Hybrid physics-based modeling and data-driven method for diagnostics of masonry structures

Abstract

Before implementing monitoring systems or reinforcements on a historic structure, it is essential to understand how crack patterns may have originated and how they affect the stability of the structure. Previous methods combining photogrammetry with physics-based modeling have been successful in diagnosing the cause of crack formation. However, a limitation of existing methods is the manual comparison process to ascertain damage origins. This research outlines a method combining physics-based modeling and data-driven approaches to automate diagnostics for existing masonry structures. This method was shown to quantitatively reproduce the cause of damage for complex, 3D structures and was validated against a laboratory-scale experimental masonry wall. The newly automated procedure increases throughput by 105 times compared to our prior method, allowing for the testing of orders of magnitude more hypotheses than were previously possible. Although the approach is demonstrated here for settlement-induced cracking, it has important implications for the broader topic of data-driven masonry diagnostics.