Vibration-Based Damage Identification of Masonry Walls Using Distinct Element Modeling

Dry-joint masonry structures are one of the oldest building techniques from ancient and historical masonry buildings. This method used in building of historical structures that are highly vulnerable today. Also in many masonry structures, mortar strength is affected strongly by duration of time and corrosion, so the structure behavior is more likely dependent on the dry-joint characteristics. To assess the existing damages of masonry walls, non-destructive dynamic-based methods are attractive tools as they are able to capture the global structural behavior. In micro-modeling method of this paper, masonry walls are represented by Distinct Element Method (DEM) as assemblies of units consist of block and mortar, which represent an idealization of their discontinuous nature governing their nonlinear mechanical behavior. Due to the heterogeneity and the complexity of the interface’s behavior between blocks and mortar, DEM seems to be the best-adapted to model this kind of structures, in particular for reproducing complex nonlinear post-elastic behavior. At the first step, micro-modeling strategy is used for masonry walls by DEM, and particularly post-elastic behavior is verified with valid experimental data. However, DEM does not directly obtain natural frequencies and mode shapes of the wall via a classic vibrational analysis. Therefore, the second objective of this study is to propose a technique to indirectly identify dynamic characteristics of masonry walls using DEM. The aim of the part is to check the capability of dynamic identification procedures, in the extraction of the dynamic characteristics of the masonry wall in the used DEM software. For this purpose, the dynamic behavior at low vibration levels of an existing masonry building subjected to forced hammer impact test, was investigated. By transforming data collected from dynamic response of the wall, from the time domain to the frequency domain, using Fast Fourier Transform (FFT), we can find natural frequencies from Fourier amplitude spectrum. The proposed technique is then validated by comparison with the results of modal analysis which was carried out using Finite Element Method (FEM). The dynamic characteristics of walls (i.e., natural frequencies and mode shapes) may change when different levels of damage are induced in the wall. The proper knowledge of these variations is a key issue in order to study the seismic demand and seismic performance of structures. Aiming at finding adequate correspondence between dynamic behavior and internal crack growth, several numerical simulations are performed, progressive damage is induced in the wall, and sequential structural frequency identification analysis is then performed at each damage stage. In this paper, frequency and drift are selected as dynamic behavior and crack growth indices, respectively. Quantifying the relative frequency drop shows, despite the shape does not vary significantly with increasing damage, there is a relation between frequency drop and damage variations, based on analyzed data. These properties are firstly modified in the elastic range, and then is developed in the inelastic range with increasing damages. It is also observed that while the failure mode of the wall is diagonal cracking, the in-plain vibration mode shapes are much affected by initiation of crack. On the other hand, modal properties of out-of-plane mode shapes undergoes fewer effects by the diagonal crack.