
however, that depends on their location relative to the formation of the
critical shear crack and direct-strut.

3.4. Stirrup condition post testing

Fig. 9. Crack diagram for BM12-150 at peak load with overlay of stirrups,
Beam specimens that potentially contained ruptured or damaged
strain gauges, and strut.
stirrups were investigated further by extracting the stirrups from the
beam. The investigation found no ruptured stirrups; however, two
beams with stirrups. stirrups showed signs of distress and are shown in Fig. 10. These stir-
Fig. 9 is a picture of the same beam at peak load and is overlaid with rups were taken from the failed shear spans of BM12-220 and BM12-
several features. The vertical dashed lines show the location of the 150. The pictured flexural tension zone bends of each stirrup contained
stirrups and the red dots show the location of the straight and bent a crack that ran through the bend in a plane perpendicular to the plane
portion gauges. A visualization of both the bottle shaped and linearized of the stirrup; the cracks followed the undulations of the fibers along
strut is also shown for reference. The critical shear crack shown in Fig. 9 the central axis of the bar. The authors are not aware of similar ob-
was on the instrumented south side of the beam, which allows for a servations having been made elsewhere and the extraction of non-
comparison with the strains plotted in Fig. 8. The labels A, B, & C in ruptured stirrups may be unique to this research programme.
Figs. 8 and 9 map stirrup strains to their location in the beam and re-
lative to cracking.
4. Analysis of experimental results
For all beams, generally, the stirrups closest to mid-span started
straining first. In each case, the straight portions experienced increases
In this section, beam behaviour is analyzed by decomposition of the
in strain after inclined shear cracks formed as did the bent portions;
experimental strength into concrete and stirrup contributions, VC and

Fig. 10. Stirrups with crack in flexural tension zone post test.

8
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

VS, respectively. Such an exercise is common for slender beams;

nevertheless, the results are insightful despite the specimens in this
study behaving like deep beams. The decomposition conforms to the
typical VE = VC + VS formulation for determining shear strength
common to many codes and standards.
In the following discussion, the shear strength, VE, of the test spe-
cimens was assumed to be equal to half of the peak load, PE, from
testing, (VE = 0.5 * PE). Peak load values are reported in Table 3, shear
strength values are reported in Table 4.

4.1. Determining the concrete contribution, VC

Two methods are presented for determining concrete contribution

to shear strength. The first method assumes the concrete contribution is
equal to the load at which the first flexural-shear crack formed. The
second method assumes that it is equal to a maximum value from a
concrete contribution curve that was back calculated from stirrup
strains. These two methods are discussed below. Fig. 11. Determining VC from load vs stirrup strain plots, BM16-s230 shown.

4.1.1. First flexural-shear crack method

In this method, the concrete contribution to shear strength was
taken as the load at which the first flexural-shear crack formed. A plot
showing the applied shear load against stirrup strains for both the bent
and straight portions was examined to make this determination, see
Fig. 11. The concrete contribution, VC, was set equal to the lowest load
on the plot corresponding to the point at which the strain curves show a
sharp change in slope. Photographs taken at discrete loads over the
course of the experiment were consulted to confirm the probable for-
mation of a flexural-shear crack at that load level.
Fig. 11 shows an example of how to determine the concrete con-
tribution for BM12-s230. This beam had four gauges because it had two
instrumented stirrups in the shear span, each with a gauge at the bent
and straight portions. The arrows in the plot show four possible loca- Fig. 12. Determining VC from shear load vs time plots, BM16-s230 shown.
tions where VC could be determined, but the lowest of the four was
selected. In this case, VC = 70.5 kN. Vi = 0.5Pi (2)

4.1.2. Stirrup strain method VS, i = εavg, i EnAv (3)

In this method, VC is back calculated by subtracting the average load
where VC,i is the concrete contribution for data point i, Vi is the applied
in the stirrups over the shear span, using the stirrup strains, from the
shear, VS,i is the stirrup contribution, Pi is the applied load, εavg,i is the
applied shear load. This computation is done for every measured point
average strain in the straight portion strain gauges for all engaged
over the course of the test, which creates a set of concrete and stirrup
stirrups in the shear span, E is the elastic modulus of the straight portion
contribution curves as shown in Fig. 12. The concrete contribution, VC,
of the stirrup leg from Table 2, n is the number of engaged stirrup legs
is then set equal to the peak of the concrete contribution curve [19,20].
in the shear span, and Av is the cross-sectional area of one stirrup leg
The computations are described by Eqs. (1)–(3).
from Table 2.
VC, i = Vi − VS, i (1) Engaged stirrups were determined by examination of the load vs

