You are on page 1of 14

Symmetric punching of reinforced

concrete slabs
P. E. Regan* BSC,PhD, DIC

THEPOLYTECHNIC OF CENTRALLONDON

S Y N OP S I S
mI boundary moment per unit width of slab
The paper reports punching tests on 28 reinforced P concentrated load or column reaction
concrete slabs. The tests were divided into five series PRex value of P corresponding to flexural capacity
primarily concerned with the efSects of thearrangement of slab calculatedby yield-line theory
offlexural reinforcement, absolute size or depth, con- p k characteristic
punching resistances, calcu-
crete strength andratio ofreinforcement, boundary lated from Codes of Practice
restraint and the size of the loaded area. The results are P, experimental ultimate value of P
compared with the predictions of four Codes of Prac- U length of control perimeter used in punching
tice-BS 8110, C P 110, A C I 318-83 and the CEB-FIP calculations (see Figure 1)
Model Code. In regard to B S 8110 it is demonstrated length of periphery of loaded area
that the Codegenerally gives a small scatter of the ratio basic characteristic shear resistance (stress)
of calculated and experimental strengths, although its partial safety factor for loads
method of defining the ratio offlexural reinforcement is partial safety factor for resistances or mater-
inferior to that of C P 110. It is also shown that BS 8110 ials
does not achieve the level of safety intended in the 5s size or depth effect factor
definition of characteristic strength. If the definition of P ratio of flexural reinforcement (average for
the ratio of reinforcement were improved, the intended orthogonal directions)
level of safety could be attained by a 10% reduction in
either the basic shear stress or the size factor CS. Reinforcement
R plain round mild steel to BS 4449
T rounddeformed(type 2) highyield bars to BS
Notation 4449
B diameterofacircularcolumn or loaded area
b sidedimension of asquarecolumn or loaded Introduction
area
d mean effective depth slab of(average of Punching is a subject on which there is no consensus
effective depths in orthogonal directions) onatheoretical level andtherearerather wide
divergenciesbetween different empirical treatments.
.LC cylindercrushing
strength
of
concrete There is thus a need for test data which may at least
fCu cubecrushingstrength
of
concrete
allow some rational assessment to be made of current
fY yield or 0.2% proof stress of reinforcement design recommendations. Such an assessment seems
h overall
thickness of slab
timely as BS81 lo(') hasintroducedanumberof
l span of slab
changes from the provisions of CP 1
* Reader in Civil Engineering, The Polytechnic of Central London, The 28 tests reported here are divided into five
35 Marylebone Road,
London, NW l 5LS. groupseach
concerned with one or two of the para-
115

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
*-
Magazine of Concrete Research: Vol. 38, No. 136: September 1986

meters thought to influence punching resistance. The


programme was not planned as a single entity but the -----l

totality of the results does provide a reasonable basis I I r---l


on which to evaluate design guidance, especially as it
can where necessary be supplemented by reference to
published data.
All the tests were of slabs supported at four
sides and
subjected toconcentratedloads at theircentres. In
most cases the loads were applied upward through
steel plates and the reactions were provided by box-
section steel beams placed on the top surfaces of the
slabs and anchored to the laboratory floor by high-
tensile bars. With the exception of the fourth group the
reaction beams were close to the slab edges and the
support condition was thus oneof simple support with
the corners of the specimens free to deflect. The more
complexarrangementsfor group IV are described
below.
The slabs of group I had monolithic column stubs
and were loaded throughthese and notby plates on the
surfaces of the slabs. Thedifference is believed to have
had no effect on punching resistance.
The main interest in this work is in the influences of
the parameters varied on the ultimate punching loads
and the relationships of the experimental strengths to
values calculated by code equations. The results are
Figure I : Control perimeters usedby Codes.
presented primarily in these terms buta limited
amount of data ondeformations is included as it has a
relevance to more theoretical approaches to the sub- (4) BS 8110 introducesa specific upper limit to
ject. resistance expressed in terms of thenominal
shear stress at the periphery of the loaded area
Codes of Practice which is related to the strength of the concrete.
Four Codes of Practice are considered in compari- The control perimeters of the ACI and CEB Codes
sons with the test data-BS 81 lo('), C P 1 IO(*), ACI 3 18- are much closer to the loaded area (0.5d from it) and
83(') and the CEB-FIP Model Code(4). the limiting nominal shear stresses for relatively small
Both British Codes work in terms of control peri- loaded areas, including all those in the present tests,
meters at relatively large distances (see Figure l ) from are higher than the values for one-way slabs. The main
the loaded areaand use the same limiting shear stresses differences between these two Codes are:
as forone-way spanning slabs. The differences between (1) In ACI 31 8 the control perimeter is taken to
the basic limit stresses in the two Codes aresmall and have the same shape as the loaded area, but in
are ignored here with the BS 81 10 equation being used CEB 78 the perimeter always hasrounded
forboth calculations.
The significant differences corners, see Figure 1.
between the two documents are: (2) In ACI 318 the limiting shear stress depends
only on the concrete strengthwhile in CEB 78 it
In BS 81 10the ratio of flexural reinforcement is
is alsoafunction of theratio of flexural
calculated for awidth of slab equal tothat of the
reinforcement and of the slab depth.
loaded areaplus 1.5d to either side of it, while in
C P 110 the width is that of the loaded area plus ' For the purposes of making comparisons with test
3 h each side. results, safety factors have been removed fromthe
In BS 81 10 the control perimeter at a distance Code equations to give expressions for characteristic
1.5d from the load has square corners whether resistances, i.e., values which should be attained by
the loaded area is square orcircular, whereas the 95% of test results. Limits on the ranges of parameters
CP 1 10 perimeter, 1 . 9 2 fromtheload,has covered by the Code formulaehave been ignored, but
rounded corners in all cases. this has little effect on the comparisons.
The rangeof slab depthsover which a size effect The uniformuse of characteristic strengths doesnot
is taken toaffect punching resistance in BS 81 10 in itself ensure a fair comparison between levels of
is 100 m m 5 d s 4 0 0 mm whilein CP 110 the safety actually achieved by differentCodes as there can
corresponding limits are 150 mm I h I 300 mm. be differences between their partial safety factors for
116

