Evaluation of Widely used Hydroplaning Risk Prediction Methods using Florida's past Crash Data

Abstract

Reliable tools for predicting hydroplaning risk in wet weather are essential for effective highway drainage designs. For this purpose, a variety of mathematical models have been developed by numerous highway agencies in the past half century. Models that can predict the thickness of water film that develops on roadways and then the corresponding threshold hydroplaning speed can be classified into three categories: empirical, analytical, and numerical. In this research, the most commonly used methods in all three categories were evaluated by use of hydroplaned crash incidents recorded on the Interstate system in Florida from 2006 to 2011. The prediction accuracy of combinations of each of these models was evaluated and summarized. The sensitivity and the consistency of errors of each model combination were also evaluated. The evaluations were used to develop a procedure for predicting the accuracy and reliability of threshold hydroplaning speed prediction for each model combination. A computer model with a graphical user interface was developed for determining the risk of hydroplaning on a given roadway section under known wet weather conditions.