DHL's optimizer decides what to ship itself vs outsource, saving $98.6m a year.

DHL Supply Chain North America runs one of the world's largest logistics networks, delivering over a billion packages a year for corporate customers, and faced decisions about bidding, fleet size and type, outsourcing and routing that had been handled separately.

Led by Ohio State PhD graduate Yibo Dang with Ted Allen's help, the team built the Transportation Network Optimizer, using innovative integer-programming and a symbiotic twin-ant-colony search to combine routing and outsourcing algorithmically.

The software solves large-scale routing and is saving an estimated $98.6 million a year and half a megaton of CO2 emissions since implementation, which made DHL and Ohio State a finalist for the 2023 INFORMS Franz Edelman Award.

Decide what to outsource inside the route model

Outsourcing and routing are one decision, not two

Future-demand signals decide what to send to a third party

Integer programming handles fleet size, type and bidding

Measured: $98.6m a year and half a megaton CO2

In a network business the biggest savings come from an integer program that couples outsourcing, fleet and routing, instead of a route optimizer that assumes the network is fixed.

The model was published in Dang's Ohio State dissertation and became the basis of DHL's network planning, while the academic-industrial collaboration show-cased how a PhD project can become production software that also cuts emissions.