Magazine of Concrete Research, 2004, 56, No.

3, April, 123138

Punching resistance of RC slabs with rectangular columns

D. R. C. Oliveira, P. E. Regan and G. S. S. A. Melo
lia University of Bras

Fifteen tests on high strength concrete slabs with rectangular supports and three different load patterns are reported. The results show that current code preovisions can overestimate punching resistance in some cases. An investigation, made using the finite element method, shows the influences of the shape of a support and the pattern of loading on the distribution of shear. Factors are proposed to allow for these effects while using the control perimeter and basic shear resistance of the CEB Model Code 90, and it is demonstrated that this approach provides strength estimates better than those of MC 90, BS 8110 and ACI 318. There remains a problem of the punching capacity of slabs almost failing in flexure and this is discussed.

Notation
c c max c min cx cy d f9 c fys m flex u0 u1 V VACI VBS VCEB Vflex VMV Vref Vtest side of a rectangular column longer side of column shorter side of column side of column perpendicular to span of a one-way slab side of column parallel to span of a oneway slab effective depth of slab (dx + dy )/2 cylinder compressive strength of concrete yield stress of steel flexural resistance per unit width perimeter of column or loaded area control perimeter shear strength unfactored shear strength from ACI 318 unfactored shear strength from BS 8110 unfactored shear strength from MC90 shear force corresponding to flexural failure shear strength predicted by Mowrer and Vanderbilts equation reference shear strength used as calculation parameter measured shear strength

Vu one-way prop punching r

ultimate shear strength shear stress for a shear failure across the width of a one-way slab shear stress proposed shear stress for a punching failure diameter of a circular column p ratio of flexural reinforcement (rx :ry )

Introduction
Flat plate floors supported by elongated rectangular columns and column-supported slabs spanning predominantly in one direction are fairly common forms of construction. In both cases designers need to consider safety in relation to punching, but most codes of practice give little if any relevant guidance. So long as any concentrated force on a slab is concentric with a load or support, the resulting shear is commonly assumed to act uniformly around the perimeter, at which shear stresses are controlled. This assumption is contrary what might reasonably be expected. This paper briefly reviews available information, describes a series of tests and presents results from finite element analyses. From this base it proposes a simple modification of the approach to punching given in CEB Model Code 90, and adopted in the current draft of EC2. The method proposed is not a complete solution to all the problems, but is shown to be satisfactory in its correlation with most available test data. 123
0024-9831 # 2004 Thomas Telford Ltd

University of Bras lia., Brazilia-OF, Brazil. (MCR 1104) Paper received 2 December 2002; last revised 6 March 2003; accepted 12 March 2003

Oliveira et al.

Existing literature
As early as 1946 tests by Forsell and Holmberg included one in which a square slab supported at four edges was centrally loaded over a 300 3 25 mm area. The punching strength is well predicted by code expres2 sions such as that of CEB Model Code 90 which take no account of the ratio c max /c min , showing that this ratio is not itself a significant parameter so long as c max /d is not large. In this instance c max /d 288d. 3 Mowrer and Vanderbilt made tests of shallow (d 51 mm) lightweight aggregate slabs supported at four sides and loaded through square central columns, with side dimensions up to 8d. The results were compared with predictions made by various methods and the best agreement was found for a modification of 4 Moes equation in which the punching strength is a function of the ratio Vu /Vflex. 5 Hawkins et al. tested slabs supported by rectangular columns with aspect ratios up to 433. In most of the tests the load was applied at the two edges perpendicular to the long sides of the columns. With the length of the column perimeter constant the punching resistance re6 duced appreciably as c max /c min increased. A proposal was made to modify p the then the ACI codes limit shear stress of 0:332 f 9 c , at a perimeter d/2 from the column, by a factor (0:625 0:75cmin =cmax ) < 1:0. This form of expression has been adopted in subsequent additions of ACI 318 but in a more conservative form. 7 Regan and Rezai-Jorabi tested one-way spanning slabs subjected to either one central load or two symmetric loads applied through steel plates. Some of the failures were by punching and others by wide beam shear. In a considerable number of tests the ultimate loads were below conventionally calculated punching resistances and normal beam shear strengths. The approach proposed was to calculate the applied shear at 8 the BS 8110 control perimeter in a two-stage process. In the first stage the concentrated force or forces are resisted by uniform upward pressure in a distribution zone and in the second the pressure is treated as a downward load spanning to the supports. Summation of the shears from the two stages gives the applied stress which is then compared with the resistance according to BS 8110. 9 Leong and Teng tested slabs supported on central columns and loaded near their edges. The major variable was the perforation of slabs by large openings near the columns, but the series included a number of solid specimens. For solid slabs Leong and Teng propose the use of the control perimeter of Fig. 1(a) which is similar to that of BS 8110 but with a reduction in length at large and/or elongated columns. 10 Al-Yousif and Regan tested slabs with elongated columns (c max /c min 5 and c max /d 625). Loads were applied through the columns and the slabs were simply supported on two or four sides. The ultimate loads varied markedly with the support conditions, 124
1

