Discussion. A Theoretical Study of the Sequence of Failure in Precast Panel Shear Walls.

Abstract

The Authors' work is in many ways similar to the work carried out at the universities of Birmingham and Aston in Birmingham on crack propagation through stabilized soil and infill panels.a Like all work on non-linear materials, such as concrete, the Authors' calculation on separation of crack propagation is simplified to the extent that it is doubtful whether it will agree well with practical cases. 43.I would like to suggest that the word 'failure' should not be used in place of the term 'physical separation of the panels', as failure means the loss of stiffness of the entire structure.Mathematically failure occurs when the determinant of the overall stiffness matrix of the entire structure reduces to zero, which is not the case when two elements separate. The Authors do not mention this fact. Do they carry out a stability test of this kind each time they introduce a separation into the structure?If not, what criterion do they use to detect the final state of collapse?44. In their calculation of the state of failure due to shear, the Authors introduce the constant K which depends on the location of the point where failure occurs, the geometry of the wall and so on. However, they do not indicate how K is estimated, nor do they point out the empirical nature of this constant.45. In the finite element analysis of the walls, the Authors assume that separation does not occur within a panel. This assumption is not necessary and the method can be extended to cater for cases where fracture of a panel may take place. This assumption was perhaps adopted so that large, panel-size elements could be used in the analysis. This indeed reduces the accuracy of the analysis. 46. When separation takes place between two elements, the Authors renumber the nodes. This is unnecessary. At the University of Aston in Birmingham a double or multiple joint numbering system has been adopted. For instance, the model shown in Fig. 7(a) is numbered so that a joint, where a crack can take place, is defined by two node numbers. In this model nodes 5 and 9 are initially the active nodes to which elements 1 , 2 , 3 and 4 are connected. However, nodes 6 and 10 are dummy nodes with no degrees of freedom. Before separation takes place the overall stiffness matrix has no rows or columns for these dummy nodes and they in no way influence the analysis. When a crack takes place between nodes 5 and 6, for instance ( Fig. 7(b)), node 6 is rendered active. Elements 1 and 3 are then connected to node 5 and elements 2 and 4 are connected to node 6. 47. When considering tensile failure of the vertical joint, the Authors apply equal and opposite holding forces representing the strength of the yielding steel. A more elegant approach would be one similar to that previously a d~p t e d~-~ in the elasticplastic analysis of frames. The stress-strain relationship of the yielding steel is simply taken to be elastic-plastic. Then, wherever the steel yields, its yield force is included in the overall stiffness equations of the structure. 48.By taking panel-size rectang...