Examining Methods for Modeling Road Surface Roughness Effects in Vehicle–Bridge Interaction Models via Physical Testing

Numerically simulating vehicle–bridge interaction (VBI) models within finite-element models (FEMs) has been a topic of interest over the last two decades. Its applications have been well-established in the structural health monitoring community for extracting the dynamic properties of bridges using instrumented vehicles. Often times, analytically generated surface profiles adopted from renowned standards such as the ISO-8608 are used to simulate road surface conditions and approach a more realistic model. However, previous analytical studies have indicated that current methodologies for modeling the effect of road surface roughness on the vehicle response tend to exaggerate the dynamic response of a vehicle and overshadow bridge frequencies in the vehicle response. To alleviate this issue, several studies have recommended relatively low-amplification roughness factors ( \(G_d\) ) or have used a moving average filter (MAF) to de-noise analytically generated surface profiles prior to prescribing it into the FEM, but to the authors’ knowledge no experimental investigation has been conducted to validate these recommendations. In this chapter, a full-scale road test is conducted using a passenger vehicle instrumented with accelerometers. A vehicle model is developed using the bicycle concept to approximate the dynamic response of the tested vehicle while including road surface roughness effects. The results are then compared to observe whether the vehicle model contains higher acceleration amplitude values than those of the experiment while using ISO-8608’s \(G_d\) factors for the appropriate road class observed in the test. It is concluded that low \(G_d\) factors, which are commonly used in VBI studies, seem to underestimate the surface roughness effects and thus produce an unrealistic VBI model. Moreover, the use of higher \(G_d\) factors without the implementation of a MAF tends to significantly exaggerate the vehicle’s response. The implementation of a MAF was successful in attenuating high frequency noise while also reducing the acceleration amplitude of the vehicle, but the \(G_d\) value still needed to be relatively low to represent the data from the road testing. Finally, the study concludes with recommendations on how to improve the current approach for modeling road surface roughness in VBI problems and suggestions for future work. For feedback and support, please see the https://github.com/omarabuodeh1994/surface_roughness.git . No promises are made in terms of timely help from the author.