The Challenge
A mid-sized cable manufacturer in Turkey produces power cables in multiple diameters (from 1.5mm to 240mm) and colors (12 standard colors plus custom). Their extrusion lines require changeovers whenever diameter or color changes.
The numbers told a painful story:
- 6 hours of changeover time per day across 4 extrusion lines
- Color changes: 15-40 minutes depending on color sequence
- Diameter changes: 30-75 minutes depending on size jump
- Planners manually sequencing jobs in Excel
- No systematic approach to minimizing changeovers
Understanding Sequence-Dependent Setups
Not all changeovers are equal. Changing from white to black requires purging the entire color system—45 minutes minimum. Changing from white to light grey? Just 15 minutes.
Similarly, stepping from 10mm to 16mm diameter takes 30 minutes. Jumping from 10mm to 95mm? That is a 90-minute changeover involving die changes, pressure adjustments, and cooling system reconfiguration.
This creates a sequence-dependent setup matrix: the time to prepare for the next job depends entirely on what the previous job was.
The Manual Approach
The planning team had developed rules of thumb over years of experience:
- "Always run light colors before dark colors"
- "Group similar diameters together"
- "Never run black before a custom color job"
These heuristics helped, but they were incomplete. With 200+ active SKUs and daily demand changes, the planners could not possibly evaluate all sequencing options. They were solving a 200-factorial problem with intuition alone.
Building the Optimization Model
We started by measuring actual changeover times across every transition. This took 6 weeks of data collection, but it was essential. The result was a 50x50 matrix for colors and a 30x30 matrix for diameters.
The constraint model included:
Decision Variables: Which job runs on which line, in which position in the sequence
Objective Function: Minimize total changeover time while meeting all due dates
Constraints:
- Each job runs exactly once
- Jobs cannot overlap on the same line
- Due dates must be respected
- Material availability constraints
- Minimum batch sizes for certain products
The Results
After implementing the optimization model:
| Metric | Before | After | Improvement |
|---|---|---|---|
| Daily changeover time | 6 hours | 3.9 hours | 35% reduction |
| On-time delivery | 81% | 91% | 10 points |
| Throughput (tons/day) | 24 | 26.9 | 12% increase |
| Planner time on scheduling | 4 hours | 1 hour | 75% reduction |
The math is straightforward: 2.1 hours saved daily x 250 working days = 525 hours per year. At their production rates, that translates to meaningful additional capacity—without any capital investment.
Key Implementation Insights
Data quality matters more than algorithm sophistication. The first model we built was theoretically elegant but produced poor results because the changeover matrix was based on estimated rather than measured times.
Operator buy-in is essential. We involved line supervisors early. They validated the changeover matrix and flagged constraints we had missed. When the first optimized schedule ran 30 minutes faster than their best manual attempt, skepticism turned to enthusiasm.
Start with one line. We piloted on a single extrusion line for 8 weeks before rolling out to all four. This contained risk and built confidence.
Integration beats standalone tools. The optimization runs automatically every night, pulling the latest demand from SAP and pushing the schedule to the shop floor terminals. No manual data transfer, no Excel exports.
Broader Applications
This approach applies wherever sequence-dependent setups exist:
- Plastics injection molding: Color and resin type changeovers
- Food processing: Allergen-based sequencing, flavor changeovers
- Pharmaceutical: Cleaning validation requirements between products
- Metal fabrication: Tool setup, material type transitions
The underlying math is the same. What changes is the domain-specific constraints and the changeover matrix.
Conclusion
Changeover optimization is not about eliminating setups—they are a physical necessity. It is about being smarter about sequence. A 35% reduction in changeover time means significantly more time making product. For this cable manufacturer, that translated directly to the bottom line.