Professional Documents
Culture Documents
This paper presents the direct decompression method for calculating the punching strengths of post-tensioned slabs. It is a method already used for determining the shear strengths of beams. In many respects it is similar to the Federation Internationale de la Precontrainte (FIP) treatment of punching in post-tensioned slabs, but it is simpler to use. The predictions of two variants of the direct decompression approach, and of the FIP method, are compared with the results of tests of slabs with various arrangements and proles of tendons. It is shown that the direct method reduces the scatter of ratios of experimental and calculated strengths although all three approaches provide reasonable results.
Ricardo Jose Carvalho Silva University of Brasilia, Brazil Paul Erskine Regan University of Westminster, London, UK
0 scp
b b0 bt c d
Notation
a ao ae
distribution factor for moments due to loads other than prestress m0 /m distribution factor for in-plane effect of 0 prestress mpo/mpo distribution factor for moments due to eccentricity and vertical effects 0 of prestress mpe/mpe ratio of bonded reinforcement p including bonded (r1xr1y), tendons. Where bonded tendons are included r1x and r1y are calculated for the sums of the areas of prestressed and non-prestressed reinforcement and the effective depths of the centroids of the combined reinforcement mean prestress of concrete over width b m m0 mpo 0 mpo mpe
r1
0 mpe
uo
scp
u1
mean prestress of concrete over width b0 breadth of test slab breadth of slab ( c 4d in the FIP method and method DD1) breadth of slab over which tendons are uniformly distributed (see Figure 6) width of support effective depth of bonded reinforcement (dx dy)/2. Where bonded tendons are included their effective depths can be calculated for the section where they cross a plane rising at an inclination of 1 : 2 from the soft at the edge of the support effective depth of tendons at centre lines of support characteristic cylinder strength of concrete mean cylinder strength of concrete overall depth of slab coefcient used in determining local moments due to prestress see equations (6c) and (6d) average moment over width b due to loads other than prestress average moment over width b0 due to loads other than prestress scph 2/6 0 scph 2/6 average moment over width b, due to eccentricity and vertical effects of prestress average moment over width b0 , due to eccentricity and vertical effects of prestress length of perimeter of support (4c for a circular support and pc for a circular one) length of control perimeter 2d from a support
V Vo Ve VE Vp
VR VR,c
VRk,c VRk Vu
punching shear force component of decompression load related to mean prestress (scp) component of decompression load related to moments due to prestress applied value of V sum of vertical components of tendon forces transmitted to support without causing shear in a slab resistance to V resistance to V for an rc slab (with the same area of bonded reinforcement as a prestressed slab) characteristic value of VR,c characteristic value of VR experimental ultimate value of V
Introduction
A generally acceptable physical theory for the punching of post-tensioned slabs has still to be developed, and there are considerable divergencies between current design recommendations. The methods given in ACI 3181 and Eurocode 22 are very simple to use, but take no account of either the plan arrangement of the tendons or their vertical eccentricities. Methods based on decompression seem more physically reasonable and do allow account to be taken of all the relevant details of the prestress. One decompression method is that of the deration Internationale de la Precontrainte Fe 3 (FIP) recommendations, and a somewhat simpler alternative can be derived from the BS 81104 treatment of shear in prestressed beams. An earlier version of this direct decompression method was given in a paper by Regan.5
10
Although decompression approaches to punching have some physical basis, the lack of a fully satisfactory ultimate load theory means that the acceptability of any design method can be judged only by comparisons with test results. With the notable exception of a report by Kordina and Nolting,6 related to proposals for German Codes, very little comparative data seems to have been published. The present paper describes the FIP method and the direct decompression alternative, including a variation, which could also be used in the FIP approach. Strengths calculated by the two methods are then compared with test data available in the literature and with the results of tests by the rst author.7 The methods are described initially in the context of typical punching test slabs, representing the negative moment regions around supports. Issues arising in their application to complete oors are discussed following the comparisons with test data. The methods of analysis are, in principle, applicable to bonded as well as unbonded post-tensioning, although all the relevant test data available for oor-slab models are for slabs with unbonded tendons. Some results for slab-bridge models with bonded tendons are discussed at the end of the section on analysis.
