The encyclopedia · Engineering & Operations · Operational decision · 1998–2003
PSA Peugeot Citroën tuned car-body lines like stochastic systems, +$130M
Simulation and Markov-chain models raised throughput on 75% of PSA's car lines with minimal capital.
PSA Peugeot Citroën
The solution
In 1998 PSA's new CEO set an ambitious strategy of growth, innovation and profitability. The car-body shops were the bottlenecks of PSA's plants, so improving their throughput without huge capital spending became the priority.
An R&D team built OR tools that combine simulation and Markov-chain models of series-parallel systems, then used them to design and continually improve the car-body lines. Over 75 percent of PSA's cars are produced on lines designed with these tools.
The models worked because they captured variability and blocking between stations rather than assuming smooth flow. The project contributed US$130 million to the bottom line in 2001 alone, and suppliers adopted the tools on their own.
Why it worked
- Variability-aware models exposed real bottlenecks
- Throughput rose without major capital spending
- The approach spread to suppliers unprompted
What can be applied
Before buying more machines, model the line as a network of queues: variability, not capacity, is often the real bottleneck.
Aftermath
PSA's suppliers acquired and used the same tools, and the 2003 Interfaces paper became a reference for stochastic modeling of production lines.
Sources
- Improving Car Body Production at PSA Peugeot Citroën
- Improving Car Body Production at PSA Peugeot Citroën (abstract)
spotted an error? The archive wants to know.