Vibration assessment to diagnose structures from a local network of accelerographs

Abstract

Establishing the needs of a structure is a complex task, especially if required information is not fully available. Hence, experimental analysis and strategies are necessary; since the guides and regulations available for the evaluation of existing structures are limited to the engineering judgment and particularities of each case, which means that they cannot be easily defined or standardized.

Therefore, the objective of this research is focused on validating the efficiency of vibration analysis, to diagnose structural problems, compared with traditional methodologies that incorporate the use of non-destructive techniques in an operational structure. For this purpose, the case study of a residential building is presented, where a system of walls that have persistently cracked was evaluated; even though some of these elements were completely replaced.

Vibration responses were monitored, and the data was used to develop mathematical models, whose error term were used to design damage indicators that, together with Machine Learning techniques, for statistical pattern recognition, it was possible to dynamically characterized the objective system and diagnose that it did not present an anomalous performance that could have influenced the deficient behavior of the masonry systems.