Table 4
Summary of VE, VC, VS.
Flexural-shear crack method Stirrup strain method

Beam ρv, % ρf, % VE, kN VC, kN VS, kN VC/VE, – VC, kN VS, kN VC/VE, –

BM12-INF – 2.51 81.6

BM16-INF – 2.23 75.1
BM25-INF – 1.82 62.5

BM12-220 0.51 2.51 191.2 60.0 131.2 0.31 84.8 106.4 0.44
BM16-220a 0.51 2.23 154.6 56.5 98.1 0.37 95.5 59.2 0.62
BM25-220 0.51 1.82 180.0 63.5 116.5 0.35 53.8 126.2 0.30

BM12-150 0.75 2.51 202.6 70.0 132.6 0.35 57.6 145.0 0.28
BM16-150 0.75 2.23 208.3 58.5 149.8 0.28 58.4 149.9 0.28
BM25-150 0.75 1.82 207.9 73.5 134.4 0.35 57.9 150.0 0.28

BM12-s230 1.19 2.18 233.5 67.0 166.5 0.29 68.5 165.0 0.29
BM16-s230 1.19 1.94 225.4 70.5 154.9 0.31 66.3 159.1 0.29
BM25-s230 1.19 1.58 222.0 69.5 152.5 0.31 55.5 166.5 0.25

Note: The BMX-INF group of beams are shown for reference.

a
Outlier, see the Results Section.

9
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

analysis found that the peak on the concrete contribution curve oc-
curred at 44 percent of peak load or lower for the specimens in this test
programme, which assuaged concerns about the applicability of this
method given some of the gauge failures that occurred within the vi-
cinity of peak load or those occurring in the un-instrumented span.

4.2. Concrete and stirrup contribution summary

Findings from the concrete contribution investigation are summar-

ized in Table 4, where VE is the applied shear at failure, VC is the
concrete contribution, and VS is the stirrup contribution calculated as
VS = VE − VC.
Generally, the results of the two methods are similar. The flexural-
shear crack method shows concrete contributions between 28 and 35
percent and the stirrup strain method between 25 and 44 percent, ex-
Fig. 13. Variation in VC against flexural reinforcement ratio, ρf. cluding the outlier, BM16-220. For the stirrup strain method, the con-
crete contribution was commonly lower than the strength of the com-
panion beam in the group of beams without shear stirrups, BMX-INF.
For the proceeding discussion, the concrete contribution, VC, from
the flexural-shear crack method will be used, because the logic re-
garding mechanical behaviour behind this method is more readily
identifiable than that of the stirrup strain method. Regardless, the re-
sults of both methods are similar and would show the same trends in the
following discussion.