in the original decompression method for beams,8 and do have some theoretical advantage, all current methods appear to use the former. For this reason, and to avoid difculties in the denition of the level of the reinforcement, surface stresses are used throughout the following sections. The basic idea of decompression is illustrated in Figure 1. Figure 1(a) shows the prestress acting alone and producing compression of the top surface at the edge of the support. Figure 1(b) shows the combined effects of the prestress and the vertical decompression load. The actions shown are radial and are accompanied by tangential forces. The system of Figure 1(b), supports the decompression load, without requiring forces in the nonprestressed reinforcement or increments of forces in the tendons. Figure 2 divides the actions of Figure 1 into two groups: one related to the effects of the horizontal forces from prestress and the other related to its vertical actions. The corresponding components of the decompression load are: Vo m0po (m0 =V) and Ve m0pe (m0 =V) (1)
There will also normally be some increases of the forces in the unbonded tendons. The details of the combination of a exural system, reliant on bond, and a vault action, dependent on end-anchored prestress, are complex. However, a simple direct addition of the two is a possibility. It implies that that the exural action requires the lower part of the radial compression zone at the support, leaving an upper zone, with a lever arm, similar to that in Figure 1(b) for the vault action. This is the basis of the direct decompression approach, in which VR VRc V (m0po m0pe ) m0 (2)
In the FIP approach the combination is more complex9 and m0po m0pe )
VR VRc Vp (V Vp )
(m0
(3)
0 0 0 where mpo s cph 2/6; s cp is the mean prestress of the concrete in the width b0 considered; mpe is the moment in the same width due to the eccentricity and out-of-plane actions of prestress; m0 /V is the ratio of the average moment in b0 , due to the loading other than prestress, to the punching force V and is a constant for a given slab and loading arrangement if the moments are determined elastically. The situation in Figure 1(b) is clearly not that at the ultimate load. At the ultimate load there will be additional forces associated with the bonded reinforcement, and the level of the centre of compression at the face of the support will have moved downward.
Here VRc is the punching resistance of an ordinary rc slab having the same bonded reinforcement as the prestressed slab; Vp is the sum of the vertical components of the forces in the tendons at their intersection with a perimeter h/2 from the support. In either equation the characteristic value of VRc can be taken as the value from the CEB-FIP model code.10 1=3 p VRk,c 018 1 200=d 100r1 fck u1 d (4) in N and mm units. r1 is the ratio of bonded p reinforcement (r1xr1y), calculated for widths equal to those of the support plus 3d to either side. Bonded tendons can be included in r1, u1 is the length of a control perimeter at a distance 2d from the support and d is the
P (a)
P h h/3 (b) s
Structural Concrete
2005 6 No 1
11
(m po )
(m ) pe
P sinq
P sinq
P sinq Vo + V e
P sinq
effective depth of the bonded reinforcement (dx dy)/2. The difference between the perimeters used for calculating shear stresses and for determining Vp seems reasonable. Any effect from the vertical components of the tendon forces can arise only for tendons crossing the potential inclined failure surface, not the vertical surface used for calculating nominal shear stresses. For the development of Vp forces to be possible there must also be a depth from the tendons to the failure surface sufcient for local failures of the concrete to be avoided. The main differences between the two approaches are the following. . The FIP equation includes the term Vp, which is absent in the direct approach, where it is assumed that all the prestress actions are included in the treatment of decompression.
. In the FIP equation the last term reduces as the load (or m0 ) increases, whereas in equation (2) the nal term is independent of m0 , provided (V/m0 ) is a constant.
In the FIP recommendations the moments are those on lines through the edges of the supports and are averaged over widths equal to those of the control perimeter u1, namely widths c 4d. This seems a reasonable local width and moderate variations of it have little effect on the calculated resistance. If a nite element analysis is used, values of 0 0 m0 , mpo and mpe can be obtained directly, but in other cases it is useful to have simple direct expressions for the average moments in widths (c 4d ) m 0 am m0po ampo m0pe ampe (5a) (5b) (5c)
where m, mpo and mpe are the average moments calculated for the full width of a test specimen. The coefcients a, ao and ae have been evaluated by nite element analysis and can
Structural Concrete
2005 6 No 1
12
a 200
a = 2 - 13 (c + 4d )/b
The FIP recommendations allow the resistance to be checked against an applied load VE, by substituting VE for V on the right-hand side of equation (3), whence VRk VRk,c Vp (VE Vp ) m0po 0 =V) m0 VE (m pe (8)
135
b 050
(c + 4d )/b
The calculation of an ultimate resistance is however a little more complicated, since, if VE and VRk are equated, the result is a quadratic equation.
