GM instrumented its plants for throughput optimization and saved over $2.1 billion

In the late 1980s General Motors set out to predict and improve the throughput of its production lines, treating productivity as a source of strategic advantage rather than a plant-level nuisance. Until then, decisions about buffers, bottlenecks and line speed rested on rules of thumb and local experience.

GM quantified throughput performance and focused improvement on the design and operation of its manufacturing systems through three coordinated activities: algorithms for estimating throughput, locating bottlenecks and optimizing buffer allocation; real-time plant-floor data collection to support those algorithms; and common processes for identifying opportunities and implementing performance improvements.

The combination let GM see which constraint in a line actually limited output and where a buffer or a speed change would pay. Across more than 30 vehicle plants in 10 countries, the program increased revenue and saved over $2.1 billion, and it became GM's standard way of designing and running lines.

Treat plant throughput as an algorithmic design variable

Throughput losses compound serially: a line's output is set by the joint design of stations, buffers and variability, not by any single station

Real-time plant-floor data kept the models honest, so estimates tracked what the line actually did

A common improvement process turned one-off analyses into a repeatable rollout plant by plant and country by country

Measured: over $2.1 billion saved and revenue gained across 30-plus vehicle plants in 10 countries

A serial line's throughput is a property of the whole design, not of any one station. Models plus live data plus a repeatable rollout process is what scales a saving across dozens of plants.

The program won GM the 2005 Franz Edelman Award, and the team published the methods as 'General Motors Increases Its Production Throughput' in Interfaces 36(1) in 2006. Bottleneck detection and buffer-allocation tools of the kind GM industrialized are now standard practice in manufacturing science.