Efficient meta-heuristic mesh adaptation strategies for NURBS upper–bound limit analysis of curved three-dimensional masonry structures. (August 2020)

• Record Type: Journal Article
• Authors: Grillanda, Nicola; Chiozzi, Andrea; Milani, Gabriele; Tralli, Antonio
• Is Part Of: Computers & structures. Volume 236(2020)
• Publication Date: 2020-08
• Subjects: Masonry -- Masonry vaults -- Limit analysis -- Meta-heuristic algorithms -- Genetic algorithm -- Prey predator algorithm -- NURBS; Structural engineering -- Data processing -- Periodicals; Electronic data processing -- Structures, Theory of -- Periodicals; 624.171
• DOI: 10.1016/j.compstruc.2020.106271
• Languages: English
• ISSNs: 0045-7949

Abstract

Highlights: Adaptive NURBS-based limit analysis allows fast assessment of general masonry structures. Several meta-heuristic mesh adaptation schemes are proposed for the analysis of masonry vaults. A new, specifically tailored, Prey Predator Algorithm (PPA) is presented. The high efficiency of the proposed schemes is illustrated by means of numerical examples. Abstract: We propose a new adaptive NURBS upper-bound limit analysis approach for the assessment of general three-dimensional curved masonry structures, based on different meta-heuristic mesh adaptation schemes. The method, which can easily be integrated within CAD modeling environments, allows to establish the actual failure mechanism and load bearing capacity of a masonry structure by iteratively adjusting a tentative pattern of yield lines, defined upon a suitable initial mesh of NURBS rigid elements, by means of a suitable meta-heuristic algorithm which searches for the minimum collapse load multiplier, thus enforcing the upper-bound theorem of limit analysis. In particular, we investigate and discuss the efficiency of several meta-heuristic algorithms in delivering the optimal solution: a specifically devised Prey Predator Algorithm (PPA) is compared with the Particle Swarm Optimization (PSO) Algorithm, the Firefly Algorithm (FA) and a suitable Genetic Algorithm (GA). Four masonry vaults have been chosen as case studies. In particular, the modified PPA proves to be the most efficient mesh-adjustment scheme for the proposed adaptive NURBS-based limit analysis procedure.