2 VRk VRk
4 Figure 3 a as a function of (c 1 4d )/b reasonably be approximated as follows (see Figures 3, 4 and 5). In order to nd expressions for the coefcients (equations (6a), (6b) and (6c)), 35 prestressed slabs7,11 13 have been analysed, via nite elements by the program Sap-2000. The nite elements used were type Shell with four nodes of size 50 mm 50 mm and with Poissons ratio equal to 0.20. From each analysis (mpo, m and mpe), the average value of the bending moment in the (c 4d) strip has been calculated and the alpha equations 0 have been found by division m0 /m, mpo/mpo 0 and mpe/mpe. c 4d ! 135 b bt ao 13 03 b bt k ae (1 k) b c 4d 20 where k 3 4 b and where bt is the width over which the tendons exert substantially uniform forces at the slab edges, and b is the slab breadth. The effect of tendon layouts more complicated than a single banding can generally be obtained by adding values of aempe or aompo for two band widths and the corresponding forces. A simplied view of this, and a treatment of bands with central gaps, are given in Figure 6. For square and circular columns the unit moments can be averaged for two orthogonal directions. For rectangular columns averages weighted in proportion to (cx 4d ) and (cy 4d ) may be preferable to simple averages.3 Once the moments have been evaluated the direct decompression method gives a simple expression for characteristic punching resistance. V (m0po m0pe ) (7) VRk VRk,c m0
(9) Both the model code and the FIP recommendations give and upper limit to punching resistance VRk,max 03fck (1 fck =250)uo d (10)
where uo is the length of the perimeter of the support. This is unlikely to be critical for a oor slab and does not affect the calculated strengths of any of the slabs in the next section.
a 2 13
Test data
All of the test data considered here relate to slabs prestressed by unbonded tendons and reported to have failed by punching. The slabs contained bonded non-prestressed tension reinforcement in the form of deformed 606 MPa). bars (455 fy The tests by the rst author were of slabs 2.0 m square on plan and about 125 mm thick. The principal variables were the level of prestress (scp 2.2 or 3.3 MPa), the size of the support (100, 200 or 300 mm) and the plan arrangement of the tendons which was generally one of banding in both directions (bt 400 to 1500 mm), although there were three slabs with a central gap within the tendon band. These details are shown in Figure 7 and Table 1. The tendon proles were circular arcs springing from points 55 mm above the bottom surface at the slab edges.
ao 130
ao = 13 - 03 (bt/b )
100
bt b
bt/b 100
Structural Concrete
2005 6 No 1
13
Analysis
The test results have been analysed in the following manner.
ae 200 bt b k = 200 bt b k = 190 k = 180 k = 170 k = 160 k = 150
150
. Values of VRk,c have been calculated from equation (4), with r1 as dened and with mean cylinder strengths fcm replacing fck. . Values of m and (V/m) have been found from the test arrangements shown in Figure 8. . Values of mpe have been calculated as 0 scph(dp 0.5h) and mpe values have been calculated with se from equation (6c). Three resistances have been calculated for each stab. . VRk,FIP the FIP value from equation (9). . VRk,DD1 the direct decompression value from equation (7); with b0 c 4d. . VRk,DD2 a simplied decompression value with m, mpo and mpe substituted for m0 , 0 0 mpo and mpe in equation (7); that is the values obtained from full width moments rather than moments in widths (c 4d ). The ratios of the experimental ultimate forces Vu to the calculated resistances (abbreviated to VFIP, VDD1 and VDD2) are given in Table 2 and a summary of statistics of the comparisons for the post-tensioned and the rc slabs is presented in Table 3. The data for ordinary rc slabs are too limited to be of wider signicance, but the mean value of Vu/VRk,c 1.0, with its coefcient of variation of 5.5%, can be borne in mind when considering the results for prestressed slabs. The punching resistances of post-tensioned slabs depend on many parameters, some of which are partially interlinked, and it is difcult to discern the effects of individual factors from a tabulated comparison. Figure 9 shows the ratios of experimental and calculated strengths plotted against Vu/VRk,c for all three methods of calculation and for all the results from post-tensioned slabs. With the FIP method there is a very clear increase in Vu/VFIP as Vu/VRk,c increases; that is, as the inuence of prestress increases.
