You are on page 1of 22


Title no. 106-S66

Punching Strength of Reinforced Concrete Footings

by Josef Hegger, Marcus Ricker, and Alaa G. Sherif

A total of 17 reinforced concrete footings were tested to investigate

the punching shear behavior of footings. The test parameters
investigated are the shear span-depth ratio (a/d), concrete
strength, and punching shear reinforcement. The (a/d) ranged
between 1.25 and 2.0, whereas the concrete strength ranged
between 20 and 40 MPa (2.9 and 5.8 ksi). To study the effect of
soil-structure interaction, five footings were realistically supported
on sand. The remaining specimens were supported on a column
stub and a uniform surface load was applied. The present experimental
investigations indicated that the angle of the failure shear crack is
steeper in punching tests on compact footings than observed in
tests on more slender slabs. Furthermore, the (a/d) significantly
affects the punching shear capacity. Based on the test results, the ACI
and Eurocode 2 provisions are critically reviewed and improvements
are proposed.

Keywords: footings; punching shear; reinforced concrete; shear span-

depth ratio; soil pressure redistribution; soil-structure interaction.

The problem of punching of reinforced concrete slabs has
been dealt with extensively in literature.1,2 Tests on reinforced
concrete footings, however, are still limited.3-6 Furthermore,
most of the available tests have unrealistic test setups. As a
consequence, the design of reinforced concrete footings is
mainly based on test results of slabs and most design codes
do not distinguish between slabs and footings in the design
rules. Hence, there is a need for a systematic experimental
Fig. 1—Critical perimeters according to: (a) ACI 318-08;
investigation on reinforced concrete footings. Therefore,
and (b) Eurocode 2.
punching tests on 17 quadratic reinforced concrete footings
were performed. The aim of this investigation is to study the
main parameters assumed to affect the punching shear ACI 318-08
strength of footings such as shear span-depth ratio (a/d), The critical section is at d/2 from the column face (Fig. 1 (a)).
concrete compression strength, shear reinforcement, and The design is based on
soil-structure interaction.
vu < φvn (1)
The punching shear capacities of footings predicted by where φ is a strength reduction factor (0.75 for shear); vu is
various codes vary significantly. By testing 17 footings, the the applied factored shear stress, using load factors 1.2 and
main parameters affecting the punching strength are 1.6 for dead and live loads; and vn is the nominal shear resistance.
systematically investigated. The parameters studied are The applied shear stress due to factored concentric shear force
a/d, concrete compressive strength, punching shear Vu is calculated as
reinforcement, and the soil-structure interaction. The provisions
of ACI 318-087 and Eurocode 28 are evaluated by comparing vu = Vu/(b0d) (2)
with the experimental results.
where b0 is the perimeter of the critical section, and d is the
DESIGN CODES distance from the extreme compression fiber to the centroid
In general, design codes do not differentiate between the
punching shear strength of flat plates and footings. The ACI Structural Journal, V. 106, No. 5, September-October 2009.
codes allow a part of the soil reaction to be subtracted from MS No. S-2008-305 received September 15, 2008, and reviewed under Institute
publication policies. Copyright © 2009, American Concrete Institute. All rights reserved,
the punching load. The amount to be deducted, however, including the making of copies unless permission is obtained from the copyright proprietors.
Pertinent discussion including author’s closure, if any, will be published in the July-
differs from one code to the other. August 2010 ACI Structural Journal if the discussion is received by March 1, 2010.

706 ACI Structural Journal/September-October 2009

ACI member Josef Hegger is a Professor at the Institute of Structural Concrete,
A v f yt
Rheinisch-Westfälische Technische Hochschule Aachen (RWTH) University, Aachen,
v s = -----------
- (9)
b0 s
Germany. He received his PhD from the Braunschweig University of Technology,
Braunschweig, Germany, in 1985. His research interests include bond behavior, shear
capacity, high-performance concrete, textile-reinforced concrete, and composite structures.
where Av is the area of shear reinforcement in one row around
Marcus Ricker is a Research Engineer at the Institute of Structural Concrete, RWTH. the column, s is the spacing of the shear reinforcement, and fyt
He received his Diploma degree in structural engineering from the Darmstadt is the yield strength of the shear reinforcement not to exceed
University of Technology, Darmstadt, Germany, in 1998. His research interests
include the punching behavior of footings and flat plates. 413 MPa (60,000 psi). The maximum allowed shear stress
vmax is determined as
ACI member Alaa G. Sherif is a Professor in the Civil Engineering Department,
Helwan University, Mataria-Cairo, Egypt. He received his BSc from Cairo University,
Giza, Egypt, in 1987, and his MSc and PhD from the University of Calgary, Calgary,
AB, Canada, in 1991 and 1996, respectively. He is a member of Joint ACI-ASCE Committee vmax = 0.5 f c′ MPa = 6 f c′ psi for stirrups (10)
352, Joints and Connections in Monolithic Concrete Structures. His research interests
include the design and serviceability of reinforced concrete structures.
vmax = 0.67 f c′ MPa = 8 f c′ psi for studs (s ≤ 0.5d) (11)
of tension reinforcement (effective depth). The shear
resistance of the concrete vc is the smallest value obtained Outside the shear-reinforced zone, the shear stress resistance
from Eq. (3), (4), and (5) of the concrete is limited to the one-way shear strength value of

vc = 0.17 ⎛ 1 + -----⎞ λ f c′ MPa = ⎛ 2 + -----⎞ λ f c′ psi

2 4

(3) vc = 0.17λ f c′ MPa = 2λ f c′ psi (12)
β c⎠ ⎝ β c⎠

αs d αs d Eurocode 2
vc = 0.083 ⎛ --------
- + 2⎞ λ f c′ MPa = ⎛ --------
- + 2⎞ λ f c′ psi (4) For concentric loading, the maximum shear stress vEd is
⎝ b0 ⎠ ⎝ b0 ⎠
calculated as

vc = 0.33λ f c′ MPa = 4λ f c′ psi (5) V Ed, red

v Ed = ----------------
- (13)
where αs is a parameter taken as 40 for interior, 30 for edge,
and 20 for corner columns; βc is the ratio of long to short side
where VEd, red is the net applied shear force, u is the control
of concentrated load or reaction area; λ is a factor accounting
for the concrete density; and b0 is the perimeter of the critical perimeter considered, and d is the effective depth. For
section in Fig. 1(a). For slabs without shear reinforcement, concentric loading, VEd, red is calculated as
the nominal shear resistance vn in Eq. (1) is vc. The isolated
footing may be assumed to be rigid, resulting in a uniform VEd,red = VEd – ΔVEd (14)
soil pressure for concentric loading. The shear force can
be reduced by the effective soil pressure within the where VEd is the column load and ΔVEd is the net upward
control perimeter. force within the control perimeter considered, that is, upward
If vu > φvc, shear reinforcement has to be used. Two critical uniform pressure from soil minus self-weight of footing.
sections are to be checked: d/2 from the column face and d/2
The punching resistance should be verified at control
from the outer shear reinforcement (Fig. 1(a)). The punching
perimeters within 2.0d from the periphery of the column
shear resistance inside the shear-reinforced zone is calculated as
(Fig. 1(b)). The lowest value of resistance at the different
sections controls the design. The punching shear stress resistance
vn = vcs + vs ≤ vmax (6) of the concrete is calculated as

where vcs is the shear stress resisted by the concrete inside

1⁄3 2d 2d
the shear-reinforced zone, vs is the shear stress resisted by v Rd, c = C Rd, c k ( 100ρ ⋅ f ck ) ---------- ≥ v min ---------- (15)
the shear reinforcement, and vmax is the maximum allowed a crit a crit
shear stress.
Acknowledging the superior anchorage performance of where CRd,c = 0.18/γc is an empirical factor derived from a
shear studs, ACI 318-08 distinguishes between shear studs regression analysis, with γc being the material resistance factor for
and stirrups as shear reinforcement. The nominal shear concrete (1.5); d is the effective depth; k = 1 + 200 ⁄ d ≤ 2.0 is
strength provided by concrete vcs inside the shear-reinforced the size factor of the effective depth (with d in mm); ρ is the
zone is calculated as flexural reinforcement ratio; fck is the characteristic cylinder
compressive concrete strength; acrit is the distance from the
vcs = 0.17λ f c′ MPa = 2λ f c′ psi for stirrups (7) column face to the control perimeter considered; and
v min = 0.035 · k 3/2· f ck1/2 is the minimum shear capacity
of the concrete, including the material resistance factor for
vcs = 0.25λ f c′ MPa = 3λ f c′ psi for studs (8) concrete γc = 1.5.
If vEd > vRd,c , shear reinforcement will be required. The
The nominal shear strength provided by shear reinforcement design of the shear reinforcement is based on the following
vs is calculated as expression

ACI Structural Journal/September-October 2009 707

Table 1—Details of test specimens
d, c, b, fc′, Ec , ρ, Av, s, Vservice , Vtest , Vflex ,
Test mm (in.) mm (in.) mm (in.) a/d MPa (ksi) MPa (ksi) % cm2 (in.2) mm (in.) kN (kips) kN (kips) kN (kips)
DF6 395 (15.6) 200 (7.9) 1200 (47.2) 1.27 19.0 (2.8) 23,450 (3401) 0.87 — — 1008 (227) 2836 (638) 7146 (1607)
DF7 395 (15.6) 200 (7.9) 1400 (55.1) 1.52 20.9 (3.0) 25,067 (3636) 0.87 — — 980 (220) 2569 (578) 6544 (1471)
DF8 250 (9.8) 200 (7.9) 1200 (47.2) 2.00 22.5 (3.3) 24,850 (3604) 0.88 — — 540 (121) 1203 (270) 3028 (681)
DF9 250 (9.8) 200 (7.9) 1200 (47.2) 2.00 20.8 (3.0) 24,700 (3582) 0.89 28.4 (4.4) 90 (3.5)* 1008 (227) 2784 (626) 3069 (690)
DF10 250 (9.8) 200 (7.9) 1200 (47.2) 2.00 38.1 (5.5) 29,500 (4278) 0.91 — — 864 (194) 1638 (368) 3179 (715)
DF11 395 (15.6) 200 (7.9) 1200 (47.2) 1.27 21.4 (3.1) 22,000 (3191) 0.87 — — 1200 (270) 2813 (632) 6845 (1539)
DF12 395 (15.6) 200 (7.9) 1400 (55.1) 1.52 21.2 (3.1) 23,700 (3437) 0.88 — — 900 (202) 2208 (496) 6732 (1513)
DF13 395 (15.6) 200 (7.9) 1800 (70.9) 2.03 21.1 (3.1) 20,100 (2915) 0.87 — — 800 (180) 1839 (413) 6004 (1350)
DF14 295 (11.6) 200 (7.9) 1400 (55.1) 2.00 21.2 (3.1) 23,700 (3437) 0.88 — — 600 (135) 1478 (332) 3766 (847)
DF15 470 (18.5) 200 (7.9) 1400 (55.1) 1.28 21.7 (3.2) 22,500 (3263) 0.85 — — 1700 (382) 2750 (618) 8996 (2022)
DF16 395 (15.6) 200 (7.9) 1200 (47.2) 1.27 20.0 (2.9) 20,900 (3031) 0.87 45.2 (7.0) 190* (7.5) 1800 (405) 3680 (827) 7114 (1599)
DF17 395 (15.6) 200 (7.9) 1400 (55.1) 1.52 20.8 (3.0) 22,000 (3191) 0.87 45.2 (7.0) * 1600 (360) 3619 (814) 6570 (1477)
190 (7.5)
DF18 395 (15.6) 200 (7.9) 1800 (70.9) 2.03 21.7 (3.2) 22,500 (3263) 0.87 45.2 (7.0) * 1500 (337) 3361 (756) 6026 (1355)
190 (7.5)
DF19† 395 (15.6) 200 (7.9) 1200 (47.2) 1.27 21.8 (3.2) 21,900 (3176) 0.87 — — 1600 (360) 2790 (627) 7132 (1603)
DF20 395 (15.6) 200 (7.9) 1200 (47.2) 1.27 35.7 (5.2) 27,400 (3974) 0.87 — — 1600 (360) 3037 (683) 7455 (1676)
DF21 395 (15.6) 200 (7.9) 1400 (55.1) 1.52 36.3 (5.3) 26,800 (3887) 0.87 — — 1200 (270) 2860 (643) 6914 (1554)
DF22 395 (15.6) 200 (7.9) 1800 (70.9) 2.03 36.4 (5.3) 26,200 (3800) 0.87 — — 1000 (225) 2405 (541) 6338 (1425)
*Spacing between column face and first row of shear reinforcement was s0 = 75 mm (3.0 in.) for Specimen DF7 and s0 = 120 mm (4.7 in.) for DF16, DF17, and DF18.

