Building a Smarter (and Cheaper) School Bus System: How a Boston-MIT Partnership Led to New Routes That Are 20% More Efficient and Saved the District $5 Million
As of August 26, Boston Public Schools faced a problem in efficiently routing its school buses. The available solutions on the market were ineffective, prompting the district to launch a public competition for external assistance. The winning team from MIT’s Operations Research Center developed an algorithm that resulted in significant savings for the Boston schools. In its first year, the algorithm saved $5 million and 1 million miles driven. It also allowed the district to reduce its fleet by 8%, the largest drop in a single year in the district’s history. Currently, after two years of implementation, the district has reduced its number of trips by almost 400, with a total of 2,800 trips and a more efficient system utilizing around 620 buses, representing a 20% increase in efficiency.
According to Will Eger, the strategic projects manager of Boston Public Schools, the team from MIT was the clear winner, as their solution achieved in just 30 minutes what used to take a team of 10 staffers 3,000 hours. This solution has the potential to assist districts nationwide.
Routing school buses presents unique challenges due to the need to start from scratch each year as students change grades, switch schools, and enter or leave the system. Arthur Delarue, an MIT Ph.D. student involved in the project, explains that the work done in the previous year becomes irrelevant, requiring a complete restart. While routing algorithms work effectively for logistics companies like FedEx and UPS, they do not apply to school districts that have specific constraints determined by community priorities, state regulations, and the local environment. Every school district faces slightly different problems, making it difficult to generalize a framework that applies to all. Routing school buses is mathematically complex and involves a vast number of possibilities. To find good solutions, it is necessary to quickly eliminate a multitude of potential poor solutions.
Boston’s transportation service for students spans 20 zip codes, allowing them the option to choose from more than 220 schools. Boston’s policy is to provide service for students living more than half a mile from a school while limiting their time spent on the bus. Bus stops are located at the closest possible intersections. This constant change requires frequent adjustments to bus stops and routes, demanding a high level of agility. The goal is to minimize disruption to student rides while responding to changes in enrollment.
The best solution turned out not to be the one with the fewest buses per school but rather one that maximizes bus reuse between schools. Each school is evaluated for the most suitable solution, taking into account how schools interact with each other. The team found that the most effective approach was for most buses to be shared by three schools, both in the morning and afternoon. This strategy considers the various types of buses and their different capacities based on students’ ages.
Efficient routing is crucial, and the savings made can be reinvested in classrooms. Additionally, the MIT solution allows the district to explore policy changes and their impact on costs accurately. For example, if the district considers providing bus service only to students living three-quarters of a mile from a school, the MIT solution can help gauge the cost. This newfound ability enables the district to make smarter policy trade-offs.
The successful collaboration with Boston has prompted MIT professor Dimitris Bertsimas to establish a company called Dynamic Ideas, which aims to market the product to other school districts. The company is currently in the early stages of adoption with other districts.
“There are unforeseen circumstances that can arise, and Boston has a highly capable team of approximately 10 transportation officers who effectively tackle these challenges and make adjustments,” Martin explains. “However, it is important to acknowledge that the process still relies on human intervention and decision-making.”
As Dynamic Ideas continues to expand, the team acknowledges that each district presents unique problems. While the algorithm can be applied in various settings, it requires time to adapt and address multiple issues.
Boston, having witnessed the success of this collaborative effort, is enthusiastic about the potential victories that can be achieved by seeking external assistance. Eger expresses excitement about partnering with exceptionally talented researchers in unconventional places. He admits to being unaware of the MIT Research and Operations Center before this project and is eager to explore other impactful problems they can tackle together.
Delarue commends Boston Public Schools for their exceptional initiative in recognizing the need to seek external solutions. Their proactive approach has propelled this conversation forward nationwide.