The other test series included are those of Correa,11 Kordina and Nolting,12 Hassanzadeh,13 Melges14 and Shehata.15 The plan arrangements of the loads and supports for all these series are drawn in Figure 8. Data for the slabs are given in Table 1 where the specimens are described as being of four types. Type 1 28 slabs with inclined tendons in single or double bands. Type 2 5 slabs with inclined tendons in single bands with central gaps. Type 3 4 slabs with horizontal tendons at mid-height.
Type 4 3
ordinary slabs.
reinforced
concrete
All of the slabs were approximately biaxially symmetric and the characteristics given are averaged for the x and y directions. Values given for scp are those at the start of testing. Values for Vp are calculated for the perimeter h/2 from the supports, as in the FIP recommendations, and relate to the tendon forces at slab failure (estimated in cases where actual values are not reported). Vp has been determined in this way to treat the problem of the force increments in test slabs
(a) example A - slab with tendons in double bands mpe = ap1mp1 + ap2mp2 mp1 = average moment in b from 4 strands mp2 = average moment in b from 3 strands ap1 & ap2 correspond to bt1/b and bt2/b
(b) example B - slab with tendons in a single band with a central gap bt = 2So + 2S
Structural Concrete
2005 6 No 1
14
5 100 mm
3 100 mm
5 250 mm
3 250 mm
5 100 mm
3 100 mm
5 150 mm
3 150 mm
5 100 mm
3 100 mm
5 150 mm
3 150 mm
100 mm
100 mm
100 mm
100 mm
3 100 mm
200 mm
400 mm
There is little, if any such trend with the direct decompression methods, and it would seem that the FIP approach underestimates the effect of prestress, by its reduction of the decompression component of resistance at loads above decompression. In Figure 10 the ratios of experimental and calculated strengths are shown as functions of the distribution of tendons over the slab width. The results shown are only those for the normal (type 1) slabs. Both the FIP method and the direct decompression method DD1 predict that strength should increase as the ratio bt/b of the width of the tendon band to the slab width decreases. The
test results show a trend for Vu/VFIP and Vu/VDD1 to increase with increasing bt/b. The second direct decompression method DD2 predicts no inuence from bt/b and the ratios Vu/VDD2 seem to be independent of bt/b. This suggests that the redistributions of forces occurring after cracking make it unnecessary to consider elastic concentrations of stresses. The effect of the support size, relative to the slab effective depth, is shown in Figure 11, again only for type 1 slabs. There is a considerable scatter. However if attention is focused on the tests by Silva, which were the only ones in which support size was varied with other parameters held constant, it can be seen that the
lowest ratios of experimental to calculated strengths for all three methods were for the slabs with smallest supports (c/d 0.9d). For the FIP method, the results from Kordina and Nolting fall conspicuously below the general trend. The main difference between these and the other slabs is their high values of Vp. The inuence of Vp, which is a separate component of resistance only in the FIP approach, is shown in Figure 12. With the exceptions of Silvas results for slabs on very small supports, and Correas, for which VRk,c was high in relation to the prestress, the data show Vu/VFIP decreasing as Vp increases,