Specimen included seven circular ties, which confine compression zone near column face (refer to text).
Note: d is effective depth; c is column dimensions; b is footing dimension; a is footing dimension measured from face of column; fc′ is cylinder concrete compression strength;
Ec is Young’s Modulus of concrete; ρ is flexural reinforcement ratio; Av is area of one leg of stirrups along a peripheral line; s is spacing between two peripheral lines of stirrups;
Vservice is estimated service load; Vtest is ultimate failure load; and Vflex is shear force that produces flexural failure according to yield-line theory.9 Footings with shear reinforcement
included layer of top reinforcement of ρ′ = 0.49% (DF9) and ρ′ = 0.31% (DF16, DF17, and DF18).

v Rd, cs = 0.75v Rd, c + 1.5 ---- A sw f ywd, ef ------ sin α (16) The experimental program included 17 footings. The
sr ud
dimensions of the test specimens were chosen to model 1/2
to 1/3 scale of a common footing. The specimens were
where Asw is the area of the shear reinforcement in one row divided in two series. The first series included five footings
around the column; sr is the radial spacing of perimeters of (DF6 to DF10) that were realistically supported on sand.
shear reinforcement; u is the control perimeter within 2.0d These footings were to supplement previous tests.6 The
from the column face (Fig. 1(b)); fywd,ef is the effective second test series included 12 footings (DF11 to DF22) that
design strength of the punching shear reinforcement, were supported on a column stub and a uniform load was
according to fywd,ef = 250 + 0.25d (with d in mm) ≤ fywd (MPa); applied to the footing. The test parameters included the a/d
and α is the angle between the shear reinforcement and the ratio, the concrete compressive strength, the shear reinforcement,
plane of the slab. and the soil-structure interaction.
The maximum punching shear resistance (at the column
face) is limited to a maximum of
Material properties
For the footings of Series I, the concrete was mixed at the
f ck ⎞
v Rd, max = 0.5 ⋅ 0.6 ⋅ ⎛ 1 – --------
- ⋅ f (with fck in MPa) (17) laboratory; for Series II, commercial ready mixed concrete
⎝ 250⎠ cd was used. The maximum coarse aggregate size was 16 mm
(0.63 in.) in all footings. Ordinary CEM III A 32.5 N portland
where fcd = αcc fck /γc is the design concrete compressive cement and a water-cement ratio (w/c) of 0.50 were used,
strength with αcc = 1.0 being a coefficient taking account for resulting in a slump of approximately 480 mm (18.9 in.).
long-term effects. The control perimeter at which shear The concrete mixtures were designed to produce a 28-day
reinforcement is not required (Fig. 1(b)) is determined by target strength of fc′ = 20 and 40 MPa (2.9 and 5.8 ksi).
German steel BSt 500 (A), with the measured yield stress
fsy = 552 MPa (80.1 ksi), tensile strength fsu = 634 MPa
V Ed (92.0 ksi), and a Young’s modulus of 200,000 MPa (29,000 ksi),
u out = ------------------
- (18)
v Rd, c ⋅ d was used for all reinforcement. Table 1 summarizes the
properties of the materials used.
The outermost perimeter of shear reinforcement should be
placed at a distance not greater than 1.5d within uout (Fig. 1(b)). Test specimens
The shear reinforcement should be provided in at least two The dimensions of the footings were 1200, 1400, and 1800 mm
rows and the spacing of the perimeters should not exceed (47.2, 55.1, and 70.9 in.) in square with the thickness of the
0.75d. The distance between the column face and the first footings varying between 300 and 530 mm (11.8 and 20.9 in.).
row of shear reinforcement should not exceed s0 = 0.5d The effective depth d varied between 250 and 470 mm (9.8 and
(Fig. 1(b)). 18.5 in.), resulting in a/d of approximately 1.25, 1.5, and 2.0. The

708 ACI Structural Journal/September-October 2009

Fig. 3—Test setups.

Fig. 2—Layout of flexural and shear reinforcement for test

Specimen DF9.

dimensions and reinforcement details of a typical test specimen

are shown in Fig. 2. The flexural reinforcement ratio varied Fig. 4—Arrangement of soil pressure gauges and steel
between 0.85 and 0.91%. The 200 x 200 mm (7.8 x 7.8 in.) strain gauges.
square column stubs were cast monolithically at the center
of the slabs using high-strength concrete and were reinforced column stub. The jacks were linked to a common manifold
with 10 mm (0.4 in.) steel plates to prevent a premature failure. and applied the same load independent of the displacement.
Full details of the footings are given in Table 1. Footings DF9 The soil pressure distribution was measured by 21 electric
and DF16 to DF18 included heavy shear reinforcement and stress sensors with electric/hydraulic pressure gauges. In
were designed to examine the maximum punching capacity. addition, thin-film pressure sensors were used to measure the
The shear reinforcement consisted of vertical stirrups with a soil pressure distribution. The main advantage of the sensors
diameter of 10 mm (0.4 in.) and a yield strength fyw of 520 MPa used is that this system allows a two-dimensional measurement
(75.4 ksi) in Specimen DF9 and 12 mm (0.5 in.) and fyw of of the soil pressure distribution whereas the soil pressure
560 MPa (81.2 ksi) in Specimens DF16 to DF18. The spacings gauges can only measure the soil pressure at definite points
of the stirrups are given in Table 1. As an example, the layout of (Fig. 4(a)).
the shear reinforcement for DF9 is shown in Fig. 2. Specimen Series II (DF11 to DF22)—The specimens of test Series II
DF19 included seven circular ties with a diameter of 400 mm were loaded by a uniform surface load using the test setup
(15.7 in.), which were horizontally arranged underneath the shown in Fig. 3(b). The footings were tested upside down.
column stub to confine the compression zone near the column The load was applied by eight hydraulic jacks (maximum
face. The bar diameter was 12 mm (0.5 in.) and the spacing was capacity 4720 kN [1061 kips]). Each jack transferred its load
40 mm (1.6 in.). The circular ties did not increase the via steel beams to two polytetrafluoroethylene (PTFE)-coated
punching shear strength compared to a similar footing sliding bearings of dimensions 140 x 140 mm (5.5 x 5.5 in.).
without confinement (DF11). Therefore, the test results are Thus, the load was applied by a total of 16 bearings simulating
not discussed further in this paper.
a uniform loading case. The reaction frame consisted of two
parallel steel beams, which were supported by four tie rods
Test setup and measurements anchored to the strong floor of the laboratory.
Series I (DF6 to DF10)—The specimens of Series I were
tested supported on sand (Fig. 3(a)). The dimension of the For both test series during testing, the vertical displacement at
sand box was 3250 x 3250 x 3200 mm (128 x 128 x 126 in.). the corners of the column stub and the slab corners were
The remaining depth of the sand bed underneath the footings measured using linear variable differential transformer
varied between 2200 and 2500 mm (87 and 98 in.), being (LVDT) gauges. The flexural steel strains were monitored at
approximately twice the width of the footings. To achieve 14 locations as shown in Fig. 4(b) as well as at some of the
homogeneous sand beds of reproducible packing, controlled stirrups. At one measuring point, two strain gauges were
pouring and tamping techniques were used to deposit sand attached to opposite side faces of the reinforcement bars to
layers of 55 mm (2.2 in.) thickness into the sand box. The obtain the strains at the center of gravity of the bars. The
load was applied by six hydraulic jacks (maximum capacity concrete strains were also recorded at 14 locations on the
3540 kN [796 kips]) placed between a steel frame and the compression faces.

ACI Structural Journal/September-October 2009 709

Fig. 5—Saw-cuts of different test specimens.

Testing procedure Load-deflection characteristic

The load was applied in increments of 108 to 196 kN The deflections at the center of the slab for several footings
(24.3 to 43.8 kips) for Series I and in increments of 100 to are shown in Fig. 6. For all series, the gradient of the curves
200 kN (22.5 to 45.0 kips) in Series II. For Series I, the decreases with increasing a/d; this can be attributed to the
increments were chosen to increase the soil pressure by increasing bending deformations. In contrast, the deformations
75 and 100 kN/m2 (10.9 and 14.5 psi), respectively. of the more compact footings are mainly induced by shear. For
After the service load Vservice was reached (Table 1), the the tested specimens, the shear deformations are somewhat
load was cycled 10 times between the service load and half smaller than the deformations due to bending irrespective of
of the service load. After this, the load was incrementally the boundary conditions, the concrete compressive strength,
increased until 80% of the calculated failure load was or the existence of shear reinforcement (Fig. 6). If the load
reached. Then the footings were continuously loaded until deflection curves of the specimens DF11 to DF13 and DF20
failure took place. to DF22 are compared, it is obvious that the gradient of the
deflection curves increases for higher concrete grades due to
EXPERIMENTAL RESULTS the increased stiffness (Fig. 6(b) and (d)). The specimen with
Cracking and failure characteristic shear reinforcement showed a more ductile behavior than the
All tests failed in punching of the footing. The failure specimen without shear reinforcement (Fig. 6(b) and (c)).
loads are listed in Table 1. The comparison with the flexural
capacities of the footings Vflex according to Gesund9 reveals Steel strains
that the flexural capacities were not reached and confirm the Measurements were made to determine the steel strain
fact that failure occurred due to punching (Table 1). After the distribution for all slabs. Typical test results of these
test, the specimens were sawn into two halves. In all slabs, measurements are shown in Fig. 7. In general, the measurements
the failure surface consisted of a wide shear crack, which showed some scatter. For the specimen without shear
formed the surface of a truncated cone (Fig. 5). For the specimens reinforcement, the bars did not reach yield at failure. For
without shear reinforcement, the observed inclinations of the the footings DF16 and DF17 with shear reinforcement, only
failure shear crack were approximately 45 degrees for the the tension reinforcement in the vicinity of the column
compact footings DF11, DF20, and DF6 (a/d = 1.25) and yielded before punching took place (Fig. 7(c)). Thus, the
less than 35 degrees for the more slender footings DF13 and flexural capacity had not been reached when the footings
DF22 (a/d = 2.0). For the footings without shear reinforcement, failed in punching.
the inclination of the failure crack seems to be mainly influenced
by a/d. Neither the concrete strength nor the support situation had Soil pressure distribution
a significant effect in this aspect. For the shear-reinforced For all tested footings of Series I, the soil pressure distribution
specimens, an outer shear crack with an inclination of was measured. Figure 8 shows the soil pressure distribution of
approximately 45 degrees, crossing the shear reinforcement, as the footings DF6 and DF9 close to failure. Although the
well as a much steeper inner crack with an inclination of measurements showed plenty of scatter, a concentration of
approximately 50 to 60 degrees were observed. Strain soil pressure underneath the column stub was more
measurements confirmed that the shear reinforcement was pronounced for DF9 (a/d = 2.0).
activated. Hence, it can be concluded that the outer, more
gently inclined shear crack developed first. The failure most DISCUSSION OF EXPERIMENTAL RESULTS
likely took place when the second, inner shear crack For the following discussion of the experimental results,
occurred. In contrast to the footings without shear reinforcement, the critical shear section specified by ACI 318-08 is used.
the influence of a/d on the inclination of the shear cracks seems The failure load Vtest is multiplied by the factor (1 – A0 /A),
to be negligible. where the parameters A0 and A are the area within the shear

