TRUMPF used one MILP model to question its inherited factory network

TRUMPF had grown for 15 years into a global high-technology machinery supplier. That growth left a production network shaped by history, not by a fresh view of total network cost. Moving products between plants was possible, but the effects were too entangled for intuition alone.

The company built a mixed-integer linear programming model for production-network design. Instead of asking the model to dictate a final answer, planners first used it to compute a theoretical reference scenario, then tested alternative feasible scenarios against that benchmark.

That framing was the ingenious part. Optimization became a negotiation instrument: it showed where the inherited network was expensive, which cost factors mattered, and what a realistic future design would give up compared with the mathematical ideal.

Use optimization as a reference case, not an oracle

A reference optimum made the cost of practical compromises visible.

Scenario analysis kept managerial constraints in the room instead of pretending they did not exist.

The model taught planners which cost drivers mattered across the whole network.

For strategic network redesign, the model's best use may be exposing trade-offs clearly enough for humans to choose the feasible move.

The published case says the model informed a strategic decision about TRUMPF's future network design and guided planners through a more systematic decision process. The work has since been cited in production-network and capacity-planning research.