being highest for four-sided support and lowest when the slab spanned one-way parallel to the long sides of the columns. Available test data were used to derive a method of reducing the BS 8110 perimeter to an effective perimeter. ueff 2fx (cx 3d ) y (cy 3d )g with   cx : : <1 x One and two-way slabs: 1 09 0 03 d   8 cy > : : > <1 < One-way slabs: 1 09 0 09 d  y cy > > <1 : Two-way slabs: 1:09 0:03 d where c y is the column dimension parallel to the span if the slab spans predominantly in one direction. The limit of this method is for for cy . 4:55d for one-way slabs for which y (cy 3d) can be taken as 514d. International codes have developed different ways of treating the effects of the column size and rectangularity, but none of them has taken any account of a slabs general flexure (one or two-way). 6 ACI 318, in which the control perimeter for punching is at a distance of d/2 from the loaded area, originally used a single limiting p shear stress with an unfactored value of 0:332 f 9 c . It then introduced a reduction factor for rectangular loaded areas and later another factor for large loaded areas. The limit stress is now as below. p ACI 0:332 f 9 (2) c where is the least of (05 + c min /c max ), (05 + 10d/u1 ) and 10, with u1 (2c max + 2c min + 4d) as ACI 318 uses square-cornered perimeters. 11 The CEB-FIP Model Code of 1978 used a control perimeter 05d from the load but with rounded corners as compared to the square corners of the ACI code. For concentric loading the distribution of characteristic shear resistance was as shown in Fig. 1(b). For large and/or elongated columns, the punching shear resistance applied only at the corners or the shorter sides. For the remainder of the perimeter the resistance was as for one-way shear ( punching 16 one-way ). The 1978 Model Codes definitions of the punching part of the column periphery was adopted in the Pre12 standard version of Eurocode 2. However the distance to the control perimeter was increased to 15d and the non-punching parts of the perimeter were assumed to be stress free in the absence of more detailed analysis. The current CEB-FIP Model Code 90 has moved the control perimeter to 2d from the loaded area, and makes no special provisions for large or rectangular columns. Recent drafts of EC2 have followed this approach.
Magazine of Concrete Research, 2004, 56, No. 3

(1)

Punching resistance of RC slabs with rectangular columns

Control perimeter

1.5 d

cmin 600 mm d 125 mm cmax 2 x 2cmin . 5 6d

Column

1.5 d

cmin

cmin 2

x cmax (a)

1.5 d

One-way shear

Punching shear

0.5d

Column 0.5b1

cmax 2 a1 2cmin 5.6d b1 cmin b1 2.8d

0.5b1 0.5a1 cmax (b) 0.5d

Fig. 1. Control perimeters of (a) Leong and Teng and (b) CEB MC 78

To make comparisons between Codes treatments of the factors in question here the effects of other parameters e.g. the ratio of main steel, the absolute effective depth and the concrete strength have to be eliminated. This can be done by relating all predicted strengths to a reference value Vref, equal to the resistance of a two-way slab at a square loaded area of side length 2d. Because the length of u1 is much more dependent on the column dimensions in ACI 318 than in MC 90, if the ACI code had not introduced the reduction coefficient , the values of VACI /Vref for large square loaded areas would be far above those of VCEB /Vref. With included, the difference between the two Codes is greatly reduced. Thus what the ACI 318 accomplishes with a column-size factor is largely accomplished in MC 90 by a different definition of the control perimeter.
Magazine of Concrete Research, 2004, 56, No. 3

Figure 2 offers comparisons between the predictions of Codes and other methods for square loaded areas and for ones with c max 5c min . The left side considers recommendations that do not distinguish between oneway and two-way spanning slabs and includes all the Codes considered above. The right side of the figure presents results for the methods that take account of the slabs overall flexural behaviour and includes lines for MC 90 to facilitate comparisons. For square loaded areas the predictions of MC 78, MC 90 and ACI 318 are all within 15% of one another, and except for the largest areas Al-Yousif and Regans predictions for two-way slabs are not much lower. However for one-way slabs Al-Yousif and Regan and Regan and Rezai-Jorabis methods give resistances significantly below those of the codes. It should be noted here that the results obtained from Regan and 125

0.5d

cmin

Oliveira et al.
3.0 cmax 5cmin 3.0 cmax 5cmin

2.5

2.5

2.0

MC78 MC90 Leong and Teng ACI EC2-92

2.0 MC90 R & R 1 way with cmax perp. to span A & R 2 way A & R 1 way with cmax perp. to span R & R 1 way with cmax parall. to span A & R 1 way with cmax parall. to span

