Seismic Risk Assessment of Non-Structural Components in Hazardous Facilities Through a Novel ANN-Based Technique

Seismic events pose a significant threat to industrial facilities, and the risk assessment of non-structural components (NSCs) within these structures is paramount for ensuring the safety and functionality of critical infrastructure. This paper presents a novel approach utilizing machine learning techniques to enhance the seismic risk assessment of NSCs in industrial facilities. The proposed methodology integrates data from multiple sources, including seismic records, structural characteristics, and NSC vulnerability parameters, to develop predictive models for evaluating the vulnerability and potential damage to non-structural components during seismic events. The study starts from a dataset generated by using a numerical effective model subjected to a set of natural records. The advantages of using an efficient model that manages to be both reliable and computationally efficient are highlighted in this paper. With this in mind, the steps to be followed to create an efficient numerical model are herein illustrated. Subsequently, an artificial neural network machine learning algorithm is adopted for training and evaluation. The latter is used for predicting the likelihood of damage to NSCs based on factors such as the intensity and duration of ground motion, the NSC's location within the structure, and its inherent vulnerability characteristics. Results are compared with traditional methods. The outcomes indicate the effectiveness of machine learning in improving the accuracy and efficiency of seismic risk assessment for NSCs in industrial plant. The research contributes to the field of seismic risk assessment by demonstrating the potential of machine learning combined with an efficient numerical model in providing more accurate and timely predictions for the vulnerability of nonstructural components, thereby aiding in the development of targeted mitigation strategies and emergency response plans. This paper serves as a foundational step towards a data-driven approach to seismic risk assessment for non-structural industrial components, ultimately reducing the economic and human losses associated with seismic events.