Structural Concrete
2005 6 No 1
15
Table 1 Data for test slabs Slab Support size: mm h: mm fcm: MPa 37.8 37.8 37.8 37.8 40.1 40.1 40.1 40.1 41.6 41.6 41.6 41.6 44.1 44.1 44.1 44.1 50.7 52.4 52.4 50.7 50.7 52.4 33.6 36.0 36.0 36.8 30.4 31.2 35.2 31.0 28.7 43.8 41.1 43.2 36.5 41.7 40.9 42.5 26.6 51.9 d: mm Silva A1 A2 A3 A4 B1 B2 B3 B4 C1 C2 C3 C4 D1 D2 D3 D4 LP1 LP2 LP3 LP4 LP5 LP6 V1 V2 V3 V5 V6 V7 V8 A1 A2 B2 B3 B4 SP1 SP4 SP5 SP6 M1 M4 100 sq 100 sq 100 sq 100 sq 200 sq 200 sq 200 sq 200 sq 300 sq 300 sq 300 sq 300 sq 200 sq 200 sq 200 sq 300 sq 150 sq 150 sq 150 sq 150 sq 150 sq 150 sq 125 127 128 129 124 124 124 124 126 122 124 123 124 123 125 125 130 130 130 130 130 130 150 150 150 150 150 150 150 180 180 220 220 220 175 175 175 175 160 160 109 113 109 104 114 110 108 106 111 105 106 102 100 106 103 111 0.62 0.47 0.62 0.51 0.60 0.48 0.63 0.50 0.61 0.50 0.64 0.52 0.68 0.50 0.51 0.48 91 97 86 86 98 94 90 89 94 89 90 85 83 90 90 95 65 65 81 81 81 113.5 113.5 113.5 75 113.5 113.5 151 144 110 191 190 135 135 135 135 120 3.31 2.14 3.16 1.98 3.39 2.23 3.12 2.16 3.33 2.26 3.48 2.31 3.34 2.23 2.27 2.22 0 2.19 4.28 0.80 1.33 1.76 1.70 1.66 3.09 0 1.77 1.77 1.77 2.79 2.74 2.12 2.21 1.99 3.94 4.81 3.28 3.50 0 1.95 0.30 0.20 0.75 0.50 0.30 0.20 0.45 0.30 0.30 0.20 0.45 0.30 0.40 0.30 0.50 0.20 0.675 0.625 0.07 0.14 0.21 0.56 0.56 0.56 0.56 0.56 0.56 0.23 0.55 0.23 0.23 0.61 0.76 0.71 0.85 0.85 0.85 10 11 0 0 33 31 13 13 42 37 19 16 11 13 0 41 0 0 0 8 14 12 80 74 140 0 0 82 85 46 0 0 66 0 45 47 22 29 0 19 1 1 1 1 1 1 1 1 1 1 1 1 2 2 2 1 4 3 3 1 1 1 1 1 1 4 3 1 1 1 2 3 1 2 1 1 1 1 4 1
r1: %
dp: mm
scp: MPa
bt/b
Vp: kN
Type
f 200 f 200 f 200 f 200 f 200 f 200 f 200 f 250 f 250 f 250 f 250 f 250
150 sq 150 sq 150 sq 150 sq 180 sq 180 sq
Correa 105 1.17 105 1.17 105 1.17 105 1.17 105 1.17 105 1.17 Kordina and Nolting 128 0.62 126 0.90 128 0.62 126 0.90 128 0.62 128 0.62 128 0.62 Hassanzadeh 150 0.18 150 0.18 190 0.29 190 0.29 190 0.29 Shehata 157 0.27 157 0.27 157 0.27 157 0.27 Melges 127 1.58 134 0.92
Note: type 1, inclined tendons, single or double bands; type 2, inclined tendons, single band with central gap; type 3, horizontal tendons at mid-height; type 4, ordinary rc slabs.
suggesting that no separate treatment of Vp is necessary. It seems fairly clear that the added complexity of the FIP method is not justied by superior accuracy as compared with the simpler direct decompression approach. Indeed its separate treatment of Vp and its use of a non-constant
Structural Concrete
term for decompression seem to reduce its accuracy. The method also has the problem of introducing a difference between prestressed and ordinary rc slabs in terms of mean ratios of experimental and calculated strengths. The direct decompression method DD2, based on moments averaged over the slab
breadth, gives the best predictions of characteristic resistance, understood as the strength of a slab in which the concrete strength is equal to fck. The mean value of Vu/VDD2 is 1.07 and thus provides a modest margin of safety, probably arising from the use of initial values for scp.