4.3. Comparison and analysis of shear contributions

Figs. 13 and 14 show the variation in concrete contribution, VC,

against the flexural and shear reinforcement ratios, ρf & ρv, respectively.
The results for the beam series with no stirrups, BMX-INF, are included
in the plot for comparison purposes and were plotted assuming the
concrete contribution was equal to their shear strength at failure.
Fig. 13 shows no correlation between the concrete contribution and
Fig. 14. Variation in VC against shear reinforcement ratio, ρv. flexural reinforcement ratio for beams with stirrups. This is because
beams exhibited deep beam behaviour where shear force is primarily
transferred by direct-strutting action. The beams without stirrups, BMX-
INF, show a clear correlation as they failed in a form of shear tension.
Fig. 14 shows no clear trend between the concrete contribution and
shear reinforcement ratio for beams with stirrups. Thus, increases in
shear reinforcement ratio may have little impact on the load level at
which flexural-shear cracks form, which was the basis for determining
VC in the current analysis.
Fig. 15 shows the variation in stirrup contribution, VS, against shear
reinforcement ratio. This figure shows that an increase in the shear
reinforcement ratio increases the stirrup contribution to shear strength.
The data also shows that the increase in stirrup contribution to shear
strength diminishes with increases in shear reinforcement ratio and that
the scatter of the data also diminishes. Specific to this experimental
investigation, stirrups, as a shear resisting mechanism, become in-
crementally less effective when additional stirrups are added to the
span, going from BMX-220 (ρv = 0.51%) to BMX-150 (ρv = 0.75%), or
when stirrup area is increased in the span, going to BMX-s230
Fig. 15. Variation in VS against shear reinforcement ratio, ρv. (ρv = 1.19%).
A normalized stirrup stress, fS, is used to explore this behaviour
further and was calculated as the average stress in the engaged stirrups
stirrup strain plots. Stirrups that did not show any significant strain
from the stirrup contribution (fS = VS/(nAv), where n and Av are as
over the entire test were not included in the average or count, such as
defined in Eqs. (1)–(3)). This is done to assess the influence that stirrup
the one at mid-span and any that were immediately adjacent to the
quantity, expressed via the area of engaged stirrups in the span, may
bearing plate. In the case of the present experimental programme, the
have on the trend identified in Fig. 15. The results are shown in Fig. 16.
stirrups were fitted with two gauges. For each stirrup in the average, the
Fig. 16 shows that normalized stirrup stress decreases as the area of
gauge used was the one that had the largest strains overall during the
stirrups in a span increases, though local stirrup stress/strains vary from
test. In every case but one, that was the straight portion gauge. The
the mean, see Fig. 8. When compared with Figs. 14 and 15, the trend
outlier was found on BM16-150 on a stirrup close to the support and
confirms that even though load carried in the span increases as shear
may be due to experimental error [19].
reinforcement increases, with that increase being provided by the
Fig. 12 shows an example of determining the concrete contribution,
stirrup contribution, the normalized stress in the stirrups also dimin-
VC, from the curve for BM16-s230 and in this plot VC = 66.3 kN. The
ished. Here, stirrups become a less effective shear resisting mechanism

10
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

shear-span-to-depth ratio, a/d.

Fig. 17 shows that beam shear strength is correlated with both the
flexural and shear reinforcement ratios and that strength increases with
increases in either ratio. It also shows diminishing increases in shear
strength with increases in the shear reinforcement ratio and mitigates
concerns about comparing thin stirrups to thick stirrups. The curves
from Nehdi et al. [17] are also included to clarify the experimental
trends. Note that the beams in the BMX-s230 group had slightly lower
flexural reinforcement ratios than their counterparts in the other groups
due to their slightly larger cross-sections and the dotted line is used for
that distinction.
A non-linear finite element model of these beam specimens using
the concrete damage plasticity model was completed by Joe Stoner
[23], who was able to adequately reproduce the observed behaviour for
beams reinforced in shear with brittle type reinforcements. That author
then conducted a parametric study on in the variation in shear-span-to-
depth ratio, a/d.
Fig. 16. Variation in normalized stirrup stress fS against shear reinforcement
ratio, ρv. 5. Conclusions

as more stirrups are, or stirrup area is, added to the span; an increase of The paper presents the results of an experimental programme on the
802 mm2 (178%), per shear span, when going from BMX-220 effectiveness of GFRP, brittle-type, stirrup reinforcement on load car-
(ρv = 0.51%) to BMX-s230 (ρv = 1.19%) only increased the average rying capacity of concrete beams and their general behaviour. Twelve
shear resistance, VE, by 82.4 kN (22%). concrete beams were studied with varying amounts of shear and flex-
In this experiment, the closed loop stirrups increased shear strength ural reinforcements.
in two ways: first, by carrying load across crack faces through the The primary conclusions are as follows:
stirrups legs and increasing aggregate interlock, and second, by con-
fining the direct-strut acting between load plate and support plate. The (1) Beams with the same stirrup arrangement, but different long-
confinement of concrete in the direct-strut increased its compressive itudinal reinforcement ratios, failed in shear at similar applied
strength, which is its resistance to crushing failure, by restricting the loads. The experimental results, when compared alongside the
growth of cracks. The complex behaviour of confining concrete at dis- Nehdi et al. [17] optimized equations for shear strength, tend to
crete locations with closed loop stirrups does not result in a linearly confirm that the shear resistance of a beam increases with both the
proportional increase in shear strength for increases in stirrup area. longitudinal and shear reinforcement ratios.
(2) Beams without stirrups failed in shear tension and showed in-
creasing shear strength with increasing longitudinal reinforcement
4.4. Comparison of experimental results with Nehdi et al. Model ratio. The trend is due to the dowel action of the longitudinal re-
inforcement acting as the primary shear resisting mechanism in
Fig. 17 shows the primary experimental result alongside the main conjunction with the aggregate interlock & shear friction me-
experimental parameters. The figure shows the variation in the product chanism.
of peak load and flexural reinforcement ratio against shear reinforce- (3) Beams with stirrups failed in shear compression with arching /
ment ratio. The plot also includes curves produced using the Nehdi direct-strutting action as the primary shear resisting mechanism.
et al. [17] ‘optimized equations’ to highlight the important trends in the Arch action plots confirmed the formation of a tied arch me-
data. The Nehdi et al. model was trained on a database of FRP re- chanism. These beams showed a degree of post peak deformability
inforced beams including both deep and slender specimens and is under displacement controlled loading with increased deform-
capable of predicting the strength of beams with wide variation in ability for higher shear reinforcement ratios. However, two

