AI-enhanced monitoring and sustainable control of bridge foundation settlement: A case study from Nepal

Abstract

Bridge foundation settlement and scour pose significant threats to Nepal’s East-West Highway, particularly during monsoon flooding. This paper presents a sustainable, data-driven framework for monitoring and controlling foundation settlement by integrating geotechnical investigation, empirical modeling, and artificial intelligence. The Barahari Bridge was chosen as a case study because of recurrent flood damages. Field work involved borehole drilling, groundwater monitoring, and in-situ soil testing to determine strength and grading properties. An empirically derived depth-damage relationship using 25 flood events showed high correlation (R² = 0.87). A Condition Assessment Scoring System (CASS) was formulated to rank critical components based on multi-attribute weighted scoring. With soil and hydraulic properties, a Random Forest regression model was developed to predict the relationship with R² = 0.81 and MSE = 0.18, indicating that floodwater depth and scour depth were the most dominant variables. Bioengineering, prefabrication, and life cycle assessment were combined to formulate environmentally sustainable
management practices. The findings clearly show that AI-assisted monitoring can offer resilient, economical, and sustainable management of bridge foundations in flood-susceptible regions of Nepal.