Date of Award


Degree Type


Degree Name

Master of Science (MS)


Civil and Environmental Engineering

First Advisor

Christopher Baxter


All bridges in the United States are required to complete a detailed scour analysis in accordance with the Federal Highway Administration's (FHWA) Hydraulic Engineering Circular Number 18 (HEC-18) standards using predetermined design flood events. Proper scour depth predictions are essential, as over conservative estimates may lead to a bridge being classified as scour critical, leading to a required plan of action and possible costly remediation efforts. Conversely, under conservative estimates could lead to reduced performance of a bridge or even complete failure (Arneson et al. 2012).

In Rhode Island there are 127 bridges that are classified as scour critical, requiring the Rhode Island Department of Transportation (RIDOT) to create a detailed plan of action for each bridge and regularly monitor them. Following historic floods in 2010, an evaluation of the scour critical bridges throughout the state was performed, and this study suggested that the current HEC-18 methodology to evaluate scour is over conservative (AECOM 2013). The experiments used to develop the current scour equations found in HEC-18 do not include detailed site characteristics such as complex hydrology, vegetation, cobbles, or soil cohesion which are all common features in Rhode Island.

The objective of this study is to evaluate the current scour methodology on selected scour critical bridges in Rhode Island. Four bridges were selected: Three riverine bridges and one bridge crossing a tidal inlet. To estimate information such as flow, velocity, and depth, each bridge site must be modeled using the US Army Corps of Engineer’s Hydraulic Engineering Center’s River Analysis System (HEC-RAS). Conditions from the 2010 flooding event were used to model the riverine bridges and Hurricane Sandy in 2012 was used to model the marine bridge. Scour at each site was predicted using the equations presented in HEC-18 and compared with both past and present scour observations. Finally, the sensitivity of multiple parameters within the HEC-18 equations were evaluated to create an upper and lower bound for the scour prediction during the event of interest.

To support the modeling effort, a detailed field testing program was conducted. At each bridge, the local bathymetry and topography was obtained using a combination of interferometric sonar, real-time kinematic, and total station surveying. Information about the bed conditions and existing scour features was obtained from analysis of grab samples, side scan sonar images, and CHIRP sub-bottom profiler. From this data, any existing scour features were noted. Minor scour was present at one of the riverine bridges. Prominent scour features were observed at the tidal bridge. Both features observed in this study have occurred in the past and have been slightly in-filled according to documentation.

Cross sections were created from the survey information to accurately model the bridge sites. Steady, one-dimensional HEC-RAS models of the 2010 flooding event were created at the three riverine sites and the associated scour was estimated. For the tidal site, complex unsteady one-dimensional models were created for Hurricane Sandy and the resulting scour was predicted. Predicted and observed values of scour were compared and it was concluded that the scour prediction equations over estimated scour at all four sites where scour analyses were completed.

A sensitivity analysis identified variations in the predicted scour depths, providing an upper and lower limit with regards to grain size, angle of attack, Froehlich’s length of active flow, and the Manning friction coefficient. It was determined that the selected median grain size plays a significant role on the prediction of scour, varying the estimates by about five feet. The angle of attack significantly affected pier scour, but had minimal effects on abutment scour. Froehlich’s length of active flow had an average change of 2.4 feet on abutment scour at the four bridge sites. Finally, the Manning friction coefficient had the smallest changes with an average change in scour of 1.1 feet. It was easily seen that the scour predictions over-predicted the depth of scour as compared to the observed, indicating that the scour prediction equations overestimate scour features in Rhode Island.