ACI STRUCTURAL JOURNAL TECHNICAL PAPER

Title no. 99-S84

Plasticity Model for Predicting Punching Shear Strengths

of Reinforced Concrete Slabs
by W. Salim and W. M. Sebastian

In this paper, the upper-bound theory of plasticity is employed to

predict the punching shear failure loads of reinforced concrete
slabs without shear reinforcement and without in-plane restraint. A
parabolic Mohr failure criterion is adopted for the concrete to
ensure that the important variation in angle of friction of the concrete
with stress state is represented, with the material assumed to be rigid-
perfectly plastic. The problem is treated as three-dimensional
axisymmetric. Using this approach, the generatrix of the failure
surface is shown to be a curve. A simplified expression, obtained
by approximating this generatrix to a line, is also derived for the
punching strength. The sensitivities of the predicted failure loads
to variations in the tension and compression effectiveness factors
used for the concrete are investigated. It is found that the predictions
correlate well with a range of experimental data for low-, normal-,
and high-strength concretes, and for both small-scale and large-
scale slabs.

Keywords: punching shear; slab; strength.

Fig. 1Modified Mohr-Coulomb failure criterion for concrete.

INTRODUCTION
Few problems in concrete have been studied more extensively
than the punching shear failure phenomenon of concrete slabs.
Despite considerable experimental and theoretical work,
however, it has proven extremely difficult to develop a
consistent, meaningful theoretical approach for punching
shear failure. As a consequence, design codes of practice
have relied on either purely empirical or semi-empirical
approaches for predicting punching loads. Therefore,
there remains a real need to develop a theoretical model in
which the important physical characteristics are clearly and
properly represented. To that end, two main modeling
approaches, namely the upper-bound approach based on
plasticity theory and the lower-bound approach founded on
satisfying equilibrium requirements, have had encouraging
success. One advantage of using the plasticity-based method
is that only effectiveness factors relating concrete compressive
and tensile strengths have to be determined through calibration
of theoretical results against experimental results.
Braestrup et al. (1976) developed a rigid-plastic model for
the analysis of punching shear failure in concrete slabs.
Predictions from this model showed good agreement with
experimental punching loads. Jiang and Shen (1986), following
Leons (1935) concept, presented an alternative plastic
method for analyzing punching shear failure of concrete
slabs in which concrete was assumed to be rigid-plastic, with
concrete material failure modeled by a parabolic intrinsic
curve. An extension was made to Braestrup et al.s plastic
theory (1976) of punching shear in flat slabs by Morley
(1979) to enable spherical shells to be treated. Also based on
Braestrup et al.s method, Bortolotti (1990) derived a plastic
solution for punching shear strength of concrete slabs in which
the influence of softening concrete was taken into consideration.
In the aforementioned studies, the level of success
achieved in predicting the punching loads has been signifi-
cantly influenced by the concrete material models used. In
particular, the effectiveness factors used for concrete in tension
and compression can play important roles in determining the
level of accuracy of the predictions. In the present paper,
alternative concrete material models are used along with
rigid plastic theory to provide new predictive analyses for
punching shear. There are three main aspects to these new
analyses. First, concrete material models are used that incor-
porate the important variation in angle of internal friction
with stress state in the concrete. Second, two alternative
predictive punching shear expressions are presented, one
based on using a curve for the generatrix of the failure surface,
and the other a simplified solution based on approximating the
failure generatrix to a straight line. Third, the sensitivity of
the predicted failure loads to variation in the concrete material
effectiveness factors is investigated.
FAILURE CRITERION FOR CONCRETE
For the concrete material modeling, Braestrup et al. (1976)
used the modified Coulomb failure criterion, which incorporates
a constant angle of internal friction (Fig. 1) to tackle the
punching shear problem. In the present study, a parabolic
Mohr failure criterion, shown in Fig. 2 and which incorporates
the variation in angle of internal friction with stress state in the
ACI Structural Journal, V. 99, No. 6, November-December 2002.
MS No. 01-386 received November 19, 2001, and reviewed under Institute publica-
tion policies. Copyright 2002, American Concrete Institute. All rights reserved,
including the making of copies unless permission is obtained from the copyright pro-
prietors. Pertinent discussion will be published in the September-October 2003 ACI
Structural Journal if received by May 1, 2003.