2005 6 No 1
16
056
06
slabs: A1, A2, A3 and A4 (Silva) m/V = 01225 slabs: B1, B2, B3, B4, D1, D2 and D3 (Silva) m/V = 0110 slabs: C1, C2, C3, C4 and D4 (Silva) m/V = 00975 slabs: LP1, LP2, LP3, LP4, LP5 and LP6 (Corra) m/V = 0116
28 025
0.62 0 .62
1. 68 2 . 30
0.25
2.8
122
043
020
183
216 085
slabs: V1, V2, V3, V5, V6, V7 and V8 (Kordina) m/V = 0125
018
25
21
235
This explanation is consistent with the increase in the safety margin with increasing size of support, since in Silvas slabs with small supports the increments of tendon forces during loading were minimal (due to the lower ultimate loads). The effect of model uncertainties can be seen in the 10.7% coefcient of variation of Vu/VDD2. For the part of VRk provided by VRk,c, the normal partial safety factor gm must account for deviations of concrete strength below fck and for model uncertainties, as it must for ordinary reinforced concrete slabs. The inuence of variations of fck is reduced by the cube root in equation (4) and a factor of about 1.3 is left for model uncertainties.
015
In order for the level of safety to be consistent in both parts of the punching resistance, a similar factor must be applied to the resistance due to prestress. A value of gp 1.25 applied to characteristic prestress has previously been used in calculations of shear resistance and would probably be adequate here. Thus the design resistance corresponding to equation (7) would be 1 1 V (m0po m0pe ) (11) VRk,c gm gp m0
VRd
The comparisons above relate entirely to slabs with unbonded tendons and there appear to be no comparable data available for slabs with bonded tendons. Some results are available5,16,17 from tests of centrally supported slabs loaded at two opposite edges to simulate conditions in slab bridges with column supports. These slabs had bonded proled tendons in the longitudinal direction, but only non-prestressed bars in the transverse direction. The experimental punching strengths can be compared with the predictions of the direct decompression methods, provided that two points are taken into account.
0 0 with mpo mpe calculated from characteristic prestressing forces gm 1.5 and gp 1.25.
184
24
Structural Concrete
2005 6 No 1
17
Table 2 Results of tests and comparisons Slab Vu: kN 380.0 315.0 352.7 321.0 582.5 488.0 519.8 458.8 720.0 556.7 636.6 497.1 497.1 385.2 395.2 531.5 327.0 355.0 415.0 390.0 475.0 437.0 450.0 525.0 570.0 349.5 375.0 475.0 518.0 668.0 564.0 827.0 1113.0 952.0 988.0 884.0 780.0 728.0 442.0 773.0 Vu/VRk,c Silva A1 A2 A3 A4 B1 B2 B3 B4 C1 C2 C3 C4 D1 D2 D3 D4 LP1 LP2 LP3 LP4 LP5 LP6 V1 V2 V3 V5 V6 V7 V8 A1 A2 B2 B3 B4 SP1 SP4 SP5 SP6 M1B M4 1.62 1.40 1.51 1.57 1.90 1.79 1.79 1.75 2.02 1.78 1.86 1.63 1.79 1.42 1.50 1.59 Correa 0.97 1.04 1.22 1.16 1.41 1.28 Kordina and Nolting 1.41 1.45 1.75 0.96 1.22 1.53 1.60 Hassanzadeh 2.42 2.09 1.62 2.23 1.87 Shehata 2.98 2.55 2.26 2.09 Melges 1.06 1.66 1.06 1.45 1.06 1.28 1.06 1.20 1.86 1.47 1.60 1.43 1.53 1.18 1.33 1.19 1.28 0.99 1.13 1.00 1.36 1.35 1.32 1.50 1.46 1.09 1.06 1.32 1.23 1.15 1.03 0.94 1.27 1.16 1.04 1.04 1.12 1.08 0.96 1.08 1.11 1.16 1.05 1.13 1.09 0.96 1.08 1.12 1.18 0.98 1.06 0.99 0.96 1.03 1.04 1.10 0.97 0.94 1.01 1.08 1.26 1.13 0.97 0.94 1.01 1.05 1.21 1.07 0.97 0.90 0.93 1.05 1.19 1.04 1.20 1.12 1.21 1.33 1.39 1.39 1.40 1.43 1.48 1.39 1.42 1.33 1.39 1.17 1.27 1.23 0.99 0.93 1.05 1.16 1.16 1.22 1.19 1.25 1.29 1.25 1.21 1.19 1.21 1.02 1.08 1.12 0.94 0.91 0.93 1.08 1.10 1.18 1.11 1.20 1.22 1.21 1.12 1.14 1.13 0.98 1.01 1.08 Vu/VFIP Vu/VDD1 Vu/VDD2