V/Vref

1.5

V/Vref

1.5

1.0

1.0

0.5

0.5

0.0 0 2 4 6 cmax /d 8 10

0.0 0 2 4 6 cmax /d 8 10

3.0

cmax cmin

MC78 ACI MC90

3.0

cmax cmin MC90

2.5

2.5

2.0

2.0

A & R 2 way R & R 1 way

V/Vref

1.5

V/Vref

Leong and Teng EC2-92

1.5

A & R 1 way R & R 1 way for l 25d and b 15d

1.0

1.0

0.5

0.5

0.0 0 2 4 6 cmax /d 8 10

0.0 0 2 4 6 cmax /d 8 10

Note: 1 Vref Strength of two-way slab and square column with c 2d 2 A & R one or two way Al-Yousif and Regans method applied to one- or two-way slab 3 R & R one way Regan and Rezai-Jorabis method applied to one-way slab 4 b and l are the slab width and the span length respectively for one-way slabs 5 Except where otherwise stated predicitions by Regan and Rezai-Jorabi are for slabs with 1 b 20d

Fig. 2. Comparisons between the predictions of codes and other methods

Rezai-Jorabi depend upon the dimensions of the slabs, which is not the case for other methods. With c max /c min 5 the differences between the codes are much greater, with ACI 318 predicting the lowest strengths for small values of c max /d and EC2 (the prestandard) giving the lowest values when c max /d is large. In the methods taking account of the overall flexural behaviour of the slab there are considerable 126

differences between the strengths obtained for different orientations of the loaded areas, i.e. between cases with c max in the direction of the span and with c max in the transverse direction. Fig. 2 does not include approaches such as Mowrer and Vanderbilts in which Vu is a function of Vu /Vflex but this form of equation does not allow some account to be taken of the differences between one-way and two-way spans.
Magazine of Concrete Research, 2004, 56, No. 3

Punching resistance of RC slabs with rectangular columns

Tests
Programme Tests were made on 15 slabs with overall dimensions 13,14 of 2280 3 1680 3 130 mm. The main reinforcement was fifteen 125 mm bars in the long direction with a nominal cover of 10 mm and twenty three 125 mm bars in the short direction, giving equal ratios of reinforcement (11%) both ways and a nominal mean effective depth of 1075 mm. At their ends these bars were lapped with 63 mm hairpin bars to assist the end anchorages. Reinforcement details are drawn in Fig. 3. Table 1 summarises the main slab properties. The slabs were supported at their centres through 50 mm thick steel plates, 120 mm wide and with lengths varying from 120 to 600 mm (112558d ). Equal loads were applied through beams close to the slab edges. In slabs type a and b the loads were at only two opposite edges (the short and long edges respectively) while in type c all four edges were loaded. The test arrangements for a type c slab are shown in Fig. 4. Those for the other types were similar but with only pairs of loading beams. The loads were measured independently by four load cells.

Table 1. Characteristics of the tested slabs

Slab d: mm 107 108 107 109 106 107 108 107 106 108 106 107 108 108 109 r f9 c: MPa 57 59 59 58 58 57 56 60 54 56 54 56 57 67 63 Column: mm c min L1a L1b L1c L2a L2b L2c L3a L3b L3c L4a L4b L4c L5a L5b L5c 00109 00108 00109 00107 00110 00109 00108 00109 00110 00108 00110 00109 00108 00108 00107 120 120 120 120 120 120 120 120 120 120 120 120 120 120 120 c max 120 120 120 240 240 240 360 360 360 480 480 480 600 600 600 2400 3224 3180 2460 3610 3308 2406 4000 3576 2508 3950 4040 2874 4264 4464 Vtest: kN

Failure modes: Group a slabs: flexural-punching Groups b and c slabs: punching

Materials The concrete used throughout had the mix proportions given in Table 2. Cement CPII F32 is an

15 # 125 mm e 117 mm 2250 mm 23 # 125 mm e 102 mm 1650 mm

100

1 2 60 3 4 5

2280

6 7 8 9 60 Staple 108 400 15

130

50

120 1680

Fig. 3. Flexural reinforcement arrangement and positions of strain gauges Magazine of Concrete Research, 2004, 56, No. 3

10

127

Oliveira et al.
1680 510

ordinary Portland cement with a 610% content of filler which is primarily crushed limestone. Compression strengths were determined from tests of 100 3 200 mm cylinders, cured and stored with the slabs. The main reinforcement was of deformed bars with a yield stress of 749 MPa and an ultimate strength of 903 MPa. The 63 mm bars used for hairpins and for the bottom steel were also deformed and had yield and ultimate stresses of 641 and 799 MPa. Stressstrain relationships from tests of the reinforcement are given in Fig. 5.
2280

90

710

Instrumentation and procedure Slab deflections were measured using dial gauges mounted from independent frames and reading onto targets on the top surfaces of the slabs. Strains of reinforcement were measured at the positions shown in Fig. 3 using pairs of gauges with 5 mm gauge lengths so that averaged values could eliminate local bending effects. Radial strains of the bottom surface of the concrete were measured also using electrical resistance gauges (gauge length 318 mm). Loads were applied in increments of 40 kN of total load and, after each, the slabs were inspected for cracking and measurements were taken.

