Development of the salt application rate tool for winter road maintenance

Road authorities in cold climates invest significant resources on winter road maintenance to ensure the safety of the commuters. Although road salt application is necessary for public safety, too much salt can have adverse effects on the road infrastructure and the environment. This study developed a new salt application tool through analysis of historic records and using advanced machine learning methods for forecasting the optimum amount of road salt application for a given winter storm event. A dataset of 227 winter storm events covering the period from 2011 to 2020 was selected for three highway salt application routes and analyzed using the Group Method of Data Handling (GMDH) and the Gene Expression Programming (GEP). These routes were selected based on the availability of climatic data and salts application data. The sensitivity of each input variables was analyzed versus target variable derived from the explicit equation for salt application rates. The developed non-linear relationship uses the three climate variables (storm duration, total precipitation, and average air temperature) as input to predict salt application rate. This tool can be integrated with the Road Weather Information System (RWIS) forecasts to provide recommended salt application rates in the next generation of smart salt trucks for each route to help optimize salt application rates for a given winter storm event. In addition to potential cost savings, reducing adverse effects to the environment and highway infrastructure are a bonus.