ACI Structural Journal/November-December 2002 827

W. Salim received his BEng from the University of Nottingham, Nottingham,
England. He is currently doing research in the Department of Civil Engineering at the
University of Bristol, Bristol, England. His research interests include the punching
shear failure of reinforced concrete slabs.

W. M. Sebastian is a lecturer in the Department of Civil Engineering at the Univer-

sity of Bristol. He received his PhD from the University of Cambridge, Cambridge,
England. His research interests include composite materials and various aspects of
reinforced and prestressed concrete.

Fig. 3Yield loci in cases of plane stress and plane strain.

c2 2
1 = f t 1 ----k ( csc 1 ) (5)
4

c2 2
3 = f t 1 ----k ( csc + 1 ) (6)
Fig. 2Parabolic Mohr-failure criterion for concrete. 4

material, is used to represent the real mechanical characteristics The yield loci for the cases of plane stress and plane strain
of concrete. As shown in Fig. 2, the parabola touches the are shown in Fig. 3.
Mohrs circles for simple compression and simple tension. It The aforementioned formulation is chosen for conve-
is important to note that the concrete material strengths used nience. It is possible to rewrite this formulation in terms of
in Fig. 2 are the effective uniaxial tensile ft and compressive the effective compressive strength fc and the effectiveness
fc strengths, as opposed to the actual measured uniaxial factor t (ratio of ft to fc) as follows
tensile and compressive strengths. This is achieved by applying
effectiveness factors to the actual concrete material
1 2
strengths, and is discussed further in the following section. 1 = f c t ----t + 1--- --- t ( 1 + t ) ( csc 1 ) (7)
The parabolic Mohr failure criterion is expressed as 2 4 2

- 2
------- + --- = 1 (1)
c k ft 1 2
ft 3 = f c t ----t + 1--- --- t ( 1 + t ) ( csc + 1 ) (8)
2 4 2
where and are shear and normal stresses on an arbitrary
plane, and ck is a parameter related to the ratio of the effec- The significance of the effective concrete material
tive material strengths of concrete as follows strengths, and of the effectiveness factors used to determine
those strengths, are discussed as follows.
f
ck = 1 + ---c 1 (2) EFFECTIVENESS FACTORS
ft
The rigid-perfectly plastic theory assumes that ultimate
conditions are attained along the failure surfaces at zero
The shear and normal stresses in the parabolic Mohr failure deformation, and that these ultimate conditions continue
envelope can be given as functions of the parameter defined to prevail via the normality rule at least up to the deformations
in Fig. 3 by corresponding to overall failure of the structure. In other words,
the slabs are assumed to be ductile in the shear failure zones.
Concrete shows very little ductility, however; in compression,
c2
= f t ----k cot (3) concrete is both nonlinear and of limited ductility, while
2 in tension, the material exhibits brittle behavior at low
stress and subsequently displays monotonic strain softening
behavior to zero stress at large strains. This is clearly at odds
c2 2
= ft 1 ----k cot (4) with the assumption of perfectly plastic material, which
4 implies a plateau in the material stress-strain behavior under
uniaxial loading. Fortunately, for the slabs considered in this
From Mohrs circle, the maximum and minimum principal study, extensive yielding of the tension steel reinforcement
stresses can be found to be occurs during the punching failure regime. As a result, the

