Improving damage detection in masonry bridges: A combination of finite–discrete element method and genetic algorithms

Abstract

Thousands of masonry structures are part of the international railway and road networks, with some remaining in use for up to two centuries. Given increasing operational demands in both traffic weight and speeds, developing effective damage detection techniques becomes essential for their proper maintenance. Currently, damage assessment is performed primarily in situ, involving a high costs. Therefore, a numerical tool based on an inverse problem for damage detection from a series of indirect and/or permanent measurements is necessary. As a first step, this paper presents a numerical approach that combines finite and discrete element methods with genetic algorithms to identify the position of one or two missing blocks in a one–span masonry bridge. Field measurements are replaced with a displacement set from a two–dimensional FemDem mesh considering missing blocks. These numerical distributions are sampled at a limited number of control nodes and are perturbed with statistical noise to simulate the variations in the actual measurements. The results are compared with a small population of new numerical cases generated by the genetic algorithm. With appropriate noise levels, it is demonstrated that the method can find the missing locations automatically and accurately identify the missing locations.