710 ACI Structural Journal/September-October 2009

deflection (in.) distance from s lab center (in.)
0.0 0.5 1.0 1.5 5 10 15 20 25 30 35
4000 5.0
(a) without shear reinforcement DF8
supponed on sand 800 ~4.5 DF7

';" 4.0
£ 3000 ·;; sua in gauges I

c. ~ 3.5 51 57 58 59 I

~ 3.0
.: 2000
~ 2.5 -------------------
.."' ~ 2.0
-~ 1.5 ~ - - DF6
- - l)f'7
- - DF6 (ald• l.25)
200 "
-~ 1.0
" 0.5
"' _,._ DFS
- - - yield strength
- - OF7 (aid= 1.5)
0 0.0
0.0 1.0 2.0 3.0 4.0 5.0 100 200 300 400 500 600 700 800 900
distance from slab cente r (mm)
deflection (mm)
distance from slab center (in.)
deflection (in.) 5 10 15 20 25 30 35
0.0 0.5 1.0 1.5 5.0
4000 DFI2

(b} without shear rcinforcment ~4.5
800 -; 4.0
·; gauges~train
~ 3.S 51 57 SS S9
z 3000 f 3.0

c. 600 j
"'.2 c.
E '_,$
~ -------------------
-g 2000 Q
~ 2.0
400 ~
c. 'f I.S ~
~ .... :::::.
- - 0 1' 11
"'" g. "
-~ 1.0 ..... - - DFI2
1000 - - oF II (ald• l.25) c _,._ DFI3
- - o F12(atd=L5) ~ 05
(b) - - - yield strcng1h
DF I3 (ald• 2.0) 0.0
100 200 300 400 500 600 700 800 900
1.0 2.0 3.0 4.0 5.0 di.stanee from slab center (mm)

deflection (mm) distance from s lab center (in.)

5 10 15 20 25 30 35
deflection (in.) 5.0
0.0 0.5
~ ~----~--~~--~----,
1.0 1.5 l4.S
(C) with reinforcement .; 4.0
800 ~ 3.5

z 3000 ~ 3.0
600 j E
c. c ~ 2.5


~ 2000
"'.... ~ 2.0
·~ 1.5
- - DFI6
- - OF17
400 ~
-~ 1.0
_,._ OF IS
"'" - -OF16(ald= l.25) "'" " 0.5
- - - yield strength
1000 200 (c)
- - OF1 7(ald= l.5) 0.0
OF 18 (ald=2.0) 0 100 200 300 400 500 600 700 800 900
o ¥-~~~==~==~~ o distance from slab center (mm)

0.0 1.0 2.0 3.0 4.0 5.0 distance from slab center (in.)
deflection (mm) 5 10 15 20 25 30 35
deflection (i n.) 5.0
0.0 0.5 1.0 1.5 ~4.5 DF21
(d) withom shear reinforcement
';;" 4.0
~ 3.5 Sl

f 3.0
strain gauges
$7 S8

£ 3000
c. ~ 2.5 -------------------
~ 2.0
~ 2000
c. ry -~ 1.5
~- - - DF20
- - DF2 1
_,._ DF22
.."' ·! 1.0
" - - - yield strength
1000 f/ -- DF20 (ald=l.25) ~ 0.5
II --OF21 (ald=1.5)
~· DF22 (ald• 2.0) 0 100 200 300 400 500 600 700 800 900
.lL----.---=;::::=;::::::::;==~ 0 distance from slab center (mm)
o.o 1.0 2.0 3.0 4.0 5.0
Fig. 7—Typical steel-strain distributions at failure.
deflection (mm)

Effect of a/d
Fig. 6—Load-deflection curves.
In Tests DF11 to DF13, the only parameter varied was a/d.
Therefore, the tests may be used to study the effect of a/d on the
critical section and the area of the footing, respectively punching behavior. As shown in Fig. 9(a), the tests indicate
(Fig. 1(a)). The term (1 – A0 /A) accounts for the soil reaction that the shear strength decreases with increasing a/d. The
within the critical section to be subtracted from the punching ACI 318-08 does not reflect this trend as it does not account
force assuming a uniform soil pressure distribution. for the effect of a/d.

ACI Structural Journal/September-October 2009 711

Fig. 8—Soil pressure distribution close to failure.
Fig. 9—Effect of: (a) a/d; and (b) concrete compressive strength
Effect of concrete compressive strength on punching shear strength of footings.
For the investigation of the influence of the concrete
strength, two test series are available. The footings of the sponding similar footing without shear reinforcement.
first series (DF11 to DF13) had a concrete compressive Figure 10(a) shows that shear reinforcement substantially
strength of approximately 20 MPa (2.9 ksi) and the test increases the punching capacity of footings. The effectiveness
specimens of the second series (DF20 to DF22) had a target of the shear reinforcement, however, depends on a/d. The
strength of 40 MPa (5.8 ksi). In Fig. 9(b), the failure loads effectiveness of shear reinforcement decreases with
are plotted as a function of the concrete compressive strength decreasing a/d. For example, the shear strength of footing
fc′. Although the test data are limited, it can be concluded DF18 (a/d = 2.0) was increased approximately 80% by the
that for slender footings (a/d = 1.5 and 2.0), the concrete stirrups. In contrast, the shear reinforcement increased the
strength seems to significantly affect the punching behavior. shear strength of the more compact footing DF16 (a/d = 1.25)
For the less slender footings DF11 and DF20 (a/d = 1.25), by only approximately 33%. In Fig. 10(b), the ratio vs,test /vs,ACI
the effect of the concrete strength is not significant. This is is compared. The shear stress v s,test is calculated by
remarkable because it was expected that the behavior of the (V test – 0.5Vc,test)/(b0d). Thus, the theoretical capacity of the
more compact footings could be described by a strut-and-tie shear reinforcement according to ACI 318-08 has not been
model. In such a case, the failure load of the compact footings reached or the concrete contribution is overestimated.
DF11 and DF20 should be controlled by the bearing capacity of
the compression strut if enough flexural reinforcement is Effect of soil-structure interaction
provided. Therefore, for compact footings, it was expected Two test series are available to investigate the influence of
that the punching shear strength increases linearly with the the soil-structure interaction. Series DF6 to DF8 were
concrete compressive strength. Two test results, however, supported on sand, whereas the companion specimens of
are not sufficient to draw definite conclusions and more test Series DF11 to DF13 were uniformly loaded. In general, the
results are needed on this aspect. footings supported on sand had a higher punching shear
resistance than those uniformly loaded (Fig. 11). This may
Effect of shear reinforcement be attributed to the soil pressure concentration underneath
To investigate the influence of shear reinforcement on the the footing (Fig. 8). This concentration increases the part of
punching behavior, two series are available. In the first the soil pressure that reduces the applied shear stress along
series, three specimens with closed stirrups as shear the critical perimeter. The test Specimen DF8 supported on
reinforcement were tested (DF16 to DF18). The second sand showed significantly higher shear strength than the
series included footings DF11 to DF13 that are identical to uniformly loaded Specimen DF13 (Fig. 11). To fit into the
the first series except they did not contain shear reinforcement. sand box, the dimensions of Specimen DF8 had to be reduced
In Fig. 10(a), the ratio Vtest /Vc,test is plotted as a function of compared to Specimen DF13. To ensure the same a/d, the slab
a/d. Vtest is the failure load of the footing containing shear thickness also had to be reduced from 450 to 300 mm (17.7 to
reinforcement, while Vc,test is the failure load of the corre- 11.8 in.) for Specimens DF13 and DF8, respectively. Therefore,

712 ACI Structural Journal/September-October 2009

Fig. 10—(a) Contribution of shear reinforcement on shear
strength of footings; and (b) comparison between theoretical
contribution of shear reinforcement to punching shear
strength according to ACI 318-08 and steel contribution
determined by testing.

Fig. 11—Effect of soil-structure interaction on punching

shear strength of footings.

Fig. 12—Comparision between punching tests and punching

the high shear strength of the footing DF8 may be partly shear resistance according to ACI 318-08.
attributed to the size effect. The present tests are not
sufficient to evaluate the size effect quantitatively. It may be Comparison with ACI 318-08
concluded that for design purpose, the soil-structure interaction The ACI 318-08 code provisions are compared with the
is not a major factor.6 test results in Fig. 12. The ACI Code does not account for the
size effect on the punching shear strength. Therefore, the
COMPARISON OF PREDICTIONS AND provisions tend to be less conservative for larger effective
EXPERIMENTAL RESULTS depths. This tendency can be observed for the specimens
For the comparison between the present tests and the supported on sand as well as for the uniformly loaded footings
codes, all material and strength reduction factors in the code (Fig. 12(a)). The ratio Vtest /VACI has a mean of 1.33 and a
equations are taken as unity. The comparison is mainly based coefficient of variation of 0.13. The ACI predictions as a
on the ratio of the failure load Vtest and the calculated function of a/d are shown in Fig. 12(b). Although the code
punching shear resistance Vcode. The ratio Vtest /Vcode is does not account for a/d, the analysis of all tested footings
plotted against the effective depth d and a/d. seem to be trend free for the concrete strengths tested. The

ACI Structural Journal/September-October 2009 713

reinforcement is plotted as a function of a/d in Fig. 12(d).
Except for Specimen DF9, which was supported on sand, the
ratio Vtest /VACI is independent of a/d.