828 ACI Structural Journal/November-December 2002

Fig. 4Stress-strain relationship for concrete.
limited ductility of the concrete is overshadowed by the Fig. 5Deforming zone between two rigid parts.
considerable ductility of steel, so that the slabs exhibit
very ductile punching failure characteristics overall. the generatix of the failure surface. In the present study, one
For analytical purposes, these overall ductile characteristics curved generatrix and one straight generatrix are considered.
permit fictitious perfectly plastic stresses to be assigned to For each generatrix, the predicted punching shear loads
the concrete in both tension and compression, consistent emerging from use of the previously mentioned four combi-
with the requirements of plastic theory. This is done by nations of effectiveness factors will be compared.
applying effectiveness factors to the actual compressive
strength of the concrete to give the effective compressive WORK EQUATION
and tensile strengths used in Eq. (1) to (8). Henceforth, the The yield zone between adjacent bodies moving in one
effectiveness factors used to determine these effective plane can be investigated by considering rigid moving
compressive and tensile strengths will be termed c and blocks separated by a homogeneous plane strain field of
t, respectively. In general, effectiveness factors are determined thickness , as shown in Fig. 5. If the angle between the relative
by calibration against test data, with the calibration used displacement rate u and the surface is , the strain rates in
depending on the material model under consideration. As the deforming zone are
illustrated in Fig. 4, one method of determining the effectiveness
factor is based on retaining the strain to failure of the material t = 0 (11)
and equating the area underneath the actual stress-strain curve to
that underneath the effective ultimate stress-strain rectangle.
u
Clearly, the effectiveness factors used will influence the n = --- sin (12)
degree of accuracy of the predictions. Hence, it seems prudent
to establish the level of sensitivity of the present predictions of
punching shear failure loads to different effectiveness factors. u
nt = --- cos (13)
For that reason, two alternative tension effectiveness factors
and two alternative compression effectiveness factors are
used in this study. The compression effectiveness factors, The principal strain rates can therefore be determined by
taken after Sigurdsson (1991) and Nielsen (2000), respectively,
are as follows
1 1 2
---- = --- [ ( n + t ) ( n t ) + nt ] = (14)
3 2
= ---------- 1 + --------------------
0.48 - ( 1 + 0.125 ) or
1.47
C (9a,b)
f c 0.001h
u-
----- ( sin 1 )
2
C = 4.22 f c
According to the normality flow rule of the theory of plasticity,
Note the dependence of the Sigurdsson factor on the tension using Eq. (5) and (6) with plastic strain rates being considered
steel reinforcement ratio , the concrete uniaxial cylinder
compressive strength, and the thickness of the slab. The two d sin 1-
alternative tension effectiveness factors considered in this --------1- = ----3 = -------------------- (15)
d 3 1 sin + 1
study, both taken from Nielsen (2000), are as follows

t = f c f t = 400 or t = f c f t = 100 (10a,b) From Eq. (14)

3 sin 1
One compression and one tension effectiveness factor ---- = ---------------------- (16)
must be used in the punching shear analysis. Equation (9) 1 sin + 1
and (10) permit up to four different potential combinations of
factors. Also, it is shown in a following section of this paper that It is evident from the comparison between Eq. (15) and
the predicted punching shear load depends on the shape of (16) that if, and only if, = can the stress state described

ACI Structural Journal/November-December 2002 829

 c2
P = 2 f t r + ------k- r dx
h

0
4r
(19)

which can also be expressed as

h
0
P = 2 f t F ( r ,r ) d x (20)

where

c2
F ( r , r ) = r r + ------k- (21)
4r
Fig. 6Failure generatrix.
Equation (19) shows that the punching shear strength depends
by the parabolic Mohr yield condition (Fig. 2) produce the on the shape of the failure surface and the related angle or, in
strain vector given by Eq. (14). Internal energy dissipation other words, it depends on the generatrix of Fig. 6. The function
per unit area in the deforming zone for the parabolic material is r(x) that minimizes the ultimate punching load can be found
by calculus of variations. The appropriate Eulers equation
c2 2 that is for a function of the form
D A = ( 1 1 + 3 3 ) = uf t 1 + ----k cot sin (17)
4
h

This equation is applicable for any condition of plane

0 F ( r , r ) dx
strain and for plane stress on the tension side of Points A, A
in Fig. 3, that is, for the ratio 1/3 exceeding a certain limit. in minimizing r = r(x), the function can be written as
For other strain ratios in plane stress, the stress point will
be at A or A in Fig. 3, and the energy dissipation will be F -----d F
------ - ------- = 0 (22)
r dx r
correspondingly different.
It is observed from Eq. (17) that the energy dissipation is
totally independent of the thickness , indicating that a yield
line is obtained in the limiting case when the value of tends Because F(r, r ) does not contain x, Eulers equation has the
toward zero. first integral