Fig. 17. Variation in product of peak load and flexural reinforcement ratio against shear reinforcement ratio.

11
M. Krall and M.A. Polak Engineering Structures 198 (2019) 109333

specimens failed abruptly after peak load. [6] Currier J, Fogstad C, Walrath D, Dolan C. Bond development of thermoplastic FRP
(4) Stirrup strains were influenced by the formation of the strut. shear reinforcement stirrups. In: Basham KD, editor. Proc of the ASCE 3rd material
eng conf, San Diego, CA. New York, NY: ASCE; 1994. p. 592–7.
Stirrups near the middle of the strut, where crack growth devel- [7] Ueda T, Sato Y, Kakuta Y, Imamura A, Kanematsu H. Failure criteria for FRP rods
opment is largest, showed the largest strains. Two stirrups extracted subjected to a combination of tensile and shear forces. In: Taerwe L, editor. Proc of
from specimens showed cracking along the centroid of the bar in a the 2nd int RILEM symp on non-metallic (FRP) reinf for conc struct, Ghent, Belgium.
E & FN Spon, London, United Kingdom; 1995. p. 26–33.
plane perpendicular to the axis of the stirrup. The cracks followed [8] Ehsani MR, Saadatmanesh H, Tao S. Bond of hooked Glass Fiber Reinforced Plastic
the undulations of the fibres and were located in the flexural ten- (GFRP) reinforcing bars to concrete. ACI Mater J 1995;92(4):391–400.
sion zone bends. [9] Nakamura H, Higai T. Evaluation of shear strength of concrete beams reinforced
with FRP. (English Translation) Concrete Library International of Japanese Society
(5) The experimental strength was decomposed into separate stirrup of Civil Engineers (JSCE) 1995; 26: 111–23.
and concrete contributions. The concrete contribution was not in- [10] Ahmed EA, El-Salakawy EF, Benmokrane B. Performance evaluation of glass fiber-
fluenced by either the longitudinal or shear reinforcement ratio for reinforced polymer shear reinforcement for concrete beams. ACI Struct J
2010;107(1):53–62.
the specimens studied. The stirrup contribution increased with in-
[11] Ahmed EA, El-Salakawy EF, Benmokrane B. Shear performance of RC bridge girders
creasing shear reinforcement ratio but diminished non-linearly; reinforced with carbon FRP stirrups. J Bridge Eng 2010;15(1):44–54. https://doi.
stirrup contribution increases occurred regardless of whether the org/10.1061/(ASCE)BE.1943-5592.0000035.
ratio increased as a result of additional stirrups in the span or as a [12] Bentz EC, Massam L, Collins MP. Shear strength of large concrete members with
FRP reinforcement. J Compos Const 2010;14(6):637–46. https://doi.org/10.1061/
result of stirrups with larger cross-sections. The stirrup contribution (ASCE)CC.1943-5614.0000108.
was a measure of the load carried by the stirrups across crack faces [13] American Concrete Institute (ACI). Guide for the design and construction of
and the confinement effect the stirrups provided to the concrete structural concrete reinforced with Fiber-Reinforced Polymer (FRP) Bars (ACI 440.
1R-15). Farmington Hills, Michigan, USA: American Concrete Institute; March
direct-strut. 2015.
[14] Imjai T, Guadagnini M, Pilakoutas K. Mechanical performance of curved FRP rebars
Acknowledgements – Part II: Analytical study. In: Smith ST, editor. Proc of the 1st Asia-Pacific conf on
FRP in struct (APFIS), Hong Kong, China; International Institute for FRP in
Construction (IIFC), Hong Kong, China; 2007. p. 339–44.