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Symmetric punching of reinforced concrete slabs

resistances (y,) and loads (yf) and between specified with p calculated fora width equalto (B+5d) or
(characteristic)loads themselves. Consideringdead +
(b 5 4 , and d in mm.
loads toavoid the last problem, the global factors
ym yf
for the Codes in question are:
BS 81 10and CP110 1.25 x 1.4= 1.75 Group I, arrangement of reinforcement
ACI 318 1.18 X 1.4= 1.65
Seven slabs were included in this group. They were
CEB-FIP 1.5 X 1.35=2.02
2.0 m square and 100 mm thick. They were tested
Thus if the characteristicresistances given by all four simply supported at four sides with spans 1.83 m and
recommendations were in thesamerelationship to with their corners free to deflect. Loads were applied
experimental strengths, theCEB-FIP documentwould through 200 mm monolithic column stubs which
give a higher and the ACI Codea lower overall factor projected aboveand below the slabs. Only tension
of safety than that of the British Codes. reinforcement was provided and this was of deformed
The Code equations used in comparisons .arelisted high yields bars to BS 4449. The bars had straight ends.
below. The comparisons with test results also include The principal variable was the arrangement of the
flexural strengths calculated by yield-line theory, with reinforcement. The first six specimens constituted
the stressblock in the concrete compression zone taken three pairs. In each pair one slab had uniformly spaced
as 0.6f,,. For codes expressed in terms of cylinder cfcc) reinforcement while the other hadthe same number of
rather thancube Cf,,)crushing strengthsof concrete it bars arranged approximately in accordance with the
has been assumed that fCc = 0.8 L,,. elastic momentdistribution.Thetotalarea of re-
BS 8110 (with the partial safety factor for materials inforcement in the outer layer was slightly less than
y, = 1.25 removed and limits o n h uneglected) that in the inner one so that the flexural capacities in
orthogonal directionswere practically equal. The pairs
differed in totalamounts of reinforcement. The
seventh slab was broadly similar to the sixth but had
U = 4 (B + 3d) for circular loaded areas some bottom steel passing through the column area
U = 4 (b + 3d) for squareloaded areas and a slightly different arrangement of tension steel.
p, = Uck ud S 1. 2 z u , d Details of the reinforcement are shown in Figure 2.
All the tests ended in punching, although the failure
with p calculated for a width equal to (B + 3 4 or
(b + 3 4 and d in mm U, = length of periphery of of slab 7 could well be said to have been primarily
loaded area = nl? or 46. flexural. Its ultimate load was 9% above the yield line
In the upperlimit of P,, ym has been assumed to be capacity. The test data are summarized in Table 1.
1-5. Figure 3shows the results as graphsof ultimate loads,
corrected forvariations of concrete strength and
CP 110 (with partial safety factorformaterials effective depth plotted against ratios of flexural re-
removed and limits on f,,neglected) inforcement determined for the widths specified in BS
81 10, CP 110 and the CEB-FIP Code. The strengths
uck = 0.27 3dmi are alsoplottedagainstthe reinforcement ratios
5, = 1.6-0.002 h 2 1.0 averaged over the full widths of the slabs. It is apparent
U = n (B + 3h) for circular loaded areas that concentrationof reinforcement toward the loaded
U = 4b + 37th for squared loaded areas
area has no significant beneficialeffect in terms of
k‘ = 5 s uck ud punching resistance. The scatter of results is greatest
with p calculated for a width equal to (B 6h) or + for BS 81 I O where the steel ratio is calculated for the
(b + 6h) and d in mm(Note:Priortothe 1974 narrowest width and least when the ratio is averaged
amendment 5, = 1.5- 0.002h 2 1.O) over the full width.
This conclusion is fully supported by the results of
ACZ318-83 (with capacity reduction factor 4 omitted) tests by Elstner and Hognestad(6),and M o ~ ( ~The ).
uck = 0.332 & available tests have all been of specimens representing
U = n (B + d ) for circular loaded areas the intersections of column strips in flat slabs. It is
U = 4(B + d ) for square loaded areas reasonable to assume that reinforcement very far from
P, = uCk u d thecolumncan have little influence on punching
resistance and it would probably be appropriate to
CEB-FZP (with partial safety factor ym = 1.5 removed) calculate steel ratios for column strips and not full
uck = 0.084 (1 + 50p)fc,2’3 column-to-column widths.
r, = 1-6-0.001 d The presentdata is not very clear as to the influence
+
u = n (B d) forcircular loaded areas of the overall steel ratiobut published
= 4b + n d for square loaded areas confirm that either the’British oc 3J100p or the CEB
= uck u d CC(1 + 5 0 p ) is satisfactory.
117

