Development and testing of a bridge weigh-in-motion method considering nonconstant vehicle speed

Bridge weigh-in-motion (BWIM) provides an alternative to conventional static weigh station for obtaining vehicle axle weights. Existing BWIM algorithms assume the vehicles being measured are traveling at a constant speed. This is a reasonable assumption for short-span highway bridges, but will yield large error for railroad bridges and bridges subjected to speed varying traffic. This paper presents a static SWIM methodology to improve the estimation accuracy on axle weights when measuring a vehicle traveling at nonconstant speed. A novel speed correction procedure is proposed and incorporated into the BWIM methodology by converting variable speed response data with nonuniform spatial spacing to uniform spacing. This new method can lead to an increased BWIM accuracy for bridges subjected to speed varying traffic. The BWIM methodology proposed in this paper is based on static analysis, and consists of the following components: Speed Detection, Lateral Load Distribution Determination, Influence Line Determination, Speed Correction, and Axle Weight Determination. In this paper, both numerical study and field study have been conducted. The numerical study is performed using finite element analysis to evaluate the performance of the BWIM results when measuring loads traveling at constant speed and variable speed. The results of the numerical study demonstrate the need of applying speed correction for BWIM analysis, and show the proposed speed correction procedure is able to improve BWIM accuracy for a variable speed vehicle to the similar level of accuracy as a constant speed vehicle. Field study is performed on a steel truss bridge located at the University of Alabama campus. With the obtained data on the bridge, the proposed BWIM methodology is applied to determine the axle weights of a vehicle moving at nonconstant speed. Results of the field study show that large error in axle weight prediction is expected when the vehicle is moving at nonconstant speed on the bridge, and correcting the response data using the proposed procedure can significantly improve the accuracy of axle weight predictions. (C) 2017 Elsevier Ltd. All rights reserved.