The encyclopedia · Engineering & Operations · Operational decision · 2006
Swift & Company schedules beef fabrication live with 45 linear programs, getting 200% ROI.
Swift & Company used 45 linear-programming models to schedule its five beef plants in real time and promise delivery dates, earning a 200% return in year one.
Swift & Company
the move
Swift & Company processes beef across five US plants and takes orders continuously, so planning was slow and many promises were made that the plant could not keep.
The obvious approach is to plan the week once and hope. Swift instead built an integrated system of 45 linear-programming models based on three model formulations that schedule fabrication dynamically in real time as orders arrive.
This let the company run a better plant and, crucially, answer the customer about a requested date in seconds using a capability-to-promise view of cattle supply and plant capacity.
The application improved order fulfilment and on-time delivery, and delivered a 200 percent return on investment in its first year.
why it works
- Real-time dynamic scheduling matches the pace at which orders actually arrive
- A single system of LP models unites the five plants rather than each planning alone
- Capable-to-promise turns a back-office system into a sales advantage
- Better promises and better delivery feed each other
what transfers
When orders arrive continuously, a model that re-plans in real time turns the schedule into a sales tool — it lets you promise only what you can actually deliver.
what came after
The application enabled Swift & Company to execute its business strategy and document improvements in order fulfilment, on-time delivery and the share of a week's production already backed by orders, earning a 200% return on investment in its first year. The Interfaces (2006) paper became a model for capable-to-promise scheduling.
references
- A scheduling and capable-to-promise application for Swift & Company
- A Scheduling and Capable-to-Promise Application for Swift & Company
spotted an error? The archive wants to know.