Skip to main navigation Skip to search Skip to main content

Determination of optimal yield line patterns governing the collapse of slabs

  • Appapillai Thavalingam

Student thesis: Doctoral ThesisDoctor of Philosophy

Abstract

he present study begins with an examination of traditional approaches to the yield line analysis for the computation of the collapse load of slabs and comparisons are made with respect to their computational efficiencies and ease of automation. The Munro and Da Fonseca method which uses triangulated mesh for the slab is extended in this thesis by adopting a geometrical optimisation technique to investigate the effect of adjustment of the geometry. Node position variables are introduced in this process in order to use the LP formulation for a stationary mesh as a non-stationary acceleration optimising function.

However, it is found that the use of general optimisation techniques to adjust the yield line patterns may run into convergence problems due to the presence of discontinuities in slope of the optimising function. The nature of these discontinuities is discussed briefly and is compared with the degeneracy of the LP solution. As a result of this investigation, a general semi-automatic computer method is described for the computation of optimal yield lines and an algorithm for this is presented. Linearised constraint equations are developed analytically and used with the conjugate gradient method to perform the mesh adjustment process efficiently. Two alternative procedures are presented for the gradient evaluations. Situations in which multiple discontinuities in the slope of the optimising function occur are studied and an additional technique is devised to identify and construct only the effective additional constraint equations required to conduct the search process efficiently. Computation procedures are given and the applications of this method to different types of slab models is illustrated with the
use of numerical examples.
Date of AwardDec 1995
Original languageEnglish
Awarding Institution
  • Queen's University Belfast
SupervisorAlan Jennings (Supervisor)

Cite this

'