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research: Vol. 38, No. 136: September 1986

TAB LE 1 : Group I test data.

3x140 ThiP.
BS CP ACI CEB P,
8110 110 318 FIP P ~ e x

2x70
1.17 1.04 0.74 0.74 0.91
40
1.01 0.98 0.78 0.77 0.76

1.08 0.97 0.76 0.84 0.90

0.93 0.90 0.83 0.78 1.02

1.25 1.10 0.94 0.96 0.92

0.93 0.89 0.83 0.69 1.06

Y direction bars outermost with


13 mm cover
0.94 0.90 0.87 0.77 1.09
W T10
bars
all in in111T8
and 115
85: 1 Slabs U 1 and //5(//Z and 1/6 similar Notes and general data
but with reinforcement uniformly
I spaced] (1) Values o f p given in column 2 are averages for the full widths of the
slabs. Local ratios for the widths specified in Codes can be seen in
Figure 3.
(2) Reinforcement of slabs 1 4 was of TI0 bars with d=77 mm and
2x180 &=SO0 N/mm2. Reinforcement of slabs 5-7 was of T8 bars with
d= 79mm and f,= 480 N/mm2.

The concretemix for the largest slab with a thickness


100 of 250 mm had a maximum aggregate size of 20 mm.
2x50 The same mix was used in one slab to each of the
smaller scales. In addition smaller slabs were tested of
concrete mixes with reduced aggregate sizes.
All the slabs failed by punching and the ultimate
loads are given in Table 2 together with the details of
the specimens and the results of comparisons with the
four Codes.
Figure 5 shows the test results as a graph of

1- i? j>Eb[o!{Jj-
N - m -

8 1
Y direction bars outermost
with 13 mm cover
all bars T10
Slab 113 1//4 similar but with
reinforcement uniformly spaced]
(2'g) plotted against d.

The correction for variations in cube strength has


Figure 2. Group I. Reinforcement details. very little influence here. If there were no size effectthe
pure linear scaling of the specimens should !ead to
Although concentration of reinforcement does not (Pu/d2)being independent of d, whereas Figure 5
increasepunching resistance in specimensof the shows it rising .significantly as d decreases. The varia-
present typeit does reduce deformations as can be seen tion is gr.eatest amongst thespecimens with a constant
from Figure 4. maximum size of aggregate and is reduced but not
eliminated where the aggregatewas scaled approxima-
tely with the slab dimensions.
Group 11, size or depth The figure alsoshows the BS 81 10 size factor
In the second group of six specimens the principal (tee 4JYd) and this agrees reasonably well with the test
variable was the size or scale of the slabs. Three sizes results and lies between the datafor scaled and
weretested. The overall dimensions,the effective unscaled aggregate.
depth, the size of the circular loadingdisc, the bar size The range of the slab depthscovered here is limited
and the bar spacing were scaled linearly in the ratios but tests at KTH Stockholm@)of slabs with effective
1 :0.64: 0.32. The spans between simple supports at depths up to619 mm also correlate satisfactorily with
four sides (corners free) were in practically the same the fourth root relationship.
ratios. Table 2 shows that in absolute termsBS 8 1 10 agrees
118

