The encyclopedia · Engineering & Operations · Technical decision · 1984–1991
Southern Company's Wescouger scheduled every generator, saving $140M in fuel
When fuel outgrew every other cost, Southern Company put load forecasting, unit commitment and power pricing in one optimizer — $140M saved in seven years.
Southern Company
The solution
Southern Company, the US utility holding group, watched fuel costs climb faster than every other cost component through the 1980s. Its Birmingham control center had to decide daily which thermal and hydro units to run and how to price power transactions with neighbors.
With ABB Power Systems it installed an operational-planning package anchored by Wescouger: load forecasting, unit commitment by dynamic programming and branch-and-bound, and price estimation for power trades, all in one loop.
Over seven years Wescouger helped Southern Company save over $140 million in fuel costs, and by 1991 it was a key daily scheduling tool at the control center, serving the pool's central duty of delivering electricity reliably and cheaply.
Why it worked
- Unit commitment is combinatorial; DP and branch-and-bound found good global schedules
- One shared model made load, generation and trading decisions consistent
- Daily use turned optimization into operations, not a study
- Fuel dominated costs, so a fuel-saving schedule moved the bottom line
What can be applied
When one input dominates cost, optimize the whole system around it: a single unit-commitment model beats local fixes because every generator's run affects the others.
Aftermath
The Wescouger architecture — forecasting, unit commitment and transaction pricing in one system — became a template for utility control-room software, and similar packages spread across the industry as deregulation approached.
Sources
- Using an Optimization Software to Lower Overall Electric Production Costs for Southern Company
- Using an Optimization Software to Lower Overall Electric Production Costs for Southern Company (abstract)
spotted an error? The archive wants to know.