Comparison with Eurocode 2

Eurocode 2 provides two equations to determine the
punching shear strength of slabs without shear reinforcement
VRd,c (Eq. (15)) and VRd,max (Eq. (17)). The smaller value
controls the design. The equation for VRd,max was originally
adapted from the Model Code 90.10 For slabs, Eurocode 2 as
well as Model Code 90 uses a critical perimeter at a distance
of 2.0d from the periphery of the applied load. The use of a
control perimeter relatively far away from the column face
leads to a good correlation with slab tests with a practical
ratio of column perimeter u0 to effective slab depth d (larger
than 4). The large control perimeter also offers the opportunity
to use the same shear strength as for one-way shear. The large
perimeter, however, leads to the fact that the shear stress
along the periphery of very small loaded areas becomes
governing.11 Therefore, Model Code 90 and Eurocode 2
demand that the shear stress at the periphery of the applied load
should not exceed the web-crushing limit for beams with
vertical stirrups.
Unfortunately, VRu,max = γcVRd,max (with γc = 1.5 being
the material resistance factor for concrete and VRd,max
according to Eq. (17)) controls the design of the present tests
(except DF10) due to the small ratios u0/d, which are
characteristic for column footings. Because the equation
of VRu,max is only a function of the concrete compressive
strength, VRu,max does not reflect the influence of the effective
depth d or a/d correctly, as shown in Fig. 13(a) and (b). In
addition, the equation significantly overestimates the
punching shear capacity for higher concrete grades, as
shown in Fig. 13(b). This is attributed to the fact that the
equation accounts for a linear effect of the concrete compressive
strength on the punching capacity, which is in contrast to test
results.12,13 It is also obvious from the statistical evaluation,
which leads to a mean value of 1.12 and a coefficient of
variation of 0.23 indicating high scatter, that the equation
for VRu,max is not able to predict the punching shear strength
of footings without shear reinforcement satisfactorily. The
equation of VRu,max is also not applicable for the maximum
punching strength of footings with shear reinforcement
(Fig. 13(d)). Due to the small ratios u0/d of 2.0 and 3.2, the
equation calculates very conservative punching shear resistances
for the shear-reinforced footings and does not count for any
shear reinforcement.
In Fig. 14, the resistance is calculated according to Eq. (15)
Fig. 13—Comparision between punching tests and punching and (16), while VRu,max is neglected. The aim of this comparison
shear resistance according to Eurocode 2. is to investigate the performance of these equations. For the
footings without shear reinforcement, VRu,c = γcVRd,c (with
ratio Vtest /VACI is significantly higher for the specimens VRd,c according to Eq. (15)) correctly reflects the influence of the
supported on sand (Fig. 12(c)). This is due to the fact that for effective depth d, the concrete strength, and a/d, and results in less
the calculations, a uniform soil pressure distribution underneath scatter (Fig. 14(a), (b), and (c)). This is also confirmed by the
the footings was assumed. The soil pressure inside the critical reduced coefficient of variation of 0.10. The mean value of
perimeter b0 at a distance of 0.5d from the column face was only 0.77, however, clearly indicates that for footings without
subtracted from the applied shear force. In the tests, a shear reinforcement, Eq. (15) would result in an unconservative
concentration of the soil pressure inside the critical perimeter design. Application of Eq. (16) leads to safety factors below 0.5
was measured. This led to a reduction of the applied shear because the footings failed in maximum punching shear due to the
force and, therefore, increased the punching shear strength. high shear reinforcement ratio (Fig. 14(d)).
Furthermore, the diameter of the punching cone differs from Eurocode 2 provides parameters that are optional for
the diameter of the assumed critical perimeter to some extent national choice. In Eq. (15), the empirical factor CRd,c can be
(Fig. 5). The ratio Vtest /VACI for the specimens with shear given by the National annex. To ensure the required safety

714 ACI Structural Journal/September-October 2009

Fig. 14—Comparision between punching tests and punching Fig. 15—Comparision between punching tests and punching
shear resistance of footings according to Eurocode 2 neglecting shear resistance according to the modification proposed to
the influence of Vmax. Eurocode 2.

level for footings, it is proposed here to reduce CRd,c to a with vRd,c being the punching shear capacity of a slab without
value of 0.12/γc. shear reinforcement, which is calculated as vRd,c = 0.18/γck
To overcome the problem that Eq. (17) for VRd,max calculates (100ρfck) 1/3. The notations are chosen in accordance with
very conservative punching shear resistances for the shear- Eurocode 2. Certainly, the applied shear force can only be
reinforced footings, a new equation for the calculation of reduced by the effective soil pressure within the column
VRd,max at the periphery of the loaded area is herein proposed perimeter u0. In Fig. 15, the proposed equations are
compared to the test results. The proposal correctly reflects
the influence of the effective depth d and a/d. In addition, the
V Rd, max = 16 d ⁄ u 0 v Rd, c u 0 d (19) scatter is reduced and the safety level is increased.

ACI Structural Journal/September-October 2009 715

Based on the results of the experimental investigation on The authors wish to express their sincere gratitude to the Deutsche
footings supported on sand, as well as on footings loaded Forschungsgemeinschaft (DFG, German Research Foundation) for the
financing. The tests of Series I were conducted in cooperation with the Institute
uniformly, the following conclusions can be drawn: of Geotechnical Engineering, RWTH Aachen University. The effective
1. For the footings without shear reinforcement, the cooperation is appreciated. This paper was written during a research visit
inclination of the failure shear crack seems to be mainly of A. G. Sherif at the RWTH Aachen University financed by the Alexander
influenced by a/d and not by the concrete strength. The von Humboldt Foundation. The support of the Alexander von Humboldt
observed inclinations of the failure crack were approximately Foundation is deeply appreciated.
45 degrees for the compact footings (a/d = 1.25) and less than
35 degrees for the more slender footings (a/d = 2.0). REFERENCES
2. The punching shear resistance is strongly influenced by 1. Regan, P. E., and Braestrup, M. W., “Punching Shear in Reinforced
Concrete,” CEB-Bulletin d’Information No. 168, Lausanne, Switzerland,
a/d. The shear strength decreases with increasing a/d. 1985, 232 pp.
3. Although the test data are limited, it can be concluded 2. Polak, M. A., ed., Punching Shear in Reinforced Concrete Slabs, SP-232,
that for slender footings (a/d = 1.5 and 2.0), the concrete American Concrete Institute, Farmington Hills, MI, 2005, 302 pp.
strength seems to significantly affect the punching behavior. 3. Richart, F. E., “Reinforced Concrete Wall and Column Footings,”
This effect seems to decline for more compact footings ACI JOURNAL, Proceedings V. 45, (Part 1) No. 2, Oct. 1948, pp. 97-127, (Part
(a/d = 1.25). 2) No. 3, Nov. 1948, pp. 237-260.
4. Dieterle, H., and Rostásy, F., “Tragverhalten quadratischer
4. Shear reinforcement can substantially increase the Einzelfundamente aus Stahlbeton,” Deutscher Ausschuss für Stahlbeton,
punching capacity of footings, but is less effective with V. 387, Berlin, Germany, 1987, 134 pp.
decreasing a/d. 5. Hallgren, M.; Kinnunen, S.; and Nylander, B., “Punching Shear Tests
5. In general, the footings supported on sand showed a on Column Footings,” Nordic Concrete Research, V. 21, No. 3, 1998, pp. 1-22.
higher punching shear resistance than those uniformly 6. Hegger, J.; Sherif, A. G.; and Ricker, M., “Experimental Investigations
loaded. This may be attributed to the soil pressure concentration on Punching Behavior of Reinforced Concrete Footings,” ACI Structural
Journal, V. 103, No. 4, July-Aug. 2006, pp. 604-613.
underneath the footing. The assumption of uniformly distributed
7. ACI Committee 318, “Building Code Requirements for Structural
soil pressure beneath the footings according to the building Concrete (ACI 318-08) and Commentary,” American Concrete Institute,
codes, however, ensures a safe design. Farmington Hills, MI, 2008, 465 pp.
6. ACI 318-08 does not account for the size effect on the 8. European Committee for Standardization (CEN), “Eurocode 2: Design
punching shear strength. Therefore, the provisions tend to be of Concrete Structures—Part 1.1: General Rules and Rules for Buildings,”
less conservative for larger effective depths. This tendency Brussels, Belgium, 2004, 225 pp.
9. Gesund, H., “Flexural Limit Analysis of Concentrically Loaded
can be observed for the specimens supported on sand as well Column Footings,” ACI JOURNAL, Proceedings V. 80, No. 3, May-June 1983,
as for the uniformly loaded footings. pp. 223-228.
7. According to Eurocode 2, the shear strength of footings 10. Committee Euro-International du Béton, “CEB-FIP Model Code
of practical dimensions, as the one tested, is governed by the 1990: Design Code,” London: Thomas Telford, 1993, 437 pp.
maximum allowed web crushing limit VRd,max at the column 11. Regan, P. E., “Punching of Slabs under Highly Concentrated Loads,”
face. Because VRd,max is only a function of the concrete Structures and Buildings, V. 157, No. 2, 2004, pp. 165-171.
compressive strength, this leads to the fact that it is not 12. Elstner, R. C., and Hognestad, E., “Shearing Strength of Reinforced
Concrete Slabs,” ACI JOURNAL, Proceedings V. 53, No. 7, July 1956, pp. 29-58.
possible to increase the punching resistance of footings by 13. Moe, J., “Shearing Strength of Reinforced Concrete Slabs and Footings
using shear reinforcement. Modifications are proposed to under Concentrated Loads,” Bulletin D47, Portland Cement Association,
overcome this deficiency. Skokie, IL, 1961, 135 pp.

716 ACI Structural Journal/September-October 2009

Disc. 106-S55/From the September-October 2009 ACI Structural Journal, p. 600

Progressive Collapse Resistance to Axially-Restrained Frame Beams. Paper by Youpo Su, Ying Tian, and
Xiaosheng Song

