The encyclopedia · Engineering & Operations · Operational decision · 2000–2018
BNSF put one OR team inside every decision, saving $110M a year
A centralized operations-research and analytics team tells BNSF's leaders the answer, not just the analysis, across the whole railroad.
BNSF Railway
the move
BNSF's operations research and advanced analytics program grew, over 25 years, from a small effort into a way of running the railroad. One centralized team advises leadership on capital decisions with modeling, optimization, machine learning and image analytics.
Rather than a separate department that hands back reports, this team works inside each part of the business — mechanical, engineering, marketing, transportation and service design — so its methods shape routing, crew scheduling, train building, fleet sizing, dispatching, fueling, maintenance and equipment distribution.
The result is that dozens of interlocking decisions are governed by a consistent modeling approach instead of a patchwork of local habits.
why it works
- It was embedded in every department, so models influenced real decisions rather than sitting on a shelf.
- A central team set a common standard and could reuse models, data and methods across the railroad.
- It stretched from day-to-day operations up to capital-intensive strategic choices.
- It applied a wide toolkit — optimization, machine learning, image analytics — where each fit.
what transfers
A single embedded analytics group that works across every department beats scattered one-off projects — the models compound because each decision is informed by the same data and methods.
what came after
The team helped BNSF save more than $110 million annually, and the integration earned BNSF the 2018 INFORMS Prize for Operations Research, awarded for long-term, organization-wide integration of OR rather than a single win.
references
- BNSF earns INFORMS award for analytics, operational research efforts
- BNSF Railway presented with the 2018 INFORMS Prize for Operations Research
spotted an error? The archive wants to know.