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Symmetric punching of reinforced concrete slabs

/
/
/
/
/’
/
k/ 0 3 O1

0
.
- */

”m
/
/’ 5 0
/
/
/

-
CP 110 b+6h

I I
1 2 1 2
REINFORCEMENT - (%) REINFORCEMENT- (%l

20

z
Y

average (whole slab)

I l
1 2 1 2
REINFORCEMENT - (%) REINFORCEMENT - (%)

Figure 3. Group I. Relationship ofpunching strength to ratios of reinforcement, determined by dzfferent Codes.

119

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research: Vol. 38, No. 136: September 1986

200 - L
1 3
S
4
,-57
/ / +2 3.', 5 7 -/ \
/'
32.
. \
/
n
't
2
a 100
50

7 1 5 0 / f j I I 1 I Maximum size
0 5 10 15 20 25 Symbol of aggregate (mm)
DEFLECTION - [mm)
0 20
Figure 4 . Group I . Load-deflection relationships.
0 10

X 5
well with the test results, while CP 110 isa little on the
safe side, with the difference being due to the different
shapes of controlperimeter for circular loaded areas.
ACI 318-83 and the CEB-FIP Code are both over-
conservative, while theACICodes neglect of size I 1
100 200
effects makes its predictions rather inconsistent for this EFFECTIVE DEPTH - (mm)
group oftests. Figure 5 . Group II, Influence of effective depth on unit resistance.

Group 111, concrete strength and reinforcement through theslabs and reacted against their top surfaces
ratio via 130 mm square plates.
This was a set of sixcircular slabs with thicknesses of The slabswere reinforced at their tension faces only.
120 mm and diameters of 1.5 m. The specimens were The bars were all provided with end hooks bent into
loaded centrally through150 mm diameter steel plates the bottoms of the slabs.In three of the slabs the ratio
and simply supported by eight tie bars arranged in a of reinforcement was 0.83% while for the other three it
circle of diameter equalto 1.37 m. The tie bars passed was 1.52%. The concrete cube strengthswere intended

TABLE 2: Group I1 test data.

Pdpu

CP CEB ACI
FIP
318 110

0.79 0.64 0.67

II/2 41.6 20 160 0.82 0.57 0.57

II/3 42.9 10 160 0.89 0.62 0.62

II/4 41.6 80 20 0.75 0.47 0.50 0.91

IIj5 42.9 80 10 0.85 0.53 0.56 0,82


-
II/6 45.2 80 5 0.86 0.55 0.58 0.82

Notes and general data


(1) All dimensions scaled practically linearly, thus with dimensions in mm: for
h = 250; d = 200, B = 250, l= 2745, reinforcement T25 at ,250; for .h = 160,
d = 128, B = 160,1= 1800, reinforcement TI6 at160; for h=80, d=64,B=80,
1=900, reinforcement T8 at 80.
(2) p=0.98%.
(3) fy=530 N/mm2 (II/1),fv=485 N/mm2 (II/2 and 3);fY=480 N/mm2 (11/46).

120

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Symmetric punching of reinforced concrete slabs

to be 15,30 and45 N/mm2 with one slab of each steei


ratio castfrom each type of concrete. In theevent there
were some variationsfrom the intended cube strengths
but the intention of having a range of concretes for
each level of reinforcement was realised.
All the slabs failed by punching without developing
very large flexural deformations (see Figure 6). Slab 3
was closest to flexural failure (Pu/Pfiex=0.88) and as
can be seen from Figure 7 an appreciable proportion,
but by no means all the reinforcement had yielded by
the ultimate load. Figure 7 also shows that the slab's
behaviour was in effect polar-symmetric in spite of the
reinforcement being orthogonal. The only significant
differences between steel strains on different radii
occurred close to the outerperimeter where conditions
were disturbed by the reactions being at eight points
rather than all around the edge.
Figure 8 shows the experimental strengths plotted
5 10
against the cube strengths of the concretes and com-
DEFLECTION - (mm)
pared with characteristic strengths calculated accord-
Figure 6 . Group III. Load-deflection relationships.
ing to CPl 10. The CP110 assumption that punching

x
l
214kN \
\
\ - B

\ \

F\ -
188kN \
\

strains on line A [averages


ll from gauges on pairs of bars
at right angles.
__ - strains on line B

154kN \ '

I I 1 I I 1 I
100 200 300 400 500 600 . 700
RADIUS - (mm1
Figure 7. Group III, Radial variations of reinforcement slrains.
121

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research: Vol. 38, No. 136: September 1986

If the limit of 40N/mm2ofthe maximum useful cube


strength, which has been ignored, were applied in
calculations by CPI 10 and BS8110 Some predictions
would of course be more conservative. However the
scatter of the ratios of calculated and experimental
strengths would be increased.
The use of the square root of concrete strength as
given in the ACI Code, also appears to over-estimate
the influence of this parameter. The ACI neglect of
reinforcement as an influential parameter produces a
difference in the ratios of calculated and experimental
strengths for the two sub-sets of slabs with different

8o t 0
p = 1.52% steel ratios.

