The $800 billion U.S. truckload industry accounts for about 80% of the nation’s entire freight costs. The industry is expected to grow by 2.3% year on year from 2019 through 2024 and reach a total of 15 billion tons of goods shipped. Since the U.S. labor market is expected to tighten, it is increasingly important to ensure equilibrium between supply and demand in the market by continuing to attract and retain truck drivers and to maximize drivers’ time on the road.
However, the pressure created by the growth in demand results in supply/demand imbalances that strains the relationships between drivers and other players in the industry.
A driver-centric market allows drivers to prioritize routes based on profitability and preference, resulting in higher numbers of load cancellations and rejections on low-density, low-profit routes. Moreover, to increase their potential earnings, drivers may reallocate capacity from long-term contracts to the more lucrative spot market, which drives up prices for shippers.
Conversely, a supplier-centric market can constrict freight rates and put driver retention at risk. As a result, drivers are encouraged to migrate to other industries that offer more competitive compensation.
Focus on productivity
One approach to solving these problems is to maximize the productivity of drivers on the road. Dwell time represents a large component of productivity in that truck drivers spend up to 30% of their work hours loading and unloading goods at shipper facilities.
Our research aims to understand the factors that drive dwell time and to ultimately predict the dwell time of a load in a shipper facility. This knowledge enables us to pinpoint operational practices and policies that cause unplanned dwell time, and provide companies with more granular information that they can use to mitigate dwell time on a per-load basis.
Shipper facilities the key
We utilized data from the leading third-party logistics company that sponsored the research, to help us understand the key factors that influence dwell time and to determine if the dwell time of a load can be predicted. The analysis used a combination of descriptive statistics and machine learning models since each approach has specific tradeoffs between predictive performance and interpretability of results.
Based on this experience, we recommend using a random forest classification model with one-hour bins. This approach provides the highest predictive performance and sufficient interpretability to drive operational action.
We found that individual cargo handling facilities have the biggest impact on dwell time. Out of 50 indicators explored, the historical performance of an individual shipper facility was the most meaningful predictor of dwell time by a wide margin. This result suggests that the stickiness and influence of the operational practices and policies followed at specific facilities are the key to predicting dwell times. However, each facility operates autonomously and has its own operational agenda that is not transparent to its partners in the industry. Therefore, future efforts to reduce dwell time must unpack these practices and policies to gain a more granular view of the leading and lagging practices within each facility.
Aside from historical facility performance, the method drivers use to update their arrival time at a facility is another major driver of dwell time prediction.
Every year, around 80 students in the MIT Center for Transportation & Logistics’ (MIT CTL) Master of Supply Chain Management (SCM) program complete approximately 45 one-year research projects. The students are early-career business professionals from multiple countries, with 2 to 10 years of experience in the industry. Most of the research projects are chosen, sponsored by, and carried out in collaboration with multinational corporations. Joint teams that include MIT SCM students and MIT CTL faculty work on the real-world problems. In this series, we summarize a selection of the latest SCM research.