Improved Moore-Penrose continuation algorithm for the computation of problems with critical points. (June 2023)

• Record Type: Journal Article
• Authors: Léger, S.; Larocque, P.; LeBlanc, D.
• Is Part Of: Computers & structures. Volume 281(2023)
• Publication Date: 2023-06
• Subjects: Moore-Penrose continuation method -- Deflated continuation -- Angle control -- Finite element method -- Algorithm -- Critical points; Structural engineering -- Data processing -- Periodicals; Electronic data processing -- Structures, Theory of -- Periodicals; 624.171
• DOI: 10.1016/j.compstruc.2023.107009
• Languages: English
• ISSNs: 0045-7949

Abstract

Highlights: This paper presents an improved algorithm for the Moore-Penrose continuation method. Its ability to compute solution curves with critical points is greatly improved. Difficulties observed using the standard approach are not encountered. Easy implementation. Abstract: Using typical solution strategies to compute the solution curve of challenging problems often leads to the break down of the algorithm. To improve the solution process, numerical continuation methods have proved to be a very efficient tool. However, these methods can still lead to undesired results. In particular, near severe limit points and cusps, the solution process frequently encounters one of the following situations: divergence of the algorithm, a change in direction which makes the algorithm backtrack on a part of the solution curve that has already been obtained and omitting important regions of the solution curve by converging to a point that is much farther than the one anticipated. Detecting these situations is not an easy task when solving practical problems since the shape of the solution curve is not known in advance. This paper will therefore present a modified Moore-Penrose continuation method that will include two key aspects to solve challenging problems: detection of problematic regions during the solution process and additional steps to deal with them. The proposed approach can either be used as a basic continuation method or simply activated when difficulties occur. Numerical examples will be presented to show the efficiency of the new approach.