90

SLAB

Fig. 4. Loading system (plan)

Results
Table 2. Mix proportions
Materials Cement (CP II F32) Silica fume Crushed limestone (16 mm) Sand (5mm down) Water Superplasticiser kg/m3 600 60 1092 512 180 72

All of the tests ended in shear failures. The type c slabs, with loads at four sides, failed in a normal punching mode with failure surfaces being truncated cones. The failure surfaces for slabs type a and b with small reaction areas were similar, but when c max > 360 mm the failure surfaces in the type a slabs did not run around the longer sides of the reactions (Fig. 6). In the type b tests, the failures of slabs L3b and L4b (c max 360 and 480 mm) were concentrated

900 800 700 600 Stress: MPa Stress: MPa 0 3 6 9 12 15 18 21 24 27 30 33 36 Strain: (a) 500 400 300 200 100 0

900 800 700 600 500 400 300 200 100 0 0 3 6 9 12 15 18 21 24 27 30 33 36 Strain: (b)

Fig. 5. Stressstrain graphs for reinforcement: (a) 630 mm; (b) 1250 mm

128

Magazine of Concrete Research, 2004, 56, No. 3

Punching resistance of RC slabs with rectangular columns

37

L3a

L5a

30

30

L3b 31 32

L5b

L3c 36 39 34

L5c 30

Fig. 6. Failure surfaces Magazine of Concrete Research, 2004, 56, No. 3

35

129

Oliveira et al. to the short sides of the reaction areas although in L5b a normal punching surface developed as in L1b and L2b. As is confirmed by calculations of flexural capacities and by the measurements discussed below all of the type a slabs were very close to flexural failure. In types b and c the failures were purely by punching. The deflection data showed that the displacements of the slabs along the lines at which measurements were made increased practically linearly with distance from the reaction area. The dial gauge readings from each line were used to calculate rotations and averaged to obtain transverse and longitudinal values. Fig. 7 shows the greater averaged rotations plotted against total load for slabs L1a, L5a, L3b and L3c. These slabs were used as there were
450 L1a 400 350 300 Load: kN 250 200 150 100 50 0 0 2 4 6 8 10 12 Strain: 14 16 18 20 1 2 3 4 5 6 7 Load: kN

450 400 350 300 Load: kN 250 200 150 100 50 0 0 5 10 15 20 Deflection: mm 25 L1a (y) L5a (y) L3b (x) L3c (y) 30 35 Vtest 400 kN Vtest 358 kN Vtest 287 kN Vtest 240 kN

Fig. 7. Load versus deflection

450 L5a 400 350 300 250 200 150 100 50 0 0 2 4 6 8 10 12 Strain: 14 16 18 20 1 2 3 4 5 6 7 8 9

450 L1b 400 350 300 Load: kN 250 200 5 150 100 50 0 0 2 4 6 8 10 12 Strain: 14 16 18 20 6 7 1 2 3 Load: kN 4

450 L5b 400 350 300 250 200 150 100 50 0 0 2 4 6 8 10 12 Strain: 14 16 18 20 1 2 3 4 5 6 7 8 9

450 400 350 300 Load: kN 250 200 150 100 50 0 0

L1c

450 1 2 3 Load: kN 4 5 6 7 400 350 300 250 200 150 100 50 0

L5c

1 2 3 4 5 6 7 8 9

8 10 12 Strain:

14

16

18

20

8 10 12 Strain:

14

16

18

20

Fig. 8. Reinforcement strains

130

Magazine of Concrete Research, 2004, 56, No. 3

Punching resistance of RC slabs with rectangular columns readings at or close to their ultimate loads. The figure shows the difference in behaviour between types a, b and c very clearly. In type a slabs, the strains measured on the bars in the direction of the main span, 60 mm from the edge of the reaction zone, showed yielding of all bars within a width 22d from the centre line. At the edge the strain reached 1156% in slab L1a and 3136% in the other slabs. None of the strains of the transverse bars exceeded 155%. In the type b slabs, spanning predominantly in the shorter direction, i.e. perpendicular to the longer sides of the reaction areas, there was no yielding of the long (transverse) bars. The main bars did not yield in L1b, but yielding was recorded in the instrumented bars of the others slabs, which were however very close to the loads (see Fig. 8). A comparison between the experimental ultimate loads and the unfactored resistances according to ACI 318, BS 8110 and CEB MC 90 is shown in Fig. 9. For the type a slabs, with the main span parallel to the longer sides of the supports, all three codes overestimate resistance and this is almost certainly because of the partly flexural nature of the failures. For the other two types ACI 318 is safe but not very consistent, while BS 8110 and MC 90 resistances are close to the actual strengths for the smaller supports but tend to become unsafe for the larger ones. The aim of the analysis was to investigate the shear force distribution around the columns and along the model code control perimeter. This perimeter was adopted to plot the results due to both its reasonable concordance with the failure surfaces from the tests and its giving good results for the integration of the shear forces. The supports were modelled as follows. For slabs type b and c the nodes at the boundaries of and within the support areas were pinned. For slabs type a only the three nodes at the short sides were pinned since central upward displacements were observed in the tests. Loads were applied uniformly along lengths of 480 mm at short edges and/or 660 mm at long edges. The shear force contours around the column for slabs L1a, L5a, L5b and L3c are presented in Fig. 10. For all slabs with cmax =cmin . 1 is possible to note clearly the influence of the column shape on the shear polarisation even for one-way slabs where the applied load is parallel to the long side of the column. This characteristic is not present in the shear distributions along the model code control perimeter shown in Fig. 11. At the model code perimeter, the shears are greatest near the support sides perpendicular to the spans for slabs of types a and b and the variation around the perimeter increases with the ratio cmax =cmin . For type a the ratio of the maximum to the average shear rises for 117 for cmax =cmin of 10164 for cmax =cmin equal to 5. For type b the variation is much smaller and reaches only 123 for slab L5b. In type c, with half the loading applied at the short edges, the concentration of shear is almost as great as for type a, with a maximum/average shear ratio of 160 for slab L5c. If the maximum shears from Fig. 11 are compared with the unfactored model code values from equation (3) r! 200 p 3 rk d 0:18 1 100r f 9 (3) cd d calculated for the parts of the perimeters where the shears are highest, the ratio of experimental (test + FE analysis) shear to rk d for types b and c is generally satisfactory (102114) but rises to 130 for slab L5c. The values for type a remain low at 082093. For type a the problem remains the proximity of flexural failure.

