Accuracy of Experimental Testing

Due to the complexity of the punching shear problem, any analytical model has to be based on, or verified with experimental test results to a certain extent. Experimental testing of punching shear presents numerous problems.

Testing of real structures is not feasible due to the tremendous costs involved and the large scale of such a test setup. The only practical option is to test representative parts of the structure either at full scale or scaled down.

Most of the punching shear tests undertaken to date were done using single column tests with little attention given to the boundary conditions of the slab portion used. The validity of this approach is explained in the following sections.

4.1.1. Single Column Tests

In most cases the dimensions of isolated slab-column are selected to coincide with the lines of contra flexure in the real structure. In other words, the region of negative moments in the real slab is used for a single slab- column test.

These tests are relatively inexpensive and allow full scale testing. However, there are a number of disadvantages in using a single column setup:

 Simulation of real boundary conditions is ignored  Confinement of the concrete is ignored

 Membrane forces in the slab is not present

 Failure shear stresses are not influenced by the size effect (The size effect causes a reduction of shear strength with increasing slab depth)

 Load redistribution is not possible

Two different configurations of a single column test are possible. The two setups should not be perceived as similar. Firstly the slab can be supported on its boundary with the load applied on the column. This setup allows the forces to distribute along the boundary. Secondly the slab specimen can be supported on the column with the load applied at a fixed distance from the column.

4.1.1.1. The effect of boundary conditions

Elstner & Hognestad (1956) tested the effect of different boundary conditions in 1956. They set up three scenarios. Firstly, a square slab with continuous simple supports along all four edges. Secondly, a similar slab with continuous simple supports on two opposing edges and lastly the four corners were simply supported.

A linear elastic finite element analysis on the three scenarios renders similar shear stress distributions for all three cases. However, the actual testing reveals a decrease in the punching shear capacity with decreasing support provided to the slab boundary.

If the failure loads are normalized with respect to '

c

f , the capacities

reduce from 100% for a slab with all four edges supported, to 85% for two opposing edges supported, to 60.4% for a slab with corner

supports only. It seems that boundary forces develop in slabs

supported on all edges. These forces enhance the shear capacity of the slab-column connection.

It should be noted that the effect of the moment to shear ratio is included in these tests. This ratio will be the highest for the slab supported on all four edges and consequently a higher punching shear capacity can be expected.

Alexander and Simmonds (1992) reported similar results by using three test scenarios providing rotational restraints with rollers. Firstly they used a slab with rotations of the corners and edges restrained, secondly, rotations of the edges alone restrained and thirdly rotations of the corners alone restrained. The normalized capacities of these tests are 100%, 89.7% and 82.1% respectively. Evidently increasing rotational restraint enhances the punching capacity of the connection.

The punching shear strength of a slab is also influenced by the shear span ratio (av d), where av is the shear distance, i.e. the distance

from the loaded area to the support, and d the effective slab depth. Although test data on the influence of the shear span ratio is rather limited, it is safe to say that shear strength rises significantly for ratios less than 1.5, but remains fairly constant for higher ratios – fib (2001). Thus if the supports are too close to the applied load, they interfere with the results. According to the fib (2001) it is reckoned that a distance of at least three slab thicknesses is necessary between the loaded area and the slab supports.

4.1.1.2. The effect of compressive membrane action

Due to the confinement of the slab-column connection by the adjacent slab it also plays a role in the punching shear capacity of the

connection.

Compressive membrane action is considered as a secondary effect, which occurs after cracking of the concrete and yielding of the reinforcing steel. As the slab fails and deflects, the surrounding concrete restrains the sagging portion of the slab by compressing around it.

It has been found that the punching shear capacity increases if the slab specimen extends beyond the nominal line of contra flexure. Testing by Bond, Long, Masterson and Rankin indicate strength

increases ranging from 30% up to 50% compared to similar single column setups.

Due to numerous reasons their test is thought to overestimate the capacity enhancement. In addition to these, real slabs undergo restrained shrinkage inducing tensile stresses, which in turn reduces the shear capacity of the slab.

For design purposes compressive membrane action should not be used as an enhancing factor for the predicted failure loads.

4.1.2. Slab subsystems

It is believed that a subsystem will render more realistic results than a single column test. Sherif (1996) tested the most realistic subsystem to date.

The slab consisted of a 150mm thick, continuously reinforced slab with realistic boundary conditions along lines of zero shear centred on an exterior column and an adjacent interior column.

Testing of this slab resulted in the conclusion that the punching shear capacity for interior slab-column connections is similar for both single column tests and full slab tests.