Union Pacific assigned empty freight cars by optimization, cutting cost 35% ROI.

A railroad has to send empty freight cars to where customers need them, balancing where cars sit, how urgent demand is and whether one car type can substitute for another.

Union Pacific worked with Purdue researchers on an optimization model that weighs all these factors at once and produces real-time assignments inside a total car-management system.

The model cut transportation cost as its simulation predicted and reduced the staff required for demand fulfillment, producing a 35% return on investment.

Assign empty cars by model, not by hand

Empty-car choices depended on many interacting factors

A model can weigh location, urgency and substitution together

Optimization made the assignment decision fast enough for real time

Measured: reduced cost and staff, with a 35% ROI

When a daily operating choice depends on many interacting factors, replacing rules of thumb with an optimization model can cut cost and staff while improving service, even in real time.

The model became part of Union Pacific's real-time car-management process and the approach was extended across the network, showing how a shared optimization core could run an operational decision that once took many specialists.