F
UPPER-BOUND SOLUTION F r ------- = C (23)
The Upper-Bound Theorem of Plasticity states that if, for r
any assumed failure mechanism, the external rate of work is
equated to the rate of dissipation of internal energy, then an where C is a constant.
upper-bound for the collapse load of the structure can be found. By substituting Eq. (21) into Eulers equation, with con-
Figure 6 shows one of the failure mechanisms assumed sideration of the following boundary conditions
herein. It is seen that the incremental movement of the
punched-out block of concrete is vertically down the page, d d
while the remaining block of peripheral concrete is assumed r ( 0 ) = ----0- and r ( h ) = ----1- (24)
2 2
not to move. As a result, the relative displacement rate vector
is oriented vertically along the entire length of the generatrix
the minimizing function of the generatrix is obtained as
of the failure surface. This, along with the requirement that
= , leads to the convenient and useful result that dr/dx =
r = tan = tan. In Fig. 6, the rate of work done by the d d xh
r = ----0- ----1- (25)
external load on the slab under punching is 2 d 0

WE = P * u (18) Once the minimizing function of the generatrix has been

found, the least upper-bound solution for the punching load
Consistent with the upper-bound procedure, one can state can be obtained from Eq. (19)
work-energy equilibrium for Fig. 6 as
2 2
2 2 2c k h
P = --- f t d 1 d 0 + -----------------------
- (26)
r = 0 DA ( 2 r ds )
r=h
WE = W A = 4 ln ( d 1 d 0 )

where ds is the increment of length along the generatrix Eq. (26) shows that the collapse load varies with the outer
of the punching shear failure surface. Combining this expression diameter of the punched cone d1. If the support diameter D
with Eq. (17), (18) and the geometric relationships r = tan is sufficiently large, there is a critical value of d1 that is found
= tan and ds = dx/cos gives the following from consideration of P/d1 = 0, which gives

830 ACI Structural Journal/November-December 2002

Fig. 7Simplified solution. (a)

d 1 d1
----
- = e
ck h
(27)
d 0

SIMPLIFIED UPPER-BOUND SOLUTION

The upper-bound solution defined in Eq. (26) can be
simplified by assuming the yield line to be straight in elevation
(Fig. 7), in which case the equation of the generatrix simplifies to

d
r = ----0- + x tan (28)
2

By substituting Eq. (28) into Eq. (19) and using Eq. (7) and (b)
(8) gives

P = f c ( t tan + z cot ) ----0- + tan h

d 2
(29)
h

where

1 1
z = ----t + --- --- t ( 1 + t ) and f c = C f c
2 4 2

The extreme of the collapse load is found from consider-

ation of P/ = 0 which gives
(c)
3 2 h
cot ----t cot --- ----- t = 0 (30)
z z d0

Equation (30) has a real root for , which is the critical

punch angle 1 given by

t h t d0
2
1 = arctan 3 ------------ 1 + 1 ------------------ + (31)
z d0 27z h
2

t h t d0
2

3 ------------ 1 1 ------------------
z d0 27z h
2

(d)
Thus, the critical diameter is
Fig. 8Correlation of predictions by complex method with
97 test results using: (a) Eq. (9a) and (10a); (b) Eq. (9a) and
d 1 = d 0 + 2h tan 1 (32)
(10b); (c) Eq. (9b) and (10a); and (d) Eq. (9b) and (10b).

