A Decision Support System for Empty Hopper Car Management

Chemical manufacturers in the United States use rail hopper cars to store and transport solid products. Customers also use the producer’s hopper cars as storage vessels, holding the car until the contained product is consumed. Uncertainty in customer hold times and transit times makes predicting the return of empty hopper cars to production sites challenging. This prediction difficulty results in increased costs due to the transfer of empty cars to and from external storage yards, and on rare occasions, production curtailment. We develop a data-driven decision support system to improve the management of empty hopper car inventory for a large U.S. chemical manufacturer. The company’s previous approach for the described task provides only reasonable estimates for cars that are in transit to the shipping facility. Our decision support system (DSS) provides a rolling forecast for a 40-day planning horizon, which improves daily and cumulative predictions by 57% and 13%, on average, when compared with the previous method. Ultimately, the DSS provides the company with increased visibility into its supply chain and allows it to make probabilistic business decisions while eliminating several data retrieval and manipulation tasks.