Discussion by Weijian Yi, Qingfeng He, and Xiao Yang

College of Civil Engineering, Hunan University, Changsha, Hunan, China

The paper presents the experimental results of a series of measured strain of the steel of Specimen B3 in the Closure
12 reinforced concrete (RC) beams restrained longitudinally of this discussion, which would be helpful in better under-
against axial deformation. Some theoretical considerations standing the paper.
are also proposed on the basis of a theory developed by Park 3. In the literature to date, there are very few measurement
and Gamble.13 The innovative design of the experiments and results of arch thrust in the testing of RC beams and slabs. It
testing results presented by the authors allowed the discussers to is very interesting that the maximum arch thrust and the
investigate the compressive arch action and tensile catenary maximum load-carrying capacity of the beams occurred at a
action in reinforced concrete beams. The discussers would different deformation state, as described in the paper. The
like to offer the following comments and suggestions: curves shown in Fig. 12 and Eq. (3) seem to give a reasonable
1. As seen in Fig. 2 to 4, there should be a great rigid explanation of the relationship between the increase in the
assembly to block horizontal displacement of the specimens cross section bending resistance capacity and the change in
under compressive arch thrust. Also, the horizontal force load-carrying capacity of the beam. In the compressive arch
should be measured. It is noted that from Fig. 5 to 9 that action stage, the effect of axial force on the bending resistance
several axial force curves take on slippage in the position of capacity of the sections and the flexural equilibrium state of
zero axial force. This slippage may have resulted from an the beam changed with deflection. For the slabs with a small
experimental device other than the behavior of the beams. In section height, the stage from the beginning of the arch
addition, the linear variable differential transformer (LVDT) action to the snap-through is very short. It can be loosely
installed in the actuator can only be used to measure the considered that the sectional bending resistance capacity and
cylinder displacement of the actuator rather than the the load-carrying capacity of the slab reach the maximum at
displacement of the beam. From Fig. 4 it can be seen that the the same time, but the section height of a beam is usually
steel frame on which the actuator was fixed may also deform much greater than that of a slab. Before the cross section
upward under loading. The displacement of the actuator’s bending resistance capacity reaches its maximum, the beam
cylinder is not equal to the displacement of the specimen. It reaches the maximum load-carrying capacity. This is the
is necessary to calibrate the flexibility of the frame, then main difference in compressive arch action between beams
subtract the displacement of the frame from the total actu- and slabs. It is also the main finding of the paper. Unfortu-
ator’s displacement to obtain the displacement of the spec- nately, the paper gives the measured relative bending
imen. In addition, the sudden drop of the load-carrying moment of only one specimen. The variation of arch thrust is
capacity in Specimens A3 and A6, as shown in Fig. 5 and 8, is limited to qualitative discussions. Using Eq. (3) in the paper,
perhaps due to the sudden release of elastic strain energy Park and Gamble13 developed an arch thrust equation of a
stored in the frame in the conversion process of arch action slab strip, as shown in Eq. (1) of the paper. Park’s equation,
to catenary action. If the stiffness of the test device was large however, implies an assumption that the “maximum load-
enough, more smooth curves13 could be obtained. carrying capacity and the maximum arch thrust occurred at
2. In Fig. 5 to 8 and Fig. 12, the load-deflection curves are the same time.” Park’s model is not applicable to the beams
marked with a yielding point, which means that the yielding presented in the paper. As shown in Fig. 12, the axial force,
of tension steel reinforcement occurred at the supports; bending moments at the midspan, and support can be
however, how to measure the strain of the steel was not expressed as the functions of deflection δ. The correct
described in the paper. It is not clear how to get the yielding method is to solve the deflection δ from the following equation
point in the testing. From the curves in these figures, such as
Specimens A1, A4, B3, B4, and others, it cannot be considered d ( M + M′ ) dN
that the behavior of the specimens at the yielding point have -------------------------- = ------- δ + N
dδ dδ
not been beyond the “linear load-deflection response,”
assuming the yielding bending moment amplifies the factor
of Specimen B3 at the midspan and support as 1.15 and 1.65, By substituting δ into Eq. (3) in the paper, the maximum
respectively (Fig. 12). The estimated yield load according to load-carrying capacity of the beam under the action of arch
Eq. (3) is at least 10% less than that listed in Table 3. In thrust can be obtained.
testing the applied load, the axial force and bending moment 4. The paper states that the tension of steel reinforcement
at the supports could be measured. Then, the bending at the midspan in all of the specimens was finally fractured.
moment at the midspan could be calculated based on Eq. (3). For example, there were three steel bars of 14 mm (0.552 in.)
The yielding of steel reinforcement, however, can only be diameter in the bottom of Specimen B3. The total tension
determined from the measured strain of the steel. The force was 247 kN (55.53 kip). Due to the elongation of the
measured strain of the steel was not listed in the paper, steel reinforcement rate of 27%, top reinforcement should
possibly due to limits on length. The authors could plot the also be in tension. Therefore, the axial force in the beam may

486 ACI Structural Journal/July-August 2010

not be less than 247 kN (55.53 kip). In Fig. 5, however, the
axial force in Specimen A3 is only approximately 100 kN
(22.48 kip). The axial tension forces in other specimens are
also smaller than the total tension force of the bottom steel
bars. In the second paragraph on p. 604, it says that “The
final failure of all specimens was announced by the fracture
of bottom reinforcement at the interface of beam and center
column stub.” According to the explanation for Fig. 12 in the
paper, the bending moment in the midspan existed until the
bottom steel reinforcement fractured. This is possible
because the bottom and top steel bars are in different tension
stress states, which results in a bending-tension state. If there
is a considerable bending moment at the midspan of a beam
in the final stage, the load-carrying capacity of the beam is
obviously underestimated by the model presented in Fig. 14
because there is not only catenary action but also bending Fig. 15—Vertical load and horizontal reaction force versus
moment in the load path. As described in the paper, “It is normalized center deflection (Specimens B3 and B2).
noted that, prior to failure, the specimen could still resist a
significant amount of bending moment at the critical vertical load P was measured solely by the load cell of the
sections. Therefore, a double curvature deformed shape was actuator. However, as indicated by the use of “transducers”
maintained in the beams until failure when the bottom rein- in this sentence, the center vertical displacement was
forcement at the midspan fractured under catenary action” measured by more than one transducer. Two LVDTs were
(first paragraph, p. 606). This description may be used to actually used in the tests: one was mounted at the actuator
explain why the measured axial force is less than the total and another one was independent. The difference in the
tension force in the bottom steel reinforcement of the beam. measurements from these two transducers was negligible at
If this explanation was acceptable, the ultimate load-carrying the peak load for each specimen, indicating that the loading
capacity would be calculated using Eq. (3) in the paper by frame placed vertically was sufficiently stiff.
simply substituting the measured axial tensile force and A sudden drop of loads occurred mainly in testing Specimens
bending moments in the equation. In this way, however, the A2, A3, A5, and A6 with small span-depth ratios. Such a
comparison shown in Fig. 14 may be meaningless. In phenomenon can be better explained by the notable effects
addition, “tensile arch action” in Table 3 should be “tensile of high axial forces developed in the beams on their flexural
catenary action.” strength rather than the strain energy stored in the vertical
5. A comparison of the test curves of Specimens B2 and loading frame. A numerical model with properly defined
B3 is very interesting because Specimen B2 contains only parameters can successfully capture the sudden loss of
one more bottom steel bar of 14 mm (0.552 in.) diameter than loading capacity as a result of concrete crushing. The horizontal
Specimen B3. According to the curves illustrated in Fig. 7 and rigidity of the supports that anchored the beams was in a
8, the curves of Specimens B2 and B3 are drawn together, as realistic range that the neighboring structural components
shown in Fig. 15. It is indicated in the figure that Specimens such as columns and slabs can offer to a frame beam. A test
B2 and B3 have an approximate yield load and ultimate load setup with extremely high axial rigidity is neither practical in
(a difference of approximately 10%), and the same maximum a test nor necessary for simulating the actual boundary
arch thrust of 210 kN (47.21 kip). The significant difference condition of a frame beam. Slippage in the measured horizontal
is the conversion from compressive arch action to tensile reaction force took place when the compressive arch action
catenary action. When the relative deflection reached 0.7, was transformed into catenary action. Such a phenomenon
the compressive arch action disappeared in Specimen B2. At can be explained by the inherent tolerance for the connection
this time, the top steel reinforcement at midspan had not yet components of the supports, especially at the pins
reached the position below the bottom steel reinforcement at connecting the steel sockets with the test bed. The slippage
the support. The “arch” should be maintained at this stage. In was sufficiently small, thereby causing negligible effects on
Specimen B3, however, the compressive arch action existed the overall performance of the specimens.
until the relative deflection reached beyond 1.10. At this Table 3 provides the measured horizontal reaction force
stage, the top face at the midspan of the beam had dropped and vertical deflection of specimens at the reach of peak load
below the bottom face at the support. In the two specimens, Pcu under compressive arch action. The discussers may have
the transition from “arch” to “catenary” could not be misinterpreted Pcu as the load causing tensile steel yielding.
explained by existing models. Because Pcu was much higher than the yield load, it is not
surprising that the “estimated yield load according to Eq. (3)
AUTHORS’ CLOSURE is at least 10% less than that listed in Table 3.”
The authors would like to thank the discussers for their The discussers claim, without providing experimental
interest in the paper. The paper focused primarily on the evidence, that “For the slabs with small section height, the
effects of compressive arch action on the load-carrying stage from the beginning of arch action to snap-through is
capacity of axially restrained frame beams. very short. It can be loosely considered that the sectional
It appears that the discussers have misunderstood the bending-resistance capacity and the load-carrying capacity
context regarding the measurement of beam deflections. The of the slab reach the maximum at the same time.” The major
paper states that “the vertical load P and deflection δ at the difference between the beam and one-way slab is the span-
center column stub were measured by a built-in load cell depth ratio that is one of the parameters governing the effects
within the actuator and displacement transducers.” The of compressive arch action. It is noteworthy, however, that

ACI Structural Journal/July-August 2010 487

the same mechanism of compressive arch action can still be approach has to be used to determine the value of δ at Pcu.
assumed for both beams and one-way slabs. Additionally, Additionally, if the same ways of defining M, M′, and N (as
due to the secondary effects resulting from large slab defor- those recommended by Park and Gamble13) are used, it is
mation far beyond initial steel yielding, the slab loading believed that the approach recommended by the discussers
capacity can be reached earlier than its flexural capacity. would lead to results identical to the analytical predictions
Such a performance has also been observed in the beam tests given in the paper.
(such as Specimen B3). The authors extended the model It is noted that N is defined in the paper as the measured
developed for one-way slabs by Park and Gamble13 to horizontal reaction force. Hence, Fig. 5 to 8 show the horizontal
axially restrained beams. It is noted that, different from what reaction force rather than the axial force actually developed
the discussers have interpreted, Eq. (1) defines P as the in the beams. At large deformation of beams subjected to
vertical load applied on the beam rather than as the “arch catenary action, the axial force will be much higher than the
thrust.” In addition, scrutinizing the context of Chapter 12 of horizontal reaction force. Therefore, it is inappropriate to
Reference 13 indicates that no assumption was adopted or directly correlate the forces shown in these figures to the
implied by the authors of this reference about a simultaneous catenary action forces. Figure 14 shows a simple model for
reach of the maximum vertical load-carrying capacity and predicting the loading capacity as well as the comparison
the maximum axial force developed in the beams under between the calculated and measured results. As described in
compressive membrane action. the paper and admitted by the discussers, the internal force
The discussers briefly described an approach to calculate may be far more complicated than that assumed in the simple
the beam loading capacity Pcu. This approach is simply a model. By presenting Fig. 14, the authors used a modest way
different mathematical method to solve the same problem: to show their disagreement with this simple model, which has
searching the maximum load at varying values. The equation been adopted in Reference 10, and alert readers that this model
provided by the discussers can be derived from Eq. (3) by may result in unreliable predictions.
taking the first-order directive of P with respect to δ as zero. The authors agree that the test results for Specimens B2
Despite the seemingly simple format of this equation, no and B3 are interesting, especially the earlier transition from
detailed formulations are provided by the discussers to a compressive arch action to a catenary action in Specimen B2.
define M, M′, and N as a function of δ. It is possible that a The authors are currently carrying out an analytical study
closed-form solution of δ cannot be obtained and a numerical that may facilitate a reasonable explanation of the different

Disc. 106-S56/From the September-October 2009 ACI Structural Journal, p. 608

Carbon Fiber-Reinforced Polymer for Continuity in Existing Reinforced Concrete Buildings Vulnerable to
Collapse. Paper by Sarah Orton, James O. Jirsa, and Oguzhan Bayrak