Finite element analysis

All the slabs were modelled and analysed using the finite element method through the program Structural Analysis Program (SAP). The mesh was the same for all slabs and the applied loads were the failure loads. The elements used were rectangular shell elements 60 3 60 mm with four joints.

600 500 V (60/f c )1/3: kN 400 300 200 Slabs a 100 0 0 Slabs c BS 1 2 3 cmax /d 4 5 Slabs b ACI CEB 6

Proposal
Is proposed here to take account of the effects of the shear polarisation by factors to be incorporated in the model code for cases of symmetrical punching for twoway and one-way slabs. For design an effective applied shear force (Vsd,eff VSd ) should be calculated such that (Vsd,eff =u1 d ) can be compared with the normal shear 131

Fig. 9. Comparison of test results with unfactored predictions from codes Magazine of Concrete Research, 2004, 56, No. 3

Oliveira et al.

70140210280350 420 N/mm

70140210280350 420 N/mm

Slab 1a

Slab 5a

200300 400500 600700 N/mm

200 400600 80010001200 N/mm

100

Slab 5b

Slab 3c

Fig. 10. Shear contours for slabs L1a, L5a, L5b and L3c

resistance rd rk =m . For comparison with test data, this implies a proposed strength equation r! 0:18 200 p 3 1 100r f 9 V Prop (4) c du1 d Values for have been derived from available test data (Table 4) taking account of the principal conditions 132

which can occur relating to the directions in which a slab spans and the orientations of the longer and shorter sides of rectangular supports. The corresponding expressions for are given in Table 3. Mowrer and Vanderbilts tests are useful as they include slabs loaded through large square columns, but they introduce the problem of lightweight aggregate concrete in association with a very small effective
Magazine of Concrete Research, 2004, 56, No. 3

Punching resistance of RC slabs with rectangular columns

L1a 154 N/mm 126 N/mm L5a 168 N/mm 54 N/mm

L3b 141 N/mm 200 N/mm

L5b 107 N/mm 187 N/mm

L3c 204 N/mm 77 N/mm

L5c 255 N/mm 38 N/mm

Fig. 11. Shear forces along the CEB control perimeter for representative slabs

Table 3. Regularisation factors

Situation Two-way slabs  : c max 0 02 1:03 d  One-way c max > c min parallel to span c max d 0:17

One-way c max perpendicular to span

 : c max 0 14 0:93 d

depth. To be able to use these data in considering the effect of column size these predictions of equation (4) have been multiplied by 09 to give an average Vtest =Vprop of 10. For the other methods of calculating resistances considered here the lightweight aggregate factors of ACI 318 and BS 8110, i.e. 085 and 080 respectively have been taken into account, while no correction has been made for MC 90 which does not treat lightweight concrete. The results obtained are given in Table 5 where it
Magazine of Concrete Research, 2004, 56, No. 3

can be seen that the proposed approach reduces the coefficient of variation of Vtest =Vprop relative to those of all three codes. The results of comparisons for normal weight concrete which exclude slabs L1aL5a are shown in Figs 12 and 13 which show that the proposal eliminates the trend of both MC 90 and BS 8110 resistances to become less safe as c max /d increases. The scatter of Vtest =CACI is larger than for the other methods principally because of ACI 318s neglect of the influence of the ratio of flexural reinforcement. It can be observed that the improvements produced by the proposed method are more relevant for one-way slabs with the long dimension of the column parallel to the span. Figure 14 considers lightweight concrete and compares Mowrer and Vanderbilts results with the predictions of ACI 318, BS 8110 and MC 90. The strengths of the present type a slabs are not predicted satisfactorily by using the factors above. The distinct difference between these and the other slabs can be appreciated by considering the results for slabs L1a and L1b. Both had square supports of the same size and both spanned one-way. Their effective depths, reinforcement ratios and concrete strengths were similar and yet the ultimate loads were 240 kN for L1a and 322 kN for L1b. The reason for the 133