ACI Structural Journal/November-December 2002 831

Table 1Slabs tested by various researchers Table 1 (continued)
Researchers and span Slab no. d0, mm d, mm , % fcu , N/mm2 Researchers and span Slab no. d0, mm d, mm , % fcu , N/mm2
A 102 57.3 1.083 33.1 1 100 40.5 0.423 38.4
B 102 57.3 1.083 35.7 2 100 40.5 0.558 38.4
C 102 57.3 1.083 32.8 3 100 40.5 0.691 38.4
D 102 57.3 1.083 34.2 4 100 40.5 0.821 43.5
Base
E 102 57.3 0.725 37.3 5 100 40.5 0.883 43.5
(span = 559 mm)
F 102 57.3 0.725 34.7 6 100 40.5 1.026 43.5
G 102 57.3 1.635 36.4 7 100 40.5 1.163 37.1
H 102 57.3 1.635 33.0 8 100 40.5 1.292 37.1
J 102 57.3 3.270 35.1 9 100 40.5 1.454 37.1
1A30a24 300 128.0 1.01 32.4 10 100 40.5 0.517 37.4
Kinnunen and 1A30a25 300 124.0 1.04 30.8. 11 100 40.5 0.802 37.4
Nylander
(span = 1710 mm) 1A15a5 150 117.0 0.80 32.9 Rankin and Long 12 100 40.5 1.107 37.4
1A15a6 150 118.0 0.79 32.1 (span = 640 mm) 13 100 40.5 0.601 42.5
S1-60 254 114.3 1.06 29.3 14 100 40.5 0.691 42.5
S5-60 203 114.3 1.06 27.8 15 100 40.5 1.994 42.5
S1-70 254 114.3 1.06 30.6 2A 100 46.5 0.691 36.0
Moe
S5-70 203 114.3 1.06 28.9 3A 100 46.5 1.293 36.0
(span = 1778 mm)
H1 254 114.3 1.15 32.6 4A 100 46.5 1.992 38.6
R2 152 114.3 1.15 33.3 3B 100 35.0 1.292 47.1
M1A 305 114.3 1.50 26.1 4B 100 35.0 1.994 38.6
SS2 200 77 1.20 29.3 1C 100 53.5 0.423 34.8
Regan SS4 200 77 0.92 40.4 2C 100 53.5 0.690 40.5
(span = 1892 mm) SS6 200 79 0.75 27.4 3C 100 53.5 1.288 40.5
SS7 200 79 0.80 38.0 4C 100 53.5 1.993 34.8
1 60 30 1.20 38.4
2 60 30 1.20 38.4
range of scales, with effective depths from 30 to 250 mm,
are considered;
3 60 60 1.20 34.1
The value of h is approximated to the effective depths
4 40 30 1.20 38.4
of the slabs;
5 60 30 0.50 27.5 The present analyses assume a circular punch, but some
6 60 30 1.20 38.4 of the experiments used square punches. In such cases,
Dragosavic and
van den Beukel 7 60 30 1.73 27.7 an effective diameter (d0) 10% greater than the side
(span = 425 mm)
15 60 30 0.60 31.1 length of the square is assumed (after Aoki and Seki
16 60 30 0.90 29.5 [1971]). This gives a circular punch of equal plan area
17 60 30 1.30 29.5 to the experimentally employed square punch;
18 60 30 1.70 29.5 In applying the compression effectiveness factor due to
19 60 30 2.10 29.5 Sigurdsson (Eq. (9a)), the tension steel reinforcement
20 60 30 2.50 29.5
ratio is required. For most of the slabs considered, this
ratio was equal along the two main perpendicular direc-
tions of the slab. For Moes tests, however, these ratios
COMPARISON WITH EXPERIMENTAL DATA
were different along the two perpendicular directions
To obtain predictions of punching shear failure loads, the (this explains why two reinforcement ratios are given
present analyses were implemented in MATLAB. Predic- for each of Moes slabs in Table 3). This issue was dealt
tions from these analyses have been compared with the results with by using, in Eq. (9a), the average of the tension
of 97 tests conducted by various researchers (Base [1959]; steel reinforcement ratios along the perpendicular
Kinnunen and Nylander [1960]; Moe [1961]; Dragosavic directions;
and van den Beukel [1974]; Elstner and Hognestad [1956]; Table 1 and 2 give the cube strength fcu and cylinder
Criswell [1974]; Regan [1978]; Rankin and Long [1987]; compressive strength fc respectively of the concrete,
and Hallgren and Kinnunen [1996]). The relevant geometric because these were the data on concrete compressive
and material-property details of the slabs are given in Table strength supplied in the literature. The present analyses,
1 and 2. The comparisons between predicted and measured however, are based on the cylinder strength. For cases
punching shear failure loads are given in Fig. 8 and 9, while where only the cube strength (and not the cylinder
a statistical summary of the correlations of Fig. 8 and 9 are strength) was given, the cylinder strength was taken as
presented in Table 3. Note the following points in relation to 80% of the cube strength; and
the predictions: Figure 8 and 9 show the comparisons based on the
Henceforth, the terms complex and simplified will refer complex (curved generatrix) and simplified (linear
to the analyses based on the curved generatrix and the generatrix) analyses, respectively. As explained pre-
linear generatrix, respectively, of failure surface; viously, four different combinations of compression
Concretes over a wide range of cylinder compressive and tension effectiveness factors, taken from Eq. (9)
strengths from 14 to 109 N/mm2 and slabs over a wide and (10), have been used with each analytical approach.

