Short-Term Prediction of Deflection for a Steel Truss Railway Bridge Induced by Train Load using Seasonal ARIMA: A Case Study at BH 77 Bridge, Lampung, Indonesia

The BH 77 Railway Bridge is one of the railway bridges that withstands the heaviest train load in Indonesia. The Babaranjang train, which crosses this track in South Sumatra, consists of 60 wagons and 2 locomotives, all loaded with coal. This heavy load is considered extreme, especially since the bridge was designed in the 1970s under an older structural design code. Therefore, a structural health monitoring system (SHMS) is essential to prevent a bridge's sudden failure. One of the key advances in Structural Health Monitoring System is predicting structural responses based on historical measurement data. This study focuses on developing a deflection prediction model using seasonal ARIMA. The bridge deflection is a critical parameter in structural health monitoring; thus, two pressure transducer sensors were installed at the end of the bridge near the abutment and reservoirs as a water level references were placed on the midspan of the bridge to measure the actual deflection induced by the train load. Eight days deflection data was analyzed, focusing on the trains loaded with coal (the heaviest train), and assuming a consistent average speed for all trains. The seasonal ARIMA was then performed to analyze the optimal frequency model. Utilizing Mean Absolute Error (MAE) and Root Mean Square Error (RMSE) evaluation techniques, this model is appropriate for short-term deflection prediction models, with results showing an MAE range from 13% to 16% and RMSE 16%-21%. The outcomes of this model's development are particularly promising and provide a general overview of the trainload-induced bridge deflection prediction.