Integrated structural health monitoring system and multi-regression models for determining load ratings for complex steel bridges. (November 2015)

• Record Type: Journal Article
• Authors: Seo, Junwon; Czaplewski, Taylor M.; Kimn, Jung-Han; Hatfield, Gary
• Is Part Of: Measurement. Volume 75(2015:Nov.)
• Page Start: 308
• Page End: 319
• Publication Date: 2015-11
• Subjects: Complex steel bridge -- Load rating -- Multi-regression analysis -- Weigh-In-Motion data -- Sensitivity analysis; Weights and measures -- Periodicals; Measurement -- Periodicals; Measurement; Weights and measures; Periodicals; 530.8
• DOI: 10.1016/j.measurement.2015.07.043
• Languages: English
• ISSNs: 0263-2241

Abstract

Abstract : Highlights: The rating framework is based on strain measurements via monitoring system. Weigh-In-Motion (WIM) data are used to characterize actual trucks. Four multi-variable regression models with measured data are developed. The best fit model is determined by comparing all ratings with baseline ones. Sensitivity study is carried out with the best fit model. Abstract: This paper presents a multi-regression-based framework to efficiently and accurately determine load ratings of complex steel bridges. To validate the efficiency and accuracy of the proposed framework, the framework was applied to an in-service steel bridge located in Iowa. A network of strain sensors was placed on critical regions of the bridge to capture its behaviors resulting from actual ambient five-axle trucks. To approximate the trucks, relevant Weigh-In-Motion (WIM) data were secured from two weigh stations nearest the bridge. As part of the load rating calculation process, strain sensors were tested for significance and compared to the WIM data to explore which trucks significantly affect the sensors. Four different regression methods were used with the significant WIM data that were condensed to match the size of conventional rating data obtained from a series of structural analyses of the bridge based upon the AASHTO Manual. The four different truck data sets obtained from the regression methods were then each combined with the conventional data to create the corresponding regression models to predict load ratings. The predicted data were compared to the conventional data to determine the best fit regression model. Using the best fit model, a sensitivity analysis was performed to test which truck characteristic parameters affect the predicted ratings the most. Findings indicated that the framework can build the best fit regression model capable of accurately and rapidly predicting load ratings given unknown truck data. It was also found that each axle weight and the largest axle spacing have a great influence on the predicted ratings, implying that these variables may be holistically considered to ensure structural safety of steel bridges subjected to unknown trucks.