Oliveira et al.
Table 4. Application of the regularisation factors for available test data
Author 1 Slab L1a L1b L1c L2a L2b L2c L3a L3b L3c L4a L4b L4c L5a L5b L5c 1 2 3 4 5 6 7 8 9 14R 15R 19R B1 B2 C1 C2 S1 OC11 OC13 OC15 OC13a1:60 C11F22 C13F22 C15F22 C13F11 2 3 4 10 11 12 A7 A8 A2a A2b A2c A7b A5 DT1 BD2 BD8 Type OW1 OW1 TW OW1 OW2 TW OW1 OW2 TW OW1 OW2 TW OW1 OW2 TW OW1 OW1 OW1 OW1 OW2 OW2 TW TW TW OW1 OW2 OW2 OW2 OW2 OW2 OW2 OW2 TW TW TW TW TW TW TW TW OW2 OW2 OW2 TW TW TW OW1 OW1 TW TW TW TW TW OW1 OW1 TW d: mm 1070 1080 1070 1090 1060 1070 1080 1070 1060 1080 1060 1070 1080 1080 1090 1173 1173 1173 1173 1173 1173 1173 1207 1207 790 790 790 955 1010 2010 2010 6190 1053 1073 1028 1098 1550 1550 1600 1590 1110 1060 1100 1040 1120 1080 1145 1145 1145 1145 1145 1145 1145 1900 1010 1010 r: % 109 108 109 107 110 109 108 109 110 108 110 109 108 108 107 112 112 112 112 112 112 086 080 076 100 100 099 076 051 075 052 057 181 171 176 167 172 166 164 107 064 067 064 068 063 065 248 248 248 248 248 248 248 128 128 128 fc 9: MPa 570 590 590 580 580 570 560 600 540 560 540 560 570 670 630 309 269 326 316 274 231 264 266 301 310 308 290 251 250 235 232 306 360 358 402 330 354 356 354 355 123 123 123 176 176 176 285 219 137 195 374 279 278 436 422 353 fys : MPa 750 750 750 750 750 750 750 750 750 750 750 750 750 750 750 419 419 419 419 419 419 419 422 422 670 670 670 709 665 711 706 622 452 452 452 470 460 460 460 520 321 321 321 321 321 321 321 530 530 530 Column: mm c min 3 c max 120 3 120 120 3 120 120 3 120 120 3 240 120 3 240 120 3 240 120 3 360 120 3 360 120 3 360 120 3 480 120 3 480 120 3 480 120 3 600 120 3 600 120 3 600 305 3 305 203 3 406 152 3 457 114 3 495 152 3 457 152 3 457 152 3 457 114 3 495 152 3 305 75 3 100 100 3 150 100 3 150 120 120 240 240 800 200 3 200 200 3 600 200 3 1000 200 3 600 250 3 250 250 3 750 250 3 1250 250 3 750 140 140 140 25 3 300 140 3 540 140 3 340 254 3 254 356 3 356 254 3 254 254 3 254 254 3 254 254 3 254 356 3 356 150 3 150 100 3 100 100 3 100 Vtest kN 240 322 318 246 361 331 241 400 358 251 395 404 287 426 446 391 358 340 337 362 342 326 321 322 154 172 170 181 180 648 547 4915 423 568 649 508 627 792 1056 769 176 172 177 186 279 265 400 436 334 400 467 512 534 780 293 251 Vprop kN 331 323 354 356 382 383 388 430 450 495 364 330 346 337 363 343 350 355 326 149 169 165 184 175 643 569 4822 369 497 627 493 664 861 1102 767 177 167 175 190 273 228 424 428 363 409 508 460 533 823 277 253 V test V prop 097 099 102 093 105 093 102 094 095 090 108 108 098 100 100 100 093 090 099 103 102 103 099 103 101 096 102 115 114 104 103 094 092 096 100 100 103 101 098 102 116 094 102 092 098 092 111 100 095 106 099 V test V flex 100 049 067 094 056 065 088 062 066 085 062 067 091 065 066 108 093 085 082 074 106 078 077 092 065 068 055 070 098 054 061 061 082 109 106 073 049 050 017 061 046 049 051 084 099 086 096 103 057 061 063 072 070 092 098 066

(continued overleaf )

134

Magazine of Concrete Research, 2004, 56, No. 3

Punching resistance of RC slabs with rectangular columns

Table 4. (continued)
Author 9 Slab 1 2 3 4 1 2 4 L42 L42a L45 L46 L41 L41a L43 L44 1 2 3 4 5 6 7 8 9 10 11 12 13 14
13