40 1 I
10
I
20 30
I
40
I
CUBE STRENGTH f,, - (Nlrnrn')
50
I I
60
Group IV, boundary restraint
The specimens for this group of tests were more
complex than the others. As shown in Figure 9 the
overall plan form was a cross with a central 1.73 m
square panel and projections of 635 mm at each side.
Figure 8 . Group III, Comparison of ultimate loads with
characteristic strengths according to CP 110. Each projection was divided by a central slot. The
specimens were 100 mmthick and their central panels
were reinforced with TI0 bars at 75 mm centres both
resistance is proportionaltothecuberoot of the
ways in the top and T8bars at 75 mm both ways in the
concrete compressive strengthcan be seen to be
bottom. Full details of the reinforcement are drawnin
reasonable. The actual levels of strength given by CP
Figure 9.
110 are alsosatisfactory over arange of concrete
An upward loadwas applied at the centre through a
strengths extending to bothsides of that defined in the
160 mm square plate and downward line loads were
code.
applied at thefour sides of a 1.83 msquare.The
BS 81 10 also takes punching strength tobe propor-
assembly was supported on rollers positioned 0.457 m
tional to the cube root of the concrete strength but in
beyond the downward loads. By varying the ratios of
absolute terms its predictions are too high for this
the upward and downward loads differing reactions
group of tests. The ACI andCEB Codes are bothsafe
for all concrete strengths but the CEB assumpion of could be produced at the roller supports, thus chang-
punching strength being proportional to compressive ingtheratio between thecentralload (P)and the
strength to the power of 2/3 gives very systemmatic restraining moments (m') at the edges of the 1.83 m
errors. square defined by the downward loads. This ratiowas
the only intentional variable in the group. With slighta
exception at low loads in test 2, the ratio was kept
T A B L E 3 : Group I11 test data.
constant in any one test but varied fromoneto

Slab fCu
7 I
ACI CEB P,
another.
All the slabsfailed by punching and thetest data are
summarised in Table 4. The ultimate loads, corrected
No. (N/mm2)l 318 FIP
fortheminorvariations in concretestrength are
Paex
plotted against ( m ' / P )in Figure 10. The strengthrises
III/l 29.0 0.59 0.84 0.54 with increasing restraining moments, but it is uncer-
tain whether the cause is really the moment or the
III/2 11.9 0.61 0.66 0.48 lateral restraint provided by the uncracked concreteat
1.01
the edges of the 1.73 m square. Figure 11 shows the
III/3 47.3 0.70 0.69 0.88
final crack patterns of slabs l and 3 and it canbe seen
III/4 14.9 1.14 0.53 0.68 0.53 that while the slots in the projections were successful in
avoidingthe
formationof
complete
a ring of
IIIj5 33.5 0.57 0.66 0.59 uncracked concrete, the application of restraining
0.62 0.78 0.64
moments in test 3 prevented the radialcracking
III/6 53.3
extending to the cornersof the central panel.
The increase of strength from thesimply supported
General data
(1) Reinforcement-
specimen tothe one with the highest restraining
for p=0.83% TI0 c/c 100 f y = 4 9 4 N/mm2, d = 9 5 mm; moment was approximately 25%. This almost corre-
for p = 1.52% T I 2 c/c 80 f,=464 N/mm2, d = 9 3 mm. sponds to the increase to be expected according to
122

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Symmetric punching of reinforced concrete slabs

l TABLE 4: Group IV test data.

Notes and general data


(1) Top reinforcement TI0 cjc 75, p = 1.31%,
fy = 525 N/mm2, d=80 mm.
(2) Bottom reinforcement T8 c/c 75, p' =0.64,
fy=510 N/mrn*, d'=18 mm.
(3) Ratios of experimental strengths to yield-line capacities are not
given as the boundary moments applied were well below yield
downward downward
moments.
updard
load load load

eventual failure surfaces. Some results are plotted in


Figure 12. It can be seen that toward failure there were
small discontinuities in the deflected profiles in the
areas of the failure surfaces. These no doubt corres-
ponded to deformations at shear cracks. However it
should be observed that the differences between the
inner and outer rotations were verymuch less than the

steel
top '1 bottom steel

Figure 9. Group I V , Test specimens and loading systems.

British Codes if theratio of reinforcement was


increasedfromtheactual 1.31% tothemaximum
effective value of 3.0%.The results thus seem to be in
accord with work by Taylor and Hayes(9) which
showed that the punching strengths of slabs with rigid
lateralrestraint,butnoboundarymoments, were
equal to those of simply supported slabs with about
3% of reinforcement. This strengthwas realised by the
restrained slabs irrespective of their reinforcement.
SincenoneoftheCodes considered takesany m'
account of lateral restraint or of boundary moments,
all of them become increasingly conservative in this
series as the ratio (rn'/P) goes up.With m'=O the
British Codes give reasonable results, the ACI Codeis
rather conservative and the CEB-FIP Code is also I I I l
0.01 0.02 0.03 0.04 0 5
conservative. m'lP
In these tests rotation of the top surfaces of the slabs
were measured a t locations inside andoutsidethe
123

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research: Vol. 38, No. 136: September 1986

Group V, size of loaded area


The slabs with a thicknessof 150 mm and plan
dimensions of 1.6 x 1.6 m Were simply supported atall
four edges on spans of 1.5 m. They were reinforced
with 13 Swedish Kam steel 12 mm bars at 120 mm
centres both ways. The clear cover to the outer steel
was 20 mm and all bars were hooked. The slabs were
without compression reinforcement.
The only significant variable between tests was the
detail of the loaded area. In tests 1 to 4 the size of steel
loadingplate was varied while in- test 5 load was
applied through a precast concrete cylinder cast into
the slab as shown in Figure 14. The purposeof the fifth
test was to simulate conditions arising where the heads
of piles penetrate into a suspended ground slab.
All the tests ended in punching failuresand the data
together with comparisons with code predictions are
summarised in Table 5. Figure 15 shows profiles of the
failure surface of slab 1 which was loaded over the
smallest area. In slab 5 where load was applied through
a precast cylinder there was a slip between the precast
and in-situ concrete at failure, but damage in the in-
situmaterialindicated that therehad been aload
transfer between the two.
Figure 16 shows the punching loads, corrected for
the small variations of cube strengths, plotted against
the dimensionsof the loaded areas and compared with
the predictions of BS 8110. It can be seen that the
general prediction 'Pk= ud' is satisfactory so long
as the load dimension exceeds about 0.75d. For smaller
loaded areas the upper limit Pk I 1.2 a u , d seems to
be necessary and gives a satisfactory result. The few
other resultsavailable(9-")from tests with small loaded
areas show a considerable scatter with respect to the
BS 81 10 upper limit of Pk.However the expression

m
TABLE 5 : Group 111test data.

Figure I I . Group W . Final crack patterns of slabs IV/I and IV/3.


rotations themselves. Thus if a simple mechanical
model of deformation is required it seems appropriate
toconsideraslabto be divided into rigid radial
segments rotating aboutaxes at the columnface, with
I

I
Lmt;d

diameter 54
I
I
P, 1
I

(kN) 8110 110 318

170 1.35 1.31 0.73


l

0.40
only negligible discontinuities at shear cracks.
diameter 170 280 1.03 0.96 0.72 0.59
Figure 13 showstangentialstrains on thecom-
pressed surfaces of the slabs both as radial distribu- diameter 110 265 0.97 0.92 0.60 0.59
tions at the maximum loads at which measurements -
were made and as load/strain curves for the locations 45.3 102 x 102 285 0.92 0.92 0.73 0.62
of maximum strain. It is clear from the distributions 41.1 diameter 150* 285 0.98 0.92 0.66 0.61
thatstrains were higher ontheradii through the -
corners of theloadsthanonthe orthogonal axes
Notes and general data
indicating a concentration of stresses at corners. From * Load arrangement penetrated sla&see Figure 14.
the load/strains graphit seems that the peak strain at (1) Mean effective depth 118 mm.
failure decreases somewhat with increasing boundary (2) 11 A,/bd= 0.80.
restraint but is of the order of 2 to 3%. (3) y/ = 628 N/mm2/. = 752 N/mm2.

124

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Symmetric punching of reinforced concrete slabs

Slab IV14
-
Slab IV/3
240 -

200 -

0 4 B 12 16 20 24
ROTATIONS (%l
Figure 12. Group IV. Rotations inside and outside shear cracks.

TABLE 6 .

I I

Code I Number of
results I 1
Mean
Coefficient of variation
(%) I
Mean ( I + 1 . 6 4 CoV)

BS 8110* 21 0.98 7.3 1.10

CP 110 24 0.92 9.1 1.06

ACI-3 18 24 0.67 17.9 0.87

CEB-FIP 24 0.67 18.8 0.88

* Excluding the results for slabs 1/1,3 and 5 .

125

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research: Vol. 38, NO. 136: September I986
3, / l precast cyclinder 6in.diameter 6in. high

I
with projecting remforcement

l j0 /

Note: profiles drawn


at 0.89P" for slab 1
and at "'6 for slabs 2-4

Locations 1-5
-Locations &l0

-4
l 252rnm Iona

I l
100
l
200
l
300
l
400
I
500
l
600

stirrups 8 0 x 8 0 internal

Figure 14. Group V , Detail of slab VIS.

rose with increasing deformation but the improvement


cannot be quantified from these testsasthe test
arrangements tended to become unstable.

Conclusions
The main conclusions that can be drawn from this
work relate to theappropriateness of thevarious
Codesconsidered. A few results are probably best
excluded from anoverall comparison-these are those
from slab V/l where the loaded areawas so small that
the BS 81 10 main equation is inapplicable and slabs
IV/2 to 4 where theboundarymomentsintroduce
factors not treatedin any of the Codesreviewed. With
STRAIN AT LOCATION (6)-(%)
these exclusions there remain 24 test results and the
Figure 13. Group I V , Concrete strains on the compressed surfaces
of the slabs.

does seem to be safe while the use of the basic equation


is unsafe for loads on very small areas.

Behaviour after punching


Behaviour afterpunching was monitored in some of
the tests. When punching occurredtheload fell
suddenly and was releasedcompletely in most cases. It
was then re-applied and it was found that in the slabs
without bottom reinforcement an almost flat plateau
of resistance was reached at which the failure pro-
gressed with thetopbars being pulled out of the
concrete in the outer partsof the slabs.Large deforma-
tions were developed at almost constant loads often as
L7 L7
low as 25 to30% of the peak resistance. Post-punching
performance was better when the slabs had bottom I I
steel passing through the loaded area.Resistance then Figure 15. Group V , Profiles of failure surface of slab V j l .

126

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Symmetric punching of reinforced concrete slabs
2

A CP 110 original1972
B CP 110 amended 1974
' 0 o/ C BS 8110
c\

/ $ 1
/ ' l
/ Pk <
1 . 2 m
(drawn for circular
/--- loaded areas).
01
l

l
/i 100
I
125
200
I
250
d (mm)
300
I
375
400
I
500
I
500 600
f load dimension h (mm)
0 - diameter of circle or
0 - side dimension of square SLAB DEPTH
Figure 17. Evolution of the size or depih fucror t8in Brifish Codes.

go further butit would have been more appropriate to


have left the value corresponding to h=250 mm or
I 1 I I
15050 100 200 d= 200 mm unchanged and to have made less than
LOAD DIMENSION - ( m m ) unity for thicker slabs. Alternatively Uck could have
Figure 16. Group V , Influence of loud dimensions on punching been reduced.
resisiunces. Table 6 showssimilar values for the ACI and CEB-
FIP Codes. The mean values of (&/P,) are 0.67 and
overall performances of the Codes in relation to these the coefficients of variation are about 18%. In both
are summarised in Table 6. cases the coefficient of variation is greater than for
The requirement for Code's
a predicted resistance to British Codes due to the use of a controlperimeter too
be a true characteristic value is approximately close to the loaded area. For the ACI Code it is also
increased by the neglect of thesize effect, whilefor the
CEBModel Codethe usein the expression for
[l + 1.64C0Vl = 1.0 U& contributes to the scatter.
p, mean
In practice applied to slabs which would typically be
where c o v is the co-efficient of variation of (&/P,). somewhat thicker than most of the test specimens the
Values greater than unity are on the unsafe side. level of safety attained by the ACI Code, would be
It is immediately apparent that BS 8 110 does not about what is intended, although the mean of (P,/P,)
achieve the intended level of safety. Its performance is would remain low. The CEB Model Code is simply
improved if the data for the slabswith variable spacing rather too conservative.
of bars are omitted but the last column of Table 6 If the overall safety situation is considered instead of
remains 10% higher than it should be although the the prediction of characteristic strengths, t h e differ-
coefficient of variation of ( P k / P , ) is remarkably low. ences between the codesare very large, At one extreme
CP 110 in its amended (1974) form is also on the the CEB-FIP Model Code actually provides a global
unsafe side though safer than BS 81 10.In the original factor y m y p 2.3 for dead loads,while the correspond-
1972 version C P 110 was a little more conservative and ing figure for BS 81 10is only about 1.5.
in relation to these tests would have given the intended In relation to morespecific points the tests indicate:
level of safety. (a) BS 81 10 is wrong to define the ratio of flexural
The evolutionof the depth factorC;,in British Codes reinforcement in relation to a width only equal to
since 1972 is shown in Figure 17. At each revision the that of the loaded areaplus 1.5dto either side of it.
range of depths affected by thefactorhas been "Width of loaded area plus 3h to either side" as
extended, and theway in which this has been done has given in CP 1 10 is preferable.
always been such asto increase tSthroughoutthe (b) In principle BS 81 10 is right to take account of a
range, while thebasic U& has remained practically size or depth effect over a relatively large range of
unchanged. The result has been a continuous rise in slab depths and the failure of the ACI Code to
permissible stresses or viewed in another way a include this effect leads to errors.
continuous reduction ofreal safety factors. Theexten- (c) In the British Codes, use of the cube root of the
sion of the range ts ofis welljustified and could indeed concrete's compressive strength in expressions for
127

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.
Magazine of Concrete Research: Vol. 38, No. 136: SeF Itember 1986

vckis appropriate while the CEB use of compres- REFERENCES


sive strength to thepower of the two thirdscauses 1. BRITISHSTANDARDS INSTITUTION. The structural use of concrete.
significant errors. London BS 8110: 1985.
2. BRITISH STANDARDS INSTITUTION. Codeofpractice for the
(d) The punching resistances of slabs can be appreci-
structural use of concrete. London. CP 110 Part 1: 1972.
able enhanced when the boundary conditions are 3. AMERICAN CONCRETE INSTITUTE. Building code requirements for
other than those of simple support. However the reinforced concrete. Detroit. ACI 318-83 1983.
factors involved are not fully understood and the 4. CEB-FIP. Modelcode for concretestructures, Comitk Euro-
effects cannot be allowed for in Codes of Practice International du BCton. Paris 1978.
at least for the present. 5. BASE, G . D. Sometests on the punchingshearstrengthof
reinforcedconcreteslabs. London,Cement and Concrete
(e) Where the loaded area is very small (side dimen- Association. July 1959. Techical Report TRA 321.
sion less than about 0.754, the BS 81 10 general 6. ELSTNER, R c. . and HOGNESTAD, E. Shearingstrength of
expression for punching resistance can give very reinforced concrete slabs. Journal of the American Concrete
significant overestimates of strength. The situation Institute, Proceedings Vol. 28, No. 1. July 1956. pp. 29-57.
seems to be covered safely by the Code's upper 7. MOE, J. Shearing strength of reinforced concrete slabs and
footings under concentrated loads, Development Department
limit on punching strength. Bulletin D47, Portland Cement Association, Skokie, Illinois
April 1961.
8. KINNUNEN, S., NYLANDER, H. and TOLF, P. Undersokningar
rorande genomstansning vid institutionen for bygnnadsstatik
KTH (Investigations of punching at the building statics
ACKNO'WLEDGEMENTS institute KTH), Nordisk Betong 3-1978, pp 25-27.
The slabsof series I and 11 were tested in the course of 9. TAYLOR, R. and HAYES, B. Some tests onthe effect of edge
aresearchprogrammefunded by theConstruction restraint on punching shear in reinforced concrete slabs,
IndustryResearch and InformationAssociation and Magazine of Concrete Research, Vol 17N o 50, March 1965, pp
3 M .
described in CIRIAReport 89. The behaviour of 10. RICHART, F. E. and KLUGE, R. W. Tests of reinforced concrete
reinforced concreteflat slabs. The authoris grateful to slabs subjected to concentrated loads, Eng. Expt. Stn. Bulletin
CIRIA forits permissionto report the tests in question. 314, University of Illinois, June 1939.
The slabs of series 111 and IV were tested by C. 11. FORSELL, c. and HOLMBERG, A. Stampellast pA plattor av betong
Hadjchristou; C . Papastavrou; C . Protopapas, and S. (Punching loads on concrete slabs) Betong 5 . Vol. 31 Nr. 2,
February 1946, pp 95-123.
Theodosiou who were students of The Polytechnic. The
author wishes to thank them for their interest in the paper shouldhe in the handsof the
Contributions discussing the above
work. Editor not later than 31 March 1987.

128

Downloaded by [ Imperial College London Library] on [20/02/17]. Copyright © ICE Publishing, all rights reserved.

You might also like