Discussion by Rafael Alves de Souza

ACI member, PhD, Universidade Estadual de Maringá, Maringá, Brazil

The authors have made a significant contribution regarding the main advantages of using CFRP instead of shoring elements
the strengthening of structures to prevent progressive collapse. in critical situations where progressive collapse is evident?
Also, the authors have presented an interesting strategy based
on the use of carbon fiber-reinforced polymers (CFRPs). EXPERIMENTAL PROCEDURE
Despite the quality of their research and their valuable findings, The authors have determined a strengthening scheme
the discusser requests clarification on some topics. using FRP that would allow a beam with discontinuity
reinforcement to survive loss of a column. For that, they
INTRODUCTION AND RESEARCH SIGNIFICANCE have investigated seven half-scale specimens based on
The authors have proposed the use of FRP as a strengthening typical information obtained from constructions built in the
alternative for beams that might lose some of their supports 1970s. Unfortunately, the concrete compressive strength
used for the specimens do not represent the majority of the
(interior columns). It is undoubtedly a situation that may
buildings constructed in that time. Also, the values presented
occur in daily practice and engineers need to come up with
in Table 3 are very different and may prompt distortions
quick and rational strategies of strengthening. The engineering
regarding the interpretation of the results.
solutions are desired to be simple, fast, and economical to
The discusser does not agree with the procedure of taking
avoid progressive collapse and allow the users to escape or
just one specimen for each proposed situation (NR-2, PM-1,
recover their belongings before the effects become significant. PM-2, NM-1, NM-2, FR-1, CR-1). At least two specimens
The use of CFRP for critical situations such as that should have been tested for each situation to effectively
presented by the authors, however, does not seem to be a discuss the results based on a minimum statistical background.
practical alternative, as this solution may demand specialized Also, there is a great variation for the compressive concrete
workers (adequate intervention) and structural engineers strength used in all specimens. As one can see in Table 3,
(design of the CFRP sheets). From a practical point of Specimens NM-2, FR-1, and CR-1 have concrete compressive
view—despite the exceptional qualities of CFRP sheets—a strengths that are significantly higher than the other situations.
shoring approach using steel or wood shores will work better Additionally, some specimens are significantly more reinforced
for a provisory situation where the loss of some columns is than others and, taking into account the great variation
evident. Also, this kind of approach is cheaper, faster, and does regarding the concrete compressive strength and reinforcements
not require a complex background. Could the authors explain (steel and FRP), some results were expected.

488 ACI Structural Journal/July-August 2010

EXPERIMENTAL RESULTS AND DISCUSSION The application of CFRP in these situations does require
To the discusser’s understanding, the authors have specialized workers and structural engineers, but there is not
mentioned two basic situations of failure for their specimens: a lack of these qualified people. The use of CFRP to
flexure strength and catenary (or cable) action. However, it strengthen structures has been long implemented and there
is not clear throughout the paper which situation is more are several firms that provide both the engineering assistance
effective for representing the level of strength of a structure to design the CFRP and the workers to correctly apply it. CFRP
based on the GSA guidelines.1 Also, it is not clear when is both flexible and lightweight so that after the surface prepara-
catenary (or cable) action may develop. How can the rotation tion, the application of the fabric can occur in less than a few
of 0.13 radians be defined to obtain catenary (or cable) action? hours with only a few skilled workers. For the beams in this
In the situation denominated “flexural strengthening,” study, the CFRP application only took approximately 1 hour.
the authors used 4.5 times the amount of CFRP used in For this study, the intent was to see whether CFRP could
Specimen NM-2. Why not use the same strengthening used be applied in such a way to provide continuity (which it can)
in Specimens PM-1/PM-2 and NM-1/NM-2 to account for and whether that continuity can aid in the resistance of
the effect of providing FRP for positive and negative progressive collapse (which it can). Although it would have
moments? Increasing CFRP in Specimen FR-1 makes a been desirable to repeat the tests of some CFRP designs, that
comparison with Specimens PM-1, PM-2, NM-1 and NM-2 was simply not within the time or budget constraints of the
difficult. Also, in Table 3, there is no description regarding projects. The evaluation of more variables was deemed more
the concrete compressive strength of Specimen NM-1. important than replicating rather expensive test specimens.
Specimens NM-1 and NM-2 did replicate results of the
negative moment strengthening. The two specimens were
the same, with the only difference being in the amount of
The authors have presented a very interesting paper CFRP applied. After the successful test of Specimen NM-1,
concerning the progressive collapse of reinforced concrete it was decided that a reduced amount of CFRP could produce
structures. The loss of a supporting column may leave a the same results, so Specimen NM-2 was tested.
beam unable to resist gravity loads and may lead to the The design concrete strength for the specimens was 27.6 MPa
collapse of either side of the lost column. In that way, engineers (4000 psi). The concrete, provided by a local concrete
need to come up with techniques that allow building occupants supplier, was unusually low for the first batch of beams, and
to recover their belongings or escape before the effects of a high for the second two. The concrete compressive strength
local failure become significant. The authors have presented of Specimen NM-1 was 33.8 MPa (4900 psi) (the same as for
a technique based on the use of CFRP sheets, which can be Specimens PM-1 and PM-2). Although the concrete strength
considered a very effective intervention because it can may not have been as intended, it did not significantly affect
provide a great level of ductility for damaged sections. the behavior of the specimens. For a specimen under catenary
In critical situations, however, this alternative of intervention action, the most important variables are the location and
may be considered more complicated and expensive than strength of the reinforcing steel, and height and depth of the
other provisory and simple situations, such as a shoring beam. As for the steel in the specimens, all specimens have
approach (steel or wood). Also, when strengthening a beam the same reinforcing steel and the same steel design, except
for flexure using CFRP, one must be aware of the effective Specimen CR-1, which was designed using current ACI 318
shear strength. If the shear strength needs to be enhanced requirements for integrity reinforcement.
once the flexural strength was increased by using CFRP All specimens (except for Specimen FR-1) exhibited some
sheets, this solution may become even more expensive and form of flexural failure (ex-beam hinging at the support in
complicated, demanding special attention. Specimens NM-1 and NM-2), then went into catenary action.
The authors are correct when they state that a statically Catenary action consistently developed at a deflection equal
applied load corresponding to 2(DL + 0.25LL), based on GSA to, or just greater than, the depth of the beam. To reach this
guidelines,1 may or may not correspond to actual progressive deflection without a complete flexural failure (bar fracture),
collapse prevention. In fact, new information concerning the beams needed to have sufficient rotational ductility. For
catenary (or cable) action is needed to better understand the these specimens, the rotational ductility needed to be
maximum strength of reinforced concrete structures. approximately 0.13 radians.
Finally, the discussed paper presents the importance of For Specimen FR-1, the intent was to increase the flexural
providing continuity for positive and negative reinforcements. strength of the beam to allow it to carry the moments induced
It is an issue that requires special attention in the design by a loss column. This was not the same intent as the other
codes. Practical experience has been shown through the specimens, so a different design approach was used. The
years that an adequately detailed structure may withstand design of the CFRP was based on basic flexural principles
incredible loads, even if some errors were committed during and used the least amount of CFRP needed to reach the
the design process. There is no doubt that the proposed required strength. As seen in Fig. 12 of Reference 19, the
intervention using FRP sheets is effective; however, it will strains measured in the CFRP were approaching the fracture
only be possible in practice if an adequate continuity of rein- strain and most of the CFRP had debonded, indicating that
forcement was defined in the damaged structure. the CFRP was almost fully used.
The discusser is correct that shear strength must be considered.
AUTHORS’ CLOSURE Improving the continuity in a beam does not help if the shear
The authors thank the discusser for his interest in the paper. It strength is lacking. Strengthening for shear was outside of
appears the discusser misinterpreted the intent of the strength- the scope of the research project.
ening procedure. The objective was to prevent a progressive
collapse if a catastrophic event occurred. The technique was REFERENCES
19. Orton, S. L.; Jirsa, J. O.; and Bayrak, O., “Design Considerations of
never intended to be used on a structure that was heavily Carbon Fiber Anchors,” Journal for Composites for Construction, ASCE,
damaged and needed to be shored to prevent collapse. V. 12, No. 6, Nov/Dec 2008, pp. 608-616.

ACI Structural Journal/July-August 2010 489

Disc. 106-S61/From the September-October 2009 ACI Structural Journal, p. 656

New Formula to Calculate Minimum Flexure Reinforcement for Thick High-Strength Concrete Plates. Paper
by E. Rizk and H. Marzouk

Discussion by Abdulkadir Cevik

Assistant Professor, Department of Civil Engineering, University of Gaziantep, Gaziantep, Turkey

The authors have proposed a new equation to calculate

minimum flexure reinforcement for thick plates and two-
way slabs. The validity of the new proposed equation is verified
by a comparison between the proposed equation with
Battista’s experimental results and with different code
formulas for calculating minimum reinforcement for flexural
members. At first sight, everything seems to be correct;
however, after closer inspection, the database (Battista’s
experimental results) used for verification cannot be used for
this purpose. First of all, the range of the dependent variable
(steel ratio) is too close where steel ratios are 0.22, 0.23, and
0.24, respectively. On the other hand, the ranges of independent
variables are quite high as compared to the dependent variable,
which means that most of the dependent variables do not Fig. 10—Interaction plot of steel ratio = 0.23 for fc versus
have any significant effect on the dependent variable. As the fy, fc versus depth, and depth versus fy.
dependent variable is grouped and the interaction diagrams
are plotted, this problem can be observed easily, as shown in
Fig. 10 to 12. As can be seen, variations of related independent
variables have no effect on the dependent variable (steel ratio).
As a result of the aforementioned arguments, Battista’s
experimental results cannot be used to verify any code formu-
lations or equations regarding the calculation of minimum
flexure reinforcement for thick plates and two-way slabs.

The authors would like to thank the discusser for his
interest in the paper, and for providing the authors the
opportunity to illustrate a few details. The developed model
is based on the theoretical assumptions based on the theory
of plates in Eq. (12) to (17) and the shear sandwich model
simplification, not any test data as explained in the paper. Fig. 11—Interaction plot of steel ratio = 0.22 for fc versus fy.
The test data of Battista is not related to the paper.
The discusser is arguing that Battista’s experimental work
cannot be used to verify any code formulations or equations
regarding the calculation of minimum flexure reinforcement
for thick plates and two-way slabs, because most of the
independent variables (concrete compressive strength fc′ , steel
yield strength fy, and slab depth d) do not have any significant
effect on the dependant variable (reinforcement ratio ρ).
The discusser’s argument is not correct because he ignored
the size-scale effect factor as an independent factor on the
amount of reinforcement ratio. It is possible to consider the
structural member size effect on the minimum reinforcement
ratio through the brittleness number concept NP, as defined
by Bosco et al.15 The brittleness number is derived from
linear elastic fracture mechanics (LEFM) concepts, as
Fig. 12—Interaction plot of steel ratio = 0.24 for fc versus
0.5 fy.
fy h
N P = ρ -----------
- (39)
K IC The critical value of the stress-intensity factor KIC can be
evaluated as follows
where ρ is the steel reinforcement ratio, KIC is the concrete
fracture toughness, fy is the yield strength of the steel, and h (40)
is the thickness of the structural member. K IC = Gf Ec

490 ACI Structural Journal/July-August 2010

Fig. 13—Brittleness number (NP): steel yield strength for Fig. 15—Brittleness number (NP): steel yield strength for
steel reinforcement ratio = 0.23%. (Note: 1 MPa = 145 psi.) steel reinforcement ratio = 0.24%. (Note: 1 MPa = 145 psi.)