832 ACI Structural Journal/November-December 2002

Table 2Slabs tested by various researchers
Researchers and span Slab no. d0, mm d, mm , % fcu, N/mm2
A1a 254 117.6 1.15 14.1
A1b 254 117.6 1.15 25.3
A1c 254 117.6 1.15 29.1
A1d 254 117.6 1.15 36.9
A1e 254 117.6 1.15 20.3
A2a 254 114.3 2.47 13.7
A2b 254 114.3 2.47 19.6
A2c 254 114.3 2.47 37.5
A7b 254 114.3 2.47 28.0
Elstner and A3a 254 114.3 3.70 12.8
Hognestad (a)
(span = 1778 mm) A3b 254 114.3 3.70 22.7
A3c 254 114.3 3.70 26.6
A3d 254 114.3 3.70 34.6
A4 356 117.6 1.15 26.2
A5 356 114.3 2.47 27.8
A6 356 114.3 3.70 25.1
B4 254 114.3 0.99 47.8
B9 254 114.3 2.00 44.0
B11 254 114.3 3.00 13.5
B14 254 114.3 3.00 50.7
S2075-1 254 120.6 0.75 32.5
S2075-2 254 122.2 0.75 29.1
Criswell S2150-1 254 124.0 1.50 29.7 (b)
(span 2134 mm) S2150-2 254 122.2 1.50 30.2
S4150-1 508 125.5 1.50 35.5
S4150-2 508 125.5 1.50 35.8
S2-60 254 114.3 1.53, 0.84 22.1
S3-60 254 114.3 2.3, 0.54 22.7
Moe-Differing
perpendicular S4-60 254 114.3 3.45, 2.65 23.9
reinforcements S3-70 254 114.3 2.3, 0.54 25.4
(span = 2388 mm)
S4-70 254 114.3 3.45, 2.65 35.2
S4A-70 254 114.3 3.45, 2.65 20.5
HSC0 250 200 0.8 90.3
Hallgren and HSC2 250 200 0.8 85.7
Kinnunen
(span = 2540 mm) HSC4 250 200 1.2 91.6
HSC6 250 200 0.6 108.8 (c)

Table 3Comparison of predictions using

different effectiveness factors
Mean
Effectiveness (predicted/ Standard Coefficient of
factors experimental) deviation variation, %
Eq. (9a) and (10a) 0.92 0.146 15.87
Complex Eq. (9a) and (10b) 1.04 0.196 18.85
approach Eq. (9b) and (10a) 0.85 0.176 20.70
Eq. (9b) and (10b) 0.85 0.216 25.41
Eq. (9a) and (10a) 0.93 0.15 16.13
Simplified Eq. (9b) and (10a) 0.92 0.18 19.56
approach Eq. (9a) and (10b) 1.18 0.25 21.19
Eq. (9b) and (10b) 1.18 0.27 22.88
(d)
Fig. 9Correlation of predictions by simplified method with
This accounts for the presence of four plots in each of 97 test results using: (a) Eq. (9a) and (10a); (b) Eq. (9a) and
Fig. 8 and 9. (10b); (c) Eq. (9b) and (10a); and (d) Eq. (9b) and (10b).
A scrutiny of Table 1 to 3, along with Fig. 8 and 9, reveals
the following essential points: correlation (with a mean of 1.04) occurs for the complex
Both the complex and simplified theoretical analyses analysis using the Sigurdsson compression effectiveness
show good correlation with test data for all possible factor and a tension effectiveness factor of value 1/100;
combinations of effectiveness factors. The closest When the tension effectiveness factor is fixed for the

