How to Plan Brownfield Changes Without Stopping Production
To plan a brownfield change without stopping production, test the order of the moves before the first machine is disconnected. A digital twin of the plant lets you run each stage of the rebuild as its own scenario: which machines move first, where material goes while an aisle is closed, how much buffer stock bridges the move and what happens to output that week. The result is a sequence of stages with dates and an output check for each one.
Why brownfield changes are harder than a new hall
In a new plant, you build the layout and then start production. In a running plant, every change happens next to orders that must ship this week. Machines move one by one. Aisles close for days. Material takes detours. The same forklifts, the same crane and the same maintenance team serve both production and the rebuild.
A project plan shows tasks and dates. It does not show how the flow behaves while the plant is half old and half new. A press moves to the new hall on Monday, and the plan assumes it runs again on Wednesday (an illustrative case). Its parts now travel 150 meters farther with the same number of forklifts. The buffer in front of assembly empties by Thursday, and the weekend becomes overtime.
Five reasons brownfield plans fail
Most problems during a rebuild come from a short list.
- Temporary routes. Detours during the rebuild make trips longer, and the forklift fleet stays the same.
- Buffers too small to bridge a move. A machine that is down for days needs stock built before it stops.
- Too many moves at once. Parallel moves compete for the same crane, technicians and electricians.
- Instant ramp-up assumed. The plan treats a moved machine as running at full rate the next day.
- The first layout idea is never tested. The order of moves follows the first drawing, even when a better final layout exists.
All five play out over time. A drawing has no time axis, so it cannot show them, and a simulation can.
How to model the stages of a rebuild
The method itself is simple. Most of the work goes into describing each stage honestly.
- Build the baseline. Model the current plant: routings, cycle times, transport, shifts and the real order mix. Use recent weeks, including the difficult ones.
- