Influence of soil characteristics on natural frequency-based bridge scour detection

The concept of detecting scour severity by analyzing the change in the Predominant Natural Frequency (PNF) of a bridge pier has been gaining increasing interest in recent years. Previous studies primarily focus on this topic using less cohesive soils such as highly erodible sands, whereas no discussions have been reported on the influence of soil characteristics, especially those of cohesive soils, on the PNF. This missing knowledge gap is critical for this application as cohesive soils are an essential part of the soil-pier interaction to determine ME's. This study aims to fill this knowledge gap by investigating three issues that are related to soil characteristics: 1) the effect of soil types on the PNF variation; 2) the questionable issue regarding the pier diameter effect for soil-pier dynamic modeling using the Vesic analytical expression; and 3) contradictory statements in the existing studies regarding the influence of the soil's elastic modulus on the PNF variation. For the purpose, a series of lab-scale tests is first conducted, and a Winkler-based numerical model is then developed and validated against the experimental results to investigate the effect of soil characteristics on the PNE's measured from systems with cohesive soils and those with less cohesive soils. We found that the soil characteristics affect the PNF by providing a different lateral stiffness to the soil-pier interaction. The strength of the soil-pier interaction mainly depends on the lateral stiffness of each type of soils. In-depth discussions are also made to clarify the pier diameter effect on the predicted PNE's from both less cohesive and cohesive soils. It was clarified that the distribution of the soil's elastic modulus determines whether the pier diameter effect needs to be considered using the Vesic analytical expression. Further simulations are finally conducted with more complex and realistic field soil conditions to mediate the contradictory statements regarding the influence of the soil's elastic modulus. The simulation results indicated that the soil's elastic modulus significantly influences the ME's. The PNF variations differ under different elastic moduli of soils and distributions of elastic moduli with soil depths. (C) 2019 Elsevier Ltd. All rights reserved.