Type OW1 TW OW2 TW OW2 OW2 OW2 TW TW TW TW TW TW TW TW TWLW TWLW TWLW TWLW TWLW TWLW TWLW TWLW TWLW TWLW TWLW TWLW TWLW TWLW

d: mm 800 800 800 800 1230 1230 1230 1390 1640 1540 1640 1390 1640 1640 1640 510 510 510 510 510 510 510 510 510 510 510 510 510 510

r: % 098 098 098 098 088 087 109 146 123 131 123 146 123 123 123 110 220 110 220 110 220 110 220 110 220 110 220 110 220

fc 9: MPa 236 232 212 220 426 449 456 432 362 420 393 447 389 387 400 286 249 211 180 155 272 233 229 280 264 278 250 249 246

fys : MPa 472 472 472 472 452 452 452 604 604 604 604 604 604 604 604 386 386 386 386 386 386 386 386 386 386 386 386 386 386

Column: mm c min 3 c max 100 3 100 3 100 3 300 3 500 500 500 300

Vtest kN 163 209 189 242 367 349 393 657 693 798 911 563 600 726 761 86 102 79 99 93 133 109 152 119 158 138 185 145 185

Vprop kN 171 217 187 216 359 364 394 613 684 753 864 540 621 698 766 79 95 84 100 87 132 113 141 134 165 147 179 155 194

V test V prop 095 096 101 112 102 096 100 107 101 106 105 104 097 104 099 108 107 094 099 106 100 097 107 089 096 094 103 094 095

V test V flex 071 054 084 062 094 120 081 051 046 056 060 045 040 049 051 114 076 103 076 118 087 125 097 125 092 137 102 135 095

10

171 171 171 200 3 200 3 200 3 200 3 150 3 150 3 150 3 150 3 102 3 102 3 152 3 152 3 203 3 203 3 254 3 254 3 305 3 305 3 356 3 356 3 406 3 406 3 400 400 600 800 250 250 450 600 102 102 152 152 203 203 254 254 305 305 356 356 406 406

11

12

Oliveira 5 Hawkins et al 7 Regan and Rezai-Jorabi 15 Nylander and Sundquist 9 Leong and Teng 1 Forssel and Holmberg 16 Elstner and Hognestad 17 Regan 10 Al-Yousif and Regan 18 Borges 19 Mahmood 3 Mowrer and Vanderbilt Legend: OW1 : One-way slab with cmax parallel to span OW2 : One-way slab with cmax perpendicular to span TW: Two-way slab TWLW: Two-way lightweight aggregate concrete

difference lies in the proximity of flexural failure in all the type a tests. 1517 Tests of two-way slabs show that while the ratio Vu /VCEB is generally independent of Vu /Vflex, there is a decrease of punching resistance by up to 20% when the load is at or very close to the flexural capacity. The main reason for this is probably a lowering of resistance due to wide cracks and high concrete strains. In the case of one-way slabs there is probably an additional effect from an increasing concentration of shear to the column faces perpendicular to the span as
Magazine of Concrete Research, 2004, 56, No. 3

transverse yield lines develop. In the extreme this could reduce the active part of the control perimeter to that obtained at two edge columns contacting a slab only at their inner faces as shown in Fig. 15 for which the shear resistance could be estimated according to equation (5) r! 200 p 3 V min 0:18 1 100r f 9 c d 3 d(2c 3d) (5) 135

Oliveira et al.
Table 5. Comparison of the results from ACI, BS 8110, CEB and proposal
Codes and proposal V test V calc:
VTest /V 1.4 1.2 1.0 OW1: Prop OW2: Prop OW2: CEB OW1: CEB

TW: Prop TW: CEB

Normal weight concrete Av ACI BS 8110 CEB Proposal

a

Lightweight concretea Cv 170 118 084 100 Sd 030 009 006 006 Cv: % 1741 789 749 621

0.8 0.6 0.4 0.2 0.0 0 2 4

SD 022 009 009 006

Cv: % 1623 853 904 577

137 101 095 100

Results obtained using the test results of Mowrer and Vanderbilt with two-way slabs loaded through square areas.

6 cmax /d

10

12

Fig. 13. Trend lines for the results from CEB with and without proposed modification

Table 6 summarises the results for the type a slabs. The data are too limited for any definite conclusion to be reached but it appears that the ultimate resistances of such slabs when very close to flexural failure might be calculated by either reducing the normal estimate of VCEB / by 30% or considering the reduced perimeter of Fig. 15. An alternative approach would be to accept the reduction in the ratio (Vu /VCEB ) on the basis that the partial safety factor on resistance can be allowed to decline from 15 for a shear failure to 115 for a flexural failure (115/15 077).

Conclusion
The results of the tests reported here and others in the literature show that the punching resistances of flat slabs are influenced by the shapes and sizes of their supports and by their overall flexural behaviour in ways not properly accounted for in current Code provisions. Elastic analyses of uncracked slabs help to explain the results obtained by illustrating the distributions of

2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

2.0 1.8 1.6 1.4 VTest /VBS ACI 0 2 4 6 cmax /d 8 10 12 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 2 4 6 cmax /d 8 10

VTest /VACI

BS 12

2.0 1.8 1.6 VTest /VProp CEB 0 2 4 6 cmax /d 8 10 12 VTest /VCEB 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0