Fig. 14—Brittleness number (NP): slab depth for steel Fig. 16—Brittleness number (NP): steel yield strength for
reinforcement ratio = 0.23%. (Note: 1 MPa = 145 psi.) steel reinforcement ratio = 0.22%. (Note: 1 MPa = 145 psi.)

where Gf is the fracture energy and Ec is the concrete ratio independent of member size. This is not true as the
modulus of elasticity determined by standard methods. The minimum reinforcement ratio is inversely proportional to the
brittleness of the structural member increases by increasing member depth.
the member size or decreasing the steel reinforcement ratio. As a result of the aforementioned arguments, Battista’s
Bosco et al.15 found that a particular value of number NP experimental results can be used to verify any code formulations
does exist, for which the moment at which the reinforcement or equations regarding the calculation of minimum flexure
yields equals the moment at first cracking. Such a condition reinforcement for thick plates and two-way slabs. The signifi-
defines the minimum amount of reinforcement ratio. Current cance of the independent variables on the reinforcement ratio
design codes suggest a constant minimum reinforcement is clear and can be observed easily, as shown in Fig. 13 to 16.

Disc. 106-S63/From the September-October 2009 ACI Structural Journal, p. 678

Evaluation of Load Transfer and Strut Strength of Deep Beams with Short Longitudinal Bar Anchorages.
Paper by Sergio F. Breña and Nathan C. Roy

Discussion by Dipak Kumar Sahoo, Bhupinder Singh, and Pradeep Bhargava

PhD, Reader, School of Engineering, Cochin University of Science and Technology, Kochi, Kerala, India; PhD, ACI member, Assistant Professor, Department of Civil Engineering,
Indian Institute of Technology Roorkee, Roorkee, India; PhD, Professor, Department of Civil Engineering, Indian Institute of Technology Roorkee

The authors are to be complimented for the interesting of web reinforcement, however, does serve to confine the
study. Based on a/d ratio, the authors have sought to identify inclined strut in the tied arch. Further, ACI 318-08 specifies
two mechanisms of load transfer in deep beams: tied-arch that diagonal struts in beams inclined in the range of 25 to
mechanism and truss mechanism. It is well established, 65 degrees with the adjoining tie are well-conditioned for
however, that besides the ratio, the truss mechanism is also strut-and-tie modeling. Therefore, for all cases wherein the
dependent on the amount of web reinforcement in the deep inclination of the strut is typically more than 25 degrees (that is,
beam, whereas the tied-arch mechanism is relatively inde- a/d < 2.14), it will be reasonable to use the tied-arch mechanism
pendent of the amount of web reinforcement. The presence as the overarching method for the analysis of deep beams.

ACI Structural Journal/July-August 2010 491

The authors have rightly observed that although vertical web
reinforcement is explicitly included in truss models, it is not
done so in tied-arch models. Moreover, the effect of horizontal
web reinforcement is usually not included in either of the
models. The discussers feel that discounting the role of
horizontal web reinforcement runs contrary to a unified
approach to strut-and-tie modeling. On the basis of their
investigations of statically determinate truss models with
vertical and with horizontal truss mechanisms, Matamoros
and Wong (2003) have concluded that though both the
mechanisms yield conservative results, they require almost
double the amount of web reinforcement compared to an
indeterminate truss model consisting of a combination of the
vertical and horizontal truss models. The discussers are of
the opinion that for a/d less than about 2.14 (strut inclination
>25 degrees), it will be simple and convenient to adopt the
tied-arch mechanism with web reinforcement—both vertical
and horizontal—being accounted for in determining the
strength of the single bottle-shaped diagonal strut joining the
load point and the support. Fig. 12—Variation of strut efficiency factor with strut angle.
The term F in Eq. (12) and (18) should be corrected to FS-truss.
The discussers suggest that the issue of apportioning the total
∑ --------sin
shear between the tied-arch and the truss mechanism can ρT = si 2
αi (23)
probably be better resolved by using the combined indeter- bs si
minate strut-and-tie model of Fig. 7(a) rather than the
determinate model approach implied in Eq. (10) through where Asi is the cross-sectional area of each layer of web
(21). Furthermore, the authors’ attempt to determine the reinforcement in the i-th orientation; bs is the strut or beam
fraction of the shear transferred through truss action on the thickness (out-of-plane); si is the spacing of web reinforce-
basis of strain measurements using Instruments L2 and L3 ment in the i-th orientation; and αi is the angle between the
may not be reliable because the strain profile across the strut strut and the bars in the i-th orientation.
axis is nonuniform due to the bottling effect. What is notable in Fig. 12 is that that the trend of the
It is interesting to note that the strut efficiency factor βs of authors’ experimentally obtained strut efficiency factors is
the diagonal struts of the nine beams that failed by strut similar to the trend of the predicted strut efficiency factors
failure when plotted against the corresponding strut inclination based on the Sahoo (2009) model.
angle α is seen to be increasing with increasing α following
the linear trend shown in Fig. 12. On the other hand, REFERENCES
irrespective of the angle of inclination of the diagonal strut Matamoros, A. B., and Wong, K. H., 2003, “Design of Simply Supported
Deep Beams Using Strut-and-Tie Models,” ACI Structural Journal, V. 100,
with the adjoining horizontal tie, ACI 318-08 Appendix A No. 6, Nov.-Dec., pp. 704-712.
recommends a constant strut efficiency factor of 0.75 for all Sahoo, D. K., 2009, “An Investigation of the Strength of Bottle-Shaped
of these reinforced bottle-shaped struts. For strut inclinations Struts,” PhD thesis, Indian Institute of Technology Roorkee, Roorkee,
India, 324 pp.
smaller than approximately 30 degrees, Fig. 12 shows that
Sahoo, D. K.; Singh, B.; and Bhargava, P., 2009, “An Appraisal of the
the ACI recommended strut efficiency factor may be uncon- ACI Strut Efficiency Factors,” Magazine of Concrete Research, V. 61, No. 6,
servative when compared to the experimental results of the Aug., pp. 445-456.
authors. Interestingly, for all of these beams, including those
with short anchorage lengths, the authors’ experimentally AUTHORS’ CLOSURE
observed strut efficiency factors do have a sufficient margin The authors appreciate the discussers’ interest in their
of safety when compared with the predicted strut efficiency paper. We would first like to thank the discussers for
factors obtained from a recent model by Sahoo (2009), shown pointing out notation errors in Eq. (12) and (18) in the paper.
as follows. This allows us to correct these equations and fix other errors
found in the manuscript. Figure 11(b) should be modified to
be consistent with the notation given in Fig. 11(a) included
0.05 α
β s = ⎛ 0.60 + ---------- + 55ρ T⎞ ------ (22) in this discussion.
⎝ rc ⎠ 90
FC(L) = FC(R) – FS-trusscosγ (12)
In Eq. (22), (valid for up to 81 MPa [11,748.05 psi]), α is
the angle of inclination of the diagonal strut with the tie in Vtruss = FS-trusssinγ (18)
degrees, and rc is the concentration ratio of the load resisted
by the diagonal strut obtained as the ratio of ws-top and half Finally, the values reported in the last column of Table 3
the strut length. The effective transverse reinforcement ratio ρT are ratios of calculated-to-test values of stresses in the strut
is computed from the corrected version of the transformation for a tied-arch model. Therefore, the header for the last
used in ACI Eq. (A-4), shown as follows (Sahoo et al. 2009) column in Table 3 should be modified to read fS-TA/fS-TA(test).

492 ACI Structural Journal/July-August 2010

The authors would now like to provide closing comments
in response to points made by the discussers. The discussers
correctly point out that the load-transfer mechanism in deep
beams is not only influenced by a/d, but also by the amount
of transverse web reinforcement. To isolate the influence of
a/d in the load-transfer mechanism, the authors designed the
deep beams in our tests with the same amount and spacing of
transverse reinforcement. The main objective of the research
was to identify the effect of short bar anchorage at the
support on the load-transfer mechanism. Certainly a more
complete study would include variations in the amount and
spacing of transverse reinforcement while holding a/d
constant—this was, however, outside the scope of our tests.
The discussers point out that by using the minimum
permissible angle between a strut and a tie in accordance
with ACI 318-08, one could determine a maximum a/d of
2.14, where load could be transferred directly into the
support using a tied-arch model. Although one could
certainly design a beam that falls in this a/d range using only
a tied-arch model and in compliance with ACI 318-08, loads
can also be transferred indirectly into the support through
truss action for a/d less than 2.0, as has been previously
demonstrated (refer to Eq. (4) from FIP [1999] in the paper).
Furthermore, the authors contend that in the case of deep
beams with anchorage details that do not ensure yielding of
the bottom tie at the face of the extended nodal zone above
the support (a requirement needed to satisfy equilibrium in a
tied-arch model), the fraction of load transferred through
truss action might be higher than for a beam with appropriate
anchorage in this region, provided that enough transverse
reinforcement exists to support this load transferred by truss
action. The fact that specimens having longitudinal bar
anchorages shorter than required to develop yielding of the
bottom tie were able to support loads comparable with those
with full development demonstrates that a fraction of the
total load was being transferred through a truss mechanism.
The discussers point out that it would be better to use an
indeterminate strut-and-tie model involving tied-arch and Fig. 11—(revised from original paper)—Truss model for load
truss mechanisms to determine the fraction of shear transferred transfer: (a) geometry of model; and (b) top node details.
by each model. The authors would like to remind the
discussers that, to solve an indeterminate strut-and-tie model,
one must either assume the fraction of load transferred by each
individual mechanism or determine this load fraction in
proportion to the individual submodel stiffnesses. The best
way to verify our experimental results was to separate the 100% of the applied shear, that would mean that the adopted
problem into two statically determinate models where the procedure to determine the individual load-transfer fractions
load being transferred could be verified by independent was flawed or that experimental measurements were not
measurements taken during the tests. These results were reliable. As mentioned in the paper and included in Table 3,
intended to provide information on the fraction of load being the largest difference between the load needed to be transferred
transferred by each potential load-transfer mechanism to by tied-arch action to ensure that 100% of the applied shear
provide guidance for the future use of indeterminate models was carried and the load determined through potentiometer
for this type of structure. The load carried by each mechanism measurements was approximately 12%. We believe that this
was estimated through independent experimental measurements small difference gives reasonable confidence about the proce-
taken along the direction of struts in the assumed tied-arch dure employed to estimate the fraction of load transferred by
and truss models for each group of beams. The authors tied-arch and truss mechanisms.
would like to emphasize that we were using potentiometers The variation of strut efficiency factor βs that the
that measured the overall axial shortening of the relevant discussers present in Fig. 12 is consistent with results plotted
struts (the potentiometer ends were attached to points located in Fig. 9. The main difference is that βs is plotted in Fig. 9 as
at the ends of struts). This allowed us to experimentally a function of a/d instead of direct strut angle α, but these two
determine the average axial force-displacement relationship quantities are directly related. The authors are also quite
of each strut and avoid basing our results on local strain satisfied that the experimentally determined βs values follow
readings. The total load (sum of forces carried by each the same trend as the discussers’ model. We thank the
individual mechanism) would of course have to add to 100% discussers for making us aware of their model and look
of the applied shear. If the total load carried did not add to forward to studying the reference they provided in detail.