(a) The decompression resistance should be averaged for the two directions. With a zero value for the transverse direction, and square or circular supports, this simply means that the resistance is taken as half the value calculated from the longitudinal moments. (b) The upper limit of shear resistance should be applied separately to the two direc-
tions. For the critical longitudinal direction, VRk,c (longitudinal) V (decompression) should be limited to the product of the stress 0.3fck(1 2 fck/250) of equation (10), uo/2 and the effective depth for the combined longitudinal reinforcement. This is necessary in view of the very uneven distribution of resistance. It may be convenient to calculate VRk,c as the
sum of longitudinal and the transverse components determined from the relevant r and d values and u1/2. On this basis the means of Vu/VDD2 for three of the four slab geometries tested are in the range from 1.05 to 1.16 with coefcients of variation of 7.6% or less and compare well with the results in Table 3. For the fourth
Structural Concrete
2005 6 No 1
18
Table 3 Mean values and coefcients of variation of Vu/Vcalculated Type of slab No. of results Average Inclined tendons, banded (type 1) Inclined tendons, band with gap (type 2) Horizontal tendons at mid-height (type 3) Ordinary rc slabs (type 4) All post-tensioned slabs (types: 1, 2, 3 and 4) 28 5 4 3 40 1.32 1.33 1.09 1.00 1.27 Vu/VFIP c. var. 14.2% 8.3% 15.2% 5.5% 15.5% Average 1.17 1.10 1.09 1.00 1.14 Vu/VDD1 c. var. 10.0% 7.0% 15.3% 5.5% 10.7% Average 1.09 1.02 1.03 1.00 1.07 Vu/VDD2 c. var. 8.8% 7.3% 16.4% 5.5% 9.6%
Solid symbols - slabs type 1 with inclined tendons in one or two bands Open symbols with points - slabs type 2 with inclined tendons in single bands with central gaps Open symbols without points - slabs type 3 with horizontal tendons at mid height 20
15
15
Vu /VDD1
Vu /VFIP
10
10
05
05
00 10 15 20 Vu /VRk,c 25 30
00 10 15 20 Vu /VRk,c 25 30
20
15
Vu /VDD2
10
05
00 10 15 20 Vu /VRk,c 25 30
Structural Concrete
2005 6 No 1
19
20
20
15
15
Vu /VDD2
Vu /VFIP
10
10
05
05
00 000
025
050 bt/b
075
100
00 000
025
050 bt/b
075
100
20
15
Vu /VDD1
10
05
00 000
025
050 bt/b
075
100
4 Figure 10 Inuence of tendon banding on the ratios of experimental and calculated strengths
type, Regans series DT, the mean is 0.94 and the coefcient of variation is 5.9%. The low mean for this type is probably due to the small size of the supports, uo/d 3.2 or c/d 0.8, together with the adverse effect of very unequal bending in the longitudinal and transverse directions. The ability to take account of bonded tendons in calculating r can be important. For example for the slabs of Nylander et al.16 where the longitudinal reinforcement ratio would otherwise be zero. However, if a reasonable minimum amount of non-prestressed reinforcement is present, the effect of bonding tendons is often small as the increase of r is largely offset by the reduction of d. It seems that the approach given for bonded tendons should be generally satisfac-
tory at least for oor slabs (and most normal bridge slabs).
