A digital tool based on Genetic Algorithms and Limit Analysis for the seismic assessment of historic masonry buildings

Abstract

New technologies are changing the way engineers work within the construction sector. Newly developed software solutions have provided effective methods to explore the design space at the interface between Structural Engineering and Architecture, allowing more efficient design strategies. These technologies are based on the integration of parametric generation and visualisation of geometries with powerful numerical solvers, employing user-customised routines. While the construction industry is rapidly moving the design of new construction towards a fully digitalised process, the assessment and the analysis of existing structures with such tools are still largely unexplored. In this context, a visual script for the structural assessment of out-of-plane mechanisms in historic masonry structures subject to seismic loading has recently been proposed by the authors. This relies on two successive steps of analysis, which are integrated into a digital work-flow. Datasets describing the geometric configuration of masonry structures are employed to automatically generate a non-linear Finite Element (FE) model and investigate possible collapse modes. A preliminary global analysis is performed using the commercial software ABAQUS CAE. This, in combination with the Control Surface Method (CSM), allows identifying the most likely failure mechanisms which are described by the geometry of the macro blocks. The parametric modelling of the macro-blocks geometry allows exploring the domain of possible solutions using the upper bound method of limit analysis. A Genetic Algorithms (GA) solver is used to refine the geometry of the macro-blocks and search the minimum of the upper-bound load multipliers, which guarantees equilibrium. The script is implemented in the visual programming environment offered by Rhino3D+Grasshopper. In this paper, a set of parametric analyses considering various input variables such as friction coefficient and opening incidence are performed to verify both the sensitivity and the accuracy of the proposed met