Seismic hazard analysis and financial impact assessment of railway infrastructure in the US West Coast: A machine learning approach

Abstract

This research examines the seismic hazard impact on railway infrastructure along the U.S. West Coast (Washington, Oregon and California), using machine learning to explore how measures of seismic hazard such as fault density, earthquake frequency, and ground shaking relate to railway infrastructure accidents. By comparing linear and non-linear models, it finds non-linear approaches superior, particularly noting that higher fault densities and stronger peak ground shaking correlate with increased infrastructure accident rates. Shallow earthquakes with magnitudes of 3.5 or greater and hypocentral depths <20 km also exhibit a pronounced correlation with the incidence of railway infrastructure accidents The study extends to financial impact analysis through Net Present Value and Monte Carlo Simulation, and evaluates damage costs from 2000–2023 to guide financial planning and risk management strategies. It highlights the crucial role of advanced financial tools in optimizing maintenance and long-term planning that could result in better preparedness in high seismic hazard regions and emphasizes the need for robust risk management strategies in enhancing railway operational safety that considers the local and regional tectonic and seismic activity and local ground shaking intensity.