by any elastic method, including equivalent frame analysis. Non-elastic moments, such as moments reduced by redistribution, should probably not be used in calculations for punching, as the additional slab rotations are likely to be harmful to shear resistance. The distributions of these moments may be given by an analysis, but it is probably satisfactory to use code-of-practice simplications, which give moments for the inner values of column strips, to dene m0 within (c 4d ). The moments due to prestress can similarly be found by analysis, but for load balancing arrangements of tendons the distributions of moments are the same as for the other loading and ae a. The results for other arrangements may differ signicantly from this. For example, an arrangement with the
Structural Concrete
2005 6 No 1
20
20
20
15
15
Vu /VFIP
10
Vu /VDD1
0 1 uo/4d 2 3
10
05
05
00
20
15
Vu /VDD2
10
05
00 0 1 uo/4d 2 3
4 Figure 11 Inuence of support size on the ratios of calculated and experimental strengths
tendons in both directions highly concentrated to the column gives ae . a and is favourable in both the FIP approach and the direct decompression method1 (DD1). The effects on punching of differences between the spans or the prestress in two directions can, within reason, be treated by averaging the moments, but the possibility of wide-beam shear failure should be considered. For rectangular columns, the FIP recommendations use a weighted average of decompression effects. This does not seem unreasonable, but, so far as the authors are aware has not been validated by any experimental work.
Conclusions
All three methods involving decompression provide reasonable comparisons with the
results of tests of post-tensioned slabs with unbonded tendons. The ratios of experimental strengths to resistances calculated by the FIP method have the highest mean values and the greatest variability. This appears to be primarily the result of the methods inclusion of a separate term for the force Vp, transferred directly from the tendons to the support, and of its reduction of the other components of resistance from prestress at loads above the decompression level. The results from the two variants of the direct decompression method are similar to one another in terms of coefcients of variation. The mean ratios of experimental to calculated strengths are a little higher for the method DD1, based on moments in widths (c 4d), than for DD2, which uses moments averaged across the widths of test slabs. The comparisons are limited to tests of models representing the negative moment
regions around columns due to the lack of data from complete slabs with realistic thicknesses. In principle the methods considered can all be applied to post-tensioned slabs with bonded tendons, by including the tendons in the calculations for r1 and d. However the only test data available for bonded posttensioned slabs with orthogonal arrangements of reinforcement appear to be from tests of models of bridge slabs spanning predominantly in one direction. The comparison with results for slabs with bonded tendons is limited to data for models of bridge slabs prestressed in only one direction.
References
1. ACI 318-02. Building Code Requirements for Structural Concrete. American Concrete Institute, Farmington Hills. Michigan, 2000.
Structural Concrete
2005 6 No 1
21
15
10
05
00 00
01 Vp /Vu
02
03
10. CEB-FIP Model Code, 1990. Thomas Telford, London, 1993. 11. Correa, G. S. Puncionamento em lajes cogumelo protendidas com cabos nao aderentes. MSc dissertation, Dept. of Civil and Environmental Engineering, University of Braslia, 2001. 12. Kordina, K. and Nolting, D. Versuche Zum Durchstanzen ohne Verbund vorgespannter Flachdecken. Lehrstuhl fur Massivbau, Technische Universitat, Braunschweig, 1984. 13. Hassanzadeh, G. Betong plattor pa pelare, Dimensioneringsmetoder for plattor med icke vidhaftande spannarmering. TRITA-BKN Bulletin 43, Institutionen for Byggkonstruktion, Kungl Tekniska Hogskolan, Stockholm, 1998. 14. Melges, J. L. P. Analise experimental da puncao em lajes de concreto armado e protendido. Dr. Eng. Thesis, Escola de Engenharia de Sao Carlos, Universidade de Sao Paulo, 2000. 15. Shehata, I. A. Punching of Prestressed and Non-prestressed Reinforced Concrete Flat Slab. MPhil thesis, Polytechnic of Central London, 1982. 16. Nylander, H., Kinnunen, S. and Ingvarsson, H. Genomstransning av pelarunderstodd plattbro av betong med spand och ospand armering. Neddelande nr 123, Institutionen for Byggnadsstatik, Kungl. Tekniska Hogskolan, Stockholm, 1977. 17. Kinnunen, S., Nylander, H. and Tolf, P. Plattjocklekens inverkan pa betongplattors hallfasthet vid genomstansning. Forsok med rektangulara plattor, Meddelande nr 137, Institutionen for Byggnadsstatik, Kungl. Tekniska Hogskolan, Stockholm, 1980.
Vu /VFIP
Structural Concrete
2005 6 No 1