ACI Structural Journal/July-August 2010 493

Disc. 106-S66/From the September-October 2009 ACI Structural Journal, p. 706

Punching Strength of Reinforced Concrete Footings. Paper by Josef Hegger, Marcus Ricker, and Alaa G. Sherif

Discussion by Andor Windisch

ACI member, PhD, Karlsfeld, Germany

The authors should be complimented for their interesting seems to be negligible.” The distance s0, the spacing between
test series and informative report. The saw cuts of different the column face and the first row of shear reinforcement is
test specimens (Fig. 5) especially allow for very informative decisive.
insights into the failure process of the footings and some
shortages of the code provisions. Steel strains
The footings without shear reinforcement did not fail in As revealed before, the position of the strain gauges was
direct punching: the failure was caused by the failure of the not optimal. The flexural steel strains measured at failure of
anchorage of the flexural reinforcement along the perimeter the footings with shear reinforcement reached yielding as the
of the specimens. This can be observed at the horizontal shear reinforcement let the inner crack develop in the neigh-
cracking along/above the flexural reinforcement at the edges borhood of the strain gauges.
of the specimens and at the many failing outer corners of
practically each and every footing without shear reinforce- DISCUSSION OF EXPERIMENTAL RESULTS
ment. These local failures could partly be caused by the Effect of a/d
loading pattern with the sliding bearings near edges The slenderness ratio a/d is an indirect indicator of the
modeling a uniform surface load, which is practical for test possible inclination of the failure shear surface only. At this
reasons but not realistic. These circumstances could lead to point, a direct factor (the inclination) should be introduced in
a decision not to consider these specimens in the discussion. the codes.
Nevertheless, as the detailing of the reinforcement corre-
sponds to the practice, it is mandatory to tackle the results as Effect of concrete compressive strength
they help to elucidate the shortages of the code provisions. The authors’ conclusion is correct: the behavior of the
footings can not be described by a strut-and-tie model. The
EXPERIMENTAL PROGRAM failure load in shear is not controlled at all by the bearing
Material properties capacity of the compressive strut. The “source” of the ultimate
It is not clear at which age the test specimens were loaded. load in shear is the shear load-bearing capacity of the
Was it around the 28th day? concrete compressive zone. Before the failure occurs, the
shear crack, as part of the failure surface, is so wide that no
Test setup aggregate interlock or similar sidelines could be drawn on to
Even if Fig. 4(b) does not yield detailed information about explain the behavior. The influence of the increasing
the real loading pattern, the discusser has the impression that concrete strength is neither linear nor follows any square
the specimens of Series II were relatively overloaded along root relationship. Further fundamental research is needed
their perimeter; the outer sliding bearings were too near to regarding this.
the edges of the specimens.
The locations of the flexural steel strain gauges as shown Effect of shear reinforcement
in Fig. 4(b) are not optimal. At choosing these locations, the As explained previously, at least in case of Specimens
staggering of the tension line due to the shear force (refer to DF16 to DF18, the shear reinforcement was not activated as
the inclined cracks in Fig. 5), too, should have been taken part of the shear load-bearing mechanism; hence, any
into consideration. conclusion—for example, based on Fig. 10—would be
Cracking and failure characteristics Effect of soil-structure interaction
All test specimens without shear reinforcement failed due It is a pity that the saw cut of the test specimen with shear
to the failure of the anchorage of the flexural reinforcement. reinforcement, Specimen DF 9, supported on sand, was not
The authors declare correctly that “the failure occurred presented in the paper.
along/due to the wide shear crack and the inclination of the
cracks in these specimens is determined by the ratio a/d.” COMPARISON OF PREDICTIONS AND
The saw cuts of the uniformly loaded footings with shear EXPERIMENTAL RESULTS
reinforcement reveal an extremely important and forward- Comparison with ACI 318-08
looking fact: as the inner work required to open the outer Figure 12(b) seems to confirm that ACI 318-08 does not
shear crack crossing the shear reinforcement was higher than consider the influence of a/d. Nevertheless, the real influence
the inner work causing the much steeper inner shear crack, of the effective depth, as shown in Fig. 12(a), should not be
even though the shear reinforcement was activated, it evaluated without considering the impact of a/d, that is, the
determined the position of the failure surface but not the size inclination of the failure surface. The specimens supported
of the ultimate failure load. The authors correctly specify on sand with different effective depths had different a/d too.
that “in contrast to the footings without shear reinforcement, The position of the critical perimeter b0, accompanied with
the influence of a/d on the inclination of the shear cracks the assumption of the uniform soil pressure distribution,

494 ACI Structural Journal/July-August 2010

Fig. 16—Crack patterns of Footings DF11 to DF13 after failure.

approximates the real behavior on the safe side (namely, address some of the points raised in the discussion to provide
conservative) two times; refer to the crack pattern in Fig. 5(iv) some clarification. Because of space limitations, the most
(Fig. 5(h) shows Specimens DF7, not DF17). The crack important points, such as the type of failure, will be
pattern of Specimen DF9 should have been shown, too, to discussed in more detail.
realize the impact of the shear reinforcement on the failure
behavior of sand-supported footings. EXPERIMENTAL PROGRAM
Failure characteristics
Comparison with Eurocode 2
The discusser assumes that the failure of the footings
The authors are correct in that:
without shear reinforcement was caused by bond failure.
• The failure crack patterns of the footings without shear
This is not correct. In Fig. 16, the crack patterns of three foot-
reinforcement seem to prove the position of the basis
ings without shear reinforcement are presented. All footings
control perimeter at 2.0d distance from the column
face; and were loaded via 16 bearings, simulating a uniform load case.
The crack patterns are typical for a punching shear failure. At
• It is odd that VRu,max according to Eq. (17) controls the
design. Being only a function of the concrete compres- first, radial cracks around the column appeared, then the first
sive strength, it seems to refer to a web-crushing limit tangential cracks developed at the column face and, later on,
although, in case of the tested footings, no crushing at higher load stages, more and more tangential cracks
could occur. appeared. The ultimate punching capacity of the slab was
The authors criticize that VRu,max does not reflect a/d achieved when the inclined failure crack reached the flexural
correctly. The discusser agrees and suggests that in the case reinforcement. Because the failure was relatively brittle and
of footings, a control like VRu,max has no meaning at all. the tests were load controlled, it was not always possible to
Instead of VRd,max, the authors propose a new equation, Eq. stop the test right in time. After the failure took place, the
(19), which is a follow-up of Eq. (15). It would be informa- load was removed and, afterward, the specimens were
tive to learn how the multiplier 16 ( d/u 0 ) has been found. reloaded to determine the bearing capacity after punching
Whereas vRd,c (Eq. (15)) yields a lower limit, VRd,max (Eq. failure. During this second loading phase, the flexural rein-
(19)) sets an upper limit of the load-bearing capacity. forcement was heavily deformed in the region of the failure
Comparing Eq. (19) to Eq. (15), the margin between them crack, which led to the spalling concrete at some edges.
seems to originate from the geometry, that is, ( d/u 0 ) . Why In contrast to Fig. 2, in the footings without shear rein-
does the geometry not influence the punching shear-stress forcement, the flexural reinforcement was twice bent-up
resistance vRd,c either? The authors are kindly asked to clarify. (Fig. 17). Thus, for the given geometry, a typical bond
It would be desirable, too, that the odd “best fit” form of failure is nearly impossible. The footings with shear rein-
(100ρ · fck)1/3 in Eq. (15) will be substituted with a physically forcement were identical to the specimens without shear
sound term in the New Model Code of fib under preparation. reinforcement in terms of flexural reinforcement and
Even if—according to Fig. 15(d)—Eq. (19) seems to yield concrete strength. Although the anchorage of the flexural
a “safe” upper limit, the rate of approximation depends on a/d reinforcement was weaker (90-degree hooks), the specimens
and is, in the case of footings supported on sand, very conser- with shear reinforcement reached a higher failure load. This
vative. For the time being, the limited number of the test should clearly indicate that at least the footings without shear
specimens with shear reinforcement, that is, with different reinforcement and 180 (2 times 90) degree hooks did not fail
ratios d/u0 and different spacings between the column face prematurely by bond failure. Hallgren et al.5 reported the
and the first row of shear reinforcement, s0, do not allow for results of 14 punching shear tests on reinforced column foot-
final acceptance of the proposed VRd,max (Eq. (19)). ings. In these tests, among other parameters, the influence of
The authors are kindly encouraged to continue their the end anchorage of reinforcement was investigated system-
interesting research work. atically. Hallgren et al.5 concluded that the anchorage of the
flexural reinforcement has only a small influence on the
AUTHORS’ CLOSURE punching strength of footings; however, curved anchorage
The authors are grateful for the comments and the interest improved the ductility. It is worth mentioning that the
in the paper. In the present closure, the authors would like to Swedish saw cuts and crack patterns are very similar to those

ACI Structural Journal/July-August 2010 495

that the failure load is sensitive to the spacing between the
column face and the first row of shear reinforcement s0. To
investigate this effect in more detail, Ricker14 conducted a
finite element analysis. In Fig. 18, the calculated failure
loads VFE are plotted against s0/d (with d being the effective
depth). The finite element analysis showed that a reduction
of the spacing between the column face and the first row to
s0 = 0.2d leads to an increase in failure load between 10 and
14%. However, to clarify the influence of this parameter,
further tests are needed.
Fig. 17—Saw cut of Footing DF11. Figure 19 presents the saw cut of Specimen DF9 with
shear reinforcement (supported on sand). The crack pattern
is comparable to those of the uniformly loaded specimens.
The concrete is slightly more crushed because the specimen
was overloaded when the bearing capacity after failure was
determined during a second loading phase. The inclination
of the failure shear crack is very steep (approximately 50 to
60 degrees). The crack propagates from the column face to
the anchorage of the first row of shear reinforcement
consisting of stirrups.


The equation for the calculation of the maximum punching
capacity was originally derived for flat plates by a regression
analysis.15 For the present paper, this equation was adapted
to footings. Due to a lack of suitable tests, Eq. (19) was
Fig. 18—Result of finite element calculations performed to adapted in such a way to determine lower-bound values for
investigate influence of distance from column face to first the maximum punching-shear capacity of footings as
row of shear reinforcement s0 on failure load. correctly mentioned by the discusser. According to Euro-
code 2, a critical perimeter at the periphery of the loaded area
was chosen for the calculation of the maximum punching-
shear capacity. In contrast, the punching-shear resistance
without shear reinforcement, according to Eq. (15), is verified
at control perimeters within 2.0d. The use of a relatively
large distance to the control perimeter has the advantage that
the correlation with test data over the normal range of
column dimensions to effective depth is reasonable (refer to
Reference 11). This can be explained by a reduction of the
influence of the uneven shear stress distribution resulting
from the type of column (for example, circular or rectangular)
and the column dimensions. Thus, for control perimeters far
Fig. 19—Saw cut of Footing DF9. away from the periphery of the loaded area, the column
dimensions need not be considered directly.

in the present tests and also showed spalling concrete at the REFERENCES
edges of the specimens. 14. Ricker, M., “Zur Zuverlässigkeit der Bemessung gegen Durchstanzen
For the footings with shear reinforcement, the discusser bei Einzelfundamenten,” doctoral thesis, RWTH Aachen University,
Aachen, Germany, 2009, 304 pp.
correctly mentions that the shear reinforcement determined 15. Hegger, J.; Häusler, F.; and Ricker, M., “Zur Durchstanzbemessung
the position of the failure surface but not the value of the von Flachdecken nach Eurocode 2,” Beton-und Stahlbetonbau, V. 103, No. 2,
ultimate failure load. The discusser also correctly presumed Feb. 2008, pp. 93-102.

496 ACI Structural Journal/July-August 2010