The authors would like to thank the Ontario Centres of Excellence, [15] Kanematsu H, Sato Y, Ueda T, Kakuta Y. A study on failure criteria of FRP rods
Natural Sciences and Engineering Research Council of Canada, and subject to tensile and shear force. Proc of FIP symposium ’93 modern prestressing
techniques and their applications. Kyoto, Japan: Japan Prestressed Concrete
Schoeck Canada (Fiberline Composites) for their financial contributions Engineering Association; 1993. p. 743–50.
to this research. [16] Canadian Standards Association (CSA). Design and construction of building struc-
tures with fibre-reinforced polymers (CAN/CSA S806-12). Mississauga, Ontario,
Canada: Canadian Standards Association; March 2012.
Funding Sources
[17] Nehdi M, El Chabib H, Aly Said A. Proposed shear design equations for FRP-re-
inforced concrete beams based on genetic algorithms approach. J Mater Civil Eng
This work was supported by the Ontario Centres of Excellence, 2007;19(12):1033–42. https://doi.org/10.1061/(ASCE)0899-1561(2007)
Natural Sciences and Engineering Research Council of Canada, and 19:12(1033).
[18] Machial R, Shahria Alam M, Rteil A. Revisiting the shear design equations for
Schoeck Canada. concrete beams reinforced with FRP rebar and stirrup. Mater Struct
2012;45(11):1593–612. https://doi.org/10.1617/s11527-012-9859-5.
References [19] Krall MD. Tests on concrete beams with GFRP flexural and shear reinforcements &
analysis method for indeterminate strut-and-tie models with brittle reinforcements.
MASc Thesis: Department of Engineering Civil and Environmental Engineering,
[1] Shehata E, Morphy R, Rizkalla S. Fibre reinforced polymer shear reinforcement for University of Waterloo, Waterloo, Ontario, Canada; 2014.
concrete members: behaviour and design guidelines. Can J Civil Eng [20] Nakamura E, Watanabe H. Shear resisting mechanism in RC beams with fractured
2000;27(5):859–72. https://doi.org/10.1139/l00-004. stirrups. In: Proc of the 22nd UW – Japan bridge engineering workshop, Seattle,
[2] Imjai T, Guadagnini M, Pilakoutas K. Mechanical performance of curved FRP rebars Washington, USA; 2006.
– Part I: Experimental study. In: Smith ST, editor. Proc of the 1st Asia-Pacific conf on [21] Andermatt MF, Lubell AS. Behavior of concrete deep beams reinforced with internal
FRP in struct (APFIS), Hong Kong, China; International Institute for FRP in fiber-reinforced polymer—experimental study. ACI Struct J 2013;110(4):585–94.
Construction (IIFC). Hong Kong, China; 2007. p. 333–8. [22] Farghaly AS, Benmokrane B Shear. Behaviour of FRP-reinforced concrete deep
[3] Ahmed EA, El-Sayed AK, El-Salakawy EF, Benmokrane B. Bend strength of FRP beams without web reinforcement. J Compos Constr 2013;17(6). https://doi.org/
stirrups: comparison and evaluation of testing methods. J Compos Const 10.1061/(ASCE)CC.1943-5614.0000385.
2010;14(1):3–10. https://doi.org/10.1061/(ASCE)CC.1943-5614.0000050. [23] Stoner J. Finite element modelling of GFRP reinforced concrete beams. MASc
[4] Maruyama T, Honma M, Okamura H. Experimental study on tensile strength of bent Thesis: Department of Engineering Civil and Environmental Engineering, University
portion of FRP rods. Spec Publ ACI 1993;138:163–76. of Waterloo, Waterloo, Ontario, Canada; 2015.
[5] Nagasaka T, Fukuyama H, Tanigaki M. Shear performance of concrete beams re-
inforced with FRP stirrups. Spec Publ ACI 1993;138:789–812.

12