2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 0 2 4 6 cmax /d 8 10 12

Fig. 12. Results from codes and proposal

136

Magazine of Concrete Research, 2004, 56, No. 3

Punching resistance of RC slabs with rectangular columns

2.4 2.2 2.0 1.8 1.6 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0.0 ACI 0 2 CEB 4 c/d BS 6 Proposal 8 10

Table 6. Results for type a slabs

Slab Column: mm c min L1a L2a L3a L4a L5a 120 120 120 120 120 c max 120 240 360 480 600 240 246 241 251 287 100 094 088 085 091 074 073 070 069 075 106 106 106 111 126 Vtest: kN V test V flex V test V CEB V test V min

Vtest /V

Fig. 14. Results for lightweight aggregate concrete

shear around the control perimeter of the CEB-FIP model code. Most of the effects observed can be taken into account, in the general context of the model code by the introduction of a factor expressing the non-uniformity of the shear distribution that can occur even at concentrically loaded supports. This factor is a function of the ratio (cmax =d ) between the larger dimension of the support and the effective depth of the slab, of the slabs overall flexure (one or two way spanning) and of the direction of c max (parallel or perpendicular to a oneway span). With the proposed values of included, the modified CEB-FIP method gives predictions of ultimate strengths which are significantly better than those of the unmodified model code, AC I318 and BS 8110. There remains the problem of the reduction of

punching capacity, which occurs when the load is very close to a slabs flexural resistance. This is discussed, and two methods of obtaining approximate ultimate loads are given, although it may be that the difference between the partial safety factors governing resistances to flexure and punching makes such calculations unnecessary.

Acknowledgements
The authors would like to acknowledge the support of Brazilian Scientific and Technological Development Agency (CNPq), FINATEC, CAPES and Imperial College (London).

References
. Sta plattor av betong. 1. Forssel C. and Holmberg A mpellast pa Betong, 1946, 31, No. 2, 95123. 2. CEB-FIP. Model Code 1990, Thomas Telford, London, 1993. 3. Mowrer R. D. and V anderbilt M. D. Shear strength of lightweight aggregate reinforced concrete. ACI Journal, 1967, 64, No. 11, 722729.

Control perimeter 22.5

Column

Fig. 15. Possible control perimeter for type a slabs Magazine of Concrete Research, 2004, 56, No. 3

137

Oliveira et al.
4. Moe J. Shearing strength of reinforced concrete slabs and footings under concentrated loads. Bulletin No. D47, Portland Cement Association Research and Development Laboratories, Skokie, Illinois, 1961, 129 pp. 5. Hawkins N. M., Falssen H. B. and Hinojosa R. C. Influence of column rectangularity on the behaviour of flat plate structures. Publication SP-30, American Concrete Institute, Detroit, 1971, pp. 127146. 6. American Concrete Institution. Building Code Requirements for Structural Concrete, ACI, Farmington Hills, Michigan, 2002, AC1318. 7. Regan P. E. and Rezai-Jorabi H. Shear resistance of one-way slabs under concentrated loads. ACI Structural Journal, 1988, 85, No. 2, 150157. 8. British Standards Institution. Structural Use of Concrete, Part 1, Code of Practice for Design and Construction, BS, London, 1997, BS 8110. 9. Leong K. K. and Teng S. Punching Shear Strength of Slabs with Openings and Supported on Rectangular Columns, Nanyang Technological University, Singapore, August, 2000. 10. Al-Yousif A. T. and Regan P. E. Punching resistances of RC slabs supported by large and/or elongated columns, The Structural Engineer, in press. 11. CEB-FIP. Model code for Concrete Structures, CEB, Paris, 1978. 12. Eurocode 2. Design of Concrete Structures, Part 1, General rules and rules for buildings, DD ENV 1992-1-1, BSI, London, 1992. Oliveira D. R. C. Experimental Analysis of Reinforced High Strength Concrete Flat Slabs with Elongated Columns, PhD thesis, University of Brasilia. Oliveira D. R. C. and Melo G. S. Reinforced concrete flat slabs with elongated columns. 6th International Symposium on Utilization of High Strength/High Performance Concrete, Leipzig, Germany, June, 2002, Vol. 1, pp. 445456. Nylander, H., Sundquist , H., Genomstansning av pelarundersto dd plattbro av betong med ospa nd armering. Meddelande Nr. 104, Institutionen fo r Byggnadsstatik, KTH Stockholm, 1972. Elstner, R. C., Hognestad, E., Shear strength of reinforced concrete slabs. ACI Journal, 1956, Proceedings 53, No. 1, July, 2957 Regan , P. E., Punching shear in prestressed concrete slab bridges. Polytechnic of Central London, January 1983. Borges, L. L. J. Comportamento ao Puncionamento de Lajes Cogumelo com Pilares Retangulares e Furos de Grandes Dimenso es, MSc. Thesis, University of Brasilia, May 2002. Mahmood, K. F. Slabs with Point Supports, M.Phil. Thesis, Polytechnic of Central London, 1978.

13.

14.

15.

16.

17. 18.

19.

Discussion contributions on this paper should reach the editor by 1 October 2004

138

Magazine of Concrete Research, 2004, 56, No. 3