Experimental results on plastic collapse stresses in flat plates containing partwall flaws have been collated. These are compared with two-dimensional limit load solutions, assuming both free rotation (pin jointing) and rigid restraint. Allowance for finite flaw lengths is made by calculating an effective flaw depth to thickness ratio, based on the area of the flaw divided by an effective sectional area. Few data on embedded flaws were found, and so an experimental program using diffusion bonded tension specimens containing internal slits was undertaken.
The results of this program, coupled with the analysis of experimental data on surface flaws, show that for surface flaws in sections experiencing some restraint, the assumption of rigid restraint is reasonable. For embedded flaws, pin jointing was assumed, rather than rigid restraint, since the equations are much simpler and the difference is smaller. The proposed correction for finite flaw length is shown to give good results, with overall safety factors under tensile loads in the range 1.08 to 1.64 being obtained. For highly work-hardening materials such as austenitic steels, the method overpredicts the collapse loads, due almost certainly to inappropriate estimates of flow strength.