The encyclopedia · Engineering & Operations · Operational decision · 2021–2025
WM automated 4,000 industrial waste routes a night, cutting manual planning.
WM rebuilt industrial waste routing as an optimizer that builds 4,000 routes in 45 minutes nightly across 500 sites, replacing manual planning.
WM (Waste Management)
the move
WM, North America's largest waste and recycling company, plans hundreds of thousands of industrial customer services a year. Post-pandemic, the factors that drive them — fluctuating demand, customer-specific requirements, different container types and sizes, and varying disposal destinations — became too dynamic for planners doing it by hand.
WM's operations research and data analytics teams built a dynamic route optimization program for the industrial line of business. It forecasts future service demand for capacity planning and runs advanced metaheuristic algorithms to generate routes automatically.
The scale is striking: about 22,000 customer services across 500 hauling sites, reduced to 4,000 routes built under 45 minutes every night for the next day, at close to 99.9% accuracy. That is a lot of data to get right before the optimizer can be trusted.
The decisive step was adoption. WM had tried twice before over 15 years and failed; this time it involved dispatchers, routers and drivers in testing the solution before going live, so field teams trusted the optimization rather than fighting it. The result was best-ever efficiency and operating margins.
why it works
- Industrial demand is dynamic, so a forecast-fed optimizer beats a manual plan.
- Metaheuristics make the 4,000-route nightly problem tractable.
- Trust from the field is what makes an optimizer actually get used.
- Automating a labor-intensive process removes the slow, suboptimal step.
what transfers
A model is only worth its adoption, and adoption comes from letting the people who run the system see it work before it goes live.
what came after
WM's dynamic route optimization was named a 2025 Franz Edelman Award finalist. It demonstrated that in a route-optimization problem the hard part is often not the solver but getting the data right and earning adoption from the people who drive the trucks. Automated routes delivered best-ever efficiency gains and operating margins, and became the foundation for further automation in the industrial line of business.
references
- 2025 Franz Edelman Award finalist: WM
- Dynamic Route Optimization and Automation of Industrial Routes at WM
spotted an error? The archive wants to know.