Utilizing Kriging Metamodeling to Provide Practical and Effective Bridge Weigh-in-Motion

Abstract

Emerging technologies are a strategic ally in the efficient management and preservation of pavements and bridges. Toward this end, different methodologies, such as weigh-in-motion (WIM), have been developed for the detection of overweight vehicles and traffic monitoring. Bridge weigh-in-motion (BWIM) systems can be installed and serviced without interrupting traffic flow. This paper presents a new approach for processing traditional strain response data. This proposal extends the BWIM concept of influence area from response time-histories. The dimensionality of the strain response waves is reduced by the calculation of the centroid and the area under the curve. Subsequently, these two dimensions are used as inputs to an ordinary Kriging (OK) metamodel to predict the gross vehicle weight (GVW) of the passing traffic. The OK methodology allows for the strategic selection of vehicles to train the BWIM metamodel. An example of the application of Kriging metamodeling (KM) to BWIM through the instrumentation of an in-service highway bridge located in Costa Rica is presented. Experimental horizontal strain data along with corresponding weight measurements from a static permanent weigh station were available for 90 trucks to validate the proposed enhanced BWIM methodology.