Electrical resistivity as a binary classifier for bridge scour evaluation

Abstract

Scour is responsible for approximately 60% of all bridge collapses in the United States. Evaluating scour using empirical equations has proven inaccurate, particularly in fine-grained soils. Furthermore, evaluating scour potential based on site specific erosion testing can be uneconomic as erosion testing is time consuming. Many soil characteristics that affect soil erosion also influence soil electrical resistivity (ER) which can be measured in the field or in the laboratory. The objectives of this study were to assess if field and laboratory ER measurements can be used interchangeably for characterizing soil erodibility and the impact of saturation as expected during a flood event, on the ER based soil erodibility model. Twenty-one bridge sites were used for in situ electrical resistivity tomography (ERT) surveys and erosion testing. Supplementary ER measurements at varying degrees of saturation were also conducted using a soil box in the laboratory. It was found that ER values do not differ significantly when ERT is conducted on streambanks as opposed to fully saturated conditions. A receiver operating characteristic curve and probability density function (PDF) were used to determine that soils with ER over 50 Ωm had an 87% probability of classifying as highly erodible based on this data set. Statistical tests also suggested that laboratory ER can be used as an alternative to field ERT surveys. While in situ ERT provides more information in terms of spatial variation in the subsurface, laboratory ER measurements on retrieved samples will allow transportation agencies to utilize the developed erodibility classifier model without the capital investment of an ERT system. Given the large number of bridges with scour susceptible foundations in the National Bridge Inventory, ER can be used to rapidly prioritize them for additional testing to measure the erosion potential or to characterize the soil erosion at proposed bridge sites.