Executive Summary : | India's old bridges, which are heavily used, suffer from various types of damages such as fatigue and corrosion. Fatigue damage is particularly dangerous for steel bridges and is difficult to identify due to their small size. Monitoring traffic loads on in-service bridges is crucial for predicting fatigue damage, efficient maintenance planning, and condition evaluation. The bridge weigh-in-motion (BWIM) system can provide information about vehicles traveling at normal highway speeds without stopping or diverting. BWIM converts the bridge into a weighing scale using two types of sensors: axle detector and weighing sensor. It has a simpler mathematical framework and requires the bridge influence line (IL) as a prerequisite. The axle weights are determined from measured bridge responses by minimizing the sum of squared residuals between the measured and theoretical bridge responses. The existing BWIM methods use an ordinary least-square method to infer unknown vehicle weights, assuming errors to be white noise. However, the measured bridge responses have a dynamic component, known as the observation error, which depends on factors like bridge characteristics, road surface roughness, vehicle types, and suspension systems. This project aims to investigate the nature of observation errors by conducting experiments using models of different bridges and vehicles, and develop an appropriate mathematical framework for modeling them. The BWIM system also suffers from system inaccuracies such as ill-conditioning effects due to closely spaced axles, inaccurate IL, multiple vehicle conditions, and transverse positioning of vehicles. The central aim is to understand these inaccuracies and address them by integrating a probabilistic inverse method into the conventional BWIM algorithm. In India, many bridges belong to local governments that suffer budget deficits. The project aims to make the BWIM system as economical as possible, making practical implementation feasible in India. |