ACI Structural Journal/November-December 2002 833

 simplified analyses, the predictions are virtually
insensitive to variation in the compression effectiveness
factor. The reverse is true for the complex analysis
using the compression effectiveness factor of C =
4.22/ f c , where the analysis is insensitive to change in
the tension effectiveness factor;
The most conservative approach (namely that with a
correlation mean of 0.85) is the complex analysis with
the compression effectiveness factor of C = 4.22/ f c .
By contrast, the simplified approaches using a tension
effectiveness factor of 1/100 give a correlation mean
18% into the nonconservative regime;
When the Sigurdsson compression effectiveness factor
is used along with the tension effectiveness factor of
value 1/400, all the correlation data are very similar for
both the simplified and complex approaches. This also
gives an impressive mean correlation of predicted to
experimental data of 0.93;
The correlation is always very good for Moes slabs
with different reinforcement percentages along the two
perpendicular sides. Hence, for these slabs, the use of
an average reinforcement ratio in the Sigurdsson
expression appears not to have adversely affected the
quality of the predictions; and
The simplified analysis which gives a mean correlation
of 0.93 also gives the closest correlation for the very
high-strength concrete slabs tested by Hallgren. This is
apparent from Fig. 8 and 9, which show that the predicted
values for Hallgrens slabs are always higher than the test
values, and that, of all the analyses, the simplified
approach of Fig. 9(a) locates the Hallgren points closest
to the 45-degree line.
Hence, the complex approach, with the compression effec-
tiveness factor due to Sigurdsson and the tension effectiveness
factor of value 1/400, and the simplified approach, with the
Sigurdsson compression effectiveness factor and the tension
effectiveness factor of value 1/400, are both strongly recom-
mended. The simplified approach, in particular, is favored for
the following reasons:
It gives an excellent mean correlation of 0.93;
This mean correlation, being slightly less than 1, indicates
a slightly conservative approach;
For Hallgrens very high-strength concrete slabs, it gives
the best correlation (the closest location of the points to
the 45-degree line of the plot) of all the analyses; and
This analytical approach can be easily implemented in
a computer program.
The aforementioned combination of factors renders this
version of the simplified analysis particularly useful in a
design office environment. Note also that both the aforemen-
tioned recommended approaches use the Sigurdsson
compression effectiveness factor, which depends on the
tension steel reinforcement ratio, the depth of the slab,
and the compressive strength of the concrete.
CONCLUSIONS
By adopting a parabolic failure criterion for concrete, a
new upper-bound rigid-perfectly plastic method for predicting
the punching shear failure loads of concrete slabs has been
derived. Two alternative predictive approaches, namely one
complex approach based on a curved generatrix of the failure
surface, and the other a simplified approach based on a linear
generatrix of the failure surface, have been presented. In
addition, two alternative effectiveness factors have been
834
considered for the concrete both in compression and tension,
giving four different possible combinations of tension and
compression effectiveness factors for use in analysis.
These combinations have been used with each of the complex
and simple approaches.
The predictions were compared with data from 97 tests
incorporating slabs from 30 to 250 mm effective depths
and concrete cylinder compressive strengths from 14 to
109 N/mm 2. The approach giving the closest correlation
of predicted to experimental failure loads (mean 1.04) is the
complex method with a compression effectiveness factor
due to Sigurdsson and a tension effectiveness factor of value
1/100. Also highly recommended is the simplified approach
using the Sigurdsson compression effectiveness factor and
the tension effectiveness factor of value 1/400. That approach
gives a mean correlation of 0.93, and also gives the closest
agreement with the test data for the highest strength concrete
slabs considered. The mean of 0.93 signifies a slightly
conservative approach which, along with the ease of imple-
mentation of the analysis, renders this simplified approach
eminently suited to use in a design office environment. Finally,
note that the Sigurdsson effectiveness factor, which is common
to both the aforementioned recommended best analytical
approaches, is dependent on the tension steel reinforcement
ratio, the concrete compressive strength, and the depth of
the slab.
NOTATION
ck = parameter related to ratio of compressive to tensile strengths of
concrete
d = effective depth of slab
d0 = punch diameter
d1 = outer diameter of punched cone
ds = increment of length along generatrix of failure surface
D = diameter of support
DA = rate of dissipation of internal energy per unit area
fc = uniaxial cube strength of concrete
fc = cylinder strength of concrete, taken as 80% of concrete cube
strength fcu
fcu = effective uniaxial compressive strength of concrete
ft = effective uniaxial tensile strength of concrete
h = slab thickness
P = punching shear load
PC = predicted punching load based on curved generatrix
PE = experimental punching load
PL = predicted punching load based on linear generatrix
r = radial (horizontal) coordinate of generatrix of failure surface
u = relative displacement rate
WA = total rate of dissipation of internal energy
WE = rate of working of external load
x = vertical coordinate
= angle between normal to - failure surface and axis
= angle between tangent to generatrix of failure surface and
relative displacement rate vector
= angle of internal friction
c = ratio of effective compressive strength to actual compressive
strength of concrete
t = ratio of effective tensile strength to effective compressive
strength of concrete
= tension steel reinforcement ratio As /bd
= normal stress on arbitrary plane
1, 3= principal stresses
= shear stress on arbitrary plane
REFERENCES
Aoki, Y., and Seki, H., 1971, Shearing Strength and Cracking in Two-
Way Slabs Subjected to Concentrated Load, Cracking, Deflection, and
Ultimate Load of Concrete Slab Systems, SP-30, E. G. Nawy, ed., Ameri-
can Concrete Institute, Farmington Hills, Mich., pp. 103-126.
Base, G. D., 1959, Some Tests on the Punching Shear Strength of
Reinforced Concrete Slabs, Technical Report TRA/321, Cement and
Concrete Association, London, July.
ACI Structural Journal/November-December 2002
 Bortolotti, L., 1990, Punching Shear Strength in Concrete Slabs, ACI
Structural Journal, V. 87, No. 2, Mar.-Apr., pp. 208-219.
Braestrup, M. W.; Nielsen, M. P.; Jensen, B. C.; and Bach, F., 1976,
Axisymmetric Punching and Reinforced Concrete, Report No. R-75,
Structural Research Laboratory, Technical University of Denmark, 33 pp.
Criswell, M. E., 1974, Static and Dynamic Response of Reinforced
Concrete Slab-Column Connections, Shear in Reinforced Concrete, SP-42,
V. 2, American Concrete Institute, Farmington Hills, Mich., pp. 721-746.
Dragosavic, M., and van den Beukel, A., 1974, Punching Shear,
Heron, V. 20, No. 2, Delft, 48 pp.
Elstner, R. C., and Hognestad, E., 1956, Shearing Strength of Rein-
forced Concrete Slabs, ACI J OURNAL, Proceedings V. 53, No. 2, July,
pp. 29-58.
Hallgren, M., and Kinnunen, S., 1996, Increase of Punching Shear
Capacity by Using High-Strength Concrete, 4th International Symposium
on Utilization of High-Strength/High-Performance Concrete, Paris,
pp. 1015-1026.
Jiang, D. H., and Shen, J. H., 1986, Strength of Concrete Slabs in
Punching Shear, Journal of Structural Engineering, ASCE, V. 112, No. 2,
Dec., pp. 2578-2591.
Jiang, D. H., and Shen, J. H., 1988, Strength of Concrete Slabs in
Punching Shear (discussion), Journal of Structural Engineering, ASCE,
V. 114, No. 9, Sept., pp. 2160-2167.
ACI Structural Journal/November-December 2002
Kinnunen, S., and Nylander, H., 1960, Punching of Concrete Slabs
without Shear Reinforcement, Transactions of the Royal Institute of
Technology, Stockholm, Sweden, No. 158, Mar., 112 pp.
Leon, A., 1935, Ueber die scherfestigkeit des betons, Beton und Eisen,
V. 34, No. 8, Apr., pp. 130-135.
Moe, J., 1961, Shearing Strength of Reinforced Concrete Slabs and
Footings under Concentrated Loads, Development Department Bulletin
D47, Portland Cement Association, Skokie, Ill., Apr., 130 pp.
Morley, C. T., 1979, Punching Shear Failure of Hollow Concrete
Spheres, Plasticity in Reinforced Concrete, Final Report, Colloquium of
International Association for Bridge and Structural Engineering, Copenhagen,
Aug., pp. 167-174.
Nielsen, M. P., 2000, Limit Analysis and Concrete Plasticity, Prentice-
Hall, New York.
Rankin, G. I. B., and Long, A. E., 1987, Predicting the Punching
Strength of Conventional Slab-Column Specimens, Proceedings of the
Institution of Civil Engineers, Part 1, V. 82, Apr., pp. 327-346.
Regan, P. E., 1978, Design of Reinforced Concrete Flat Slabs, Technical
Note to Project Record 220, Construction Industry Research and Information
Association, London.
Sigurdsson, T. G., 1991, Punching Shear by Eccentric Loading, MSc
thesis, Department of Structural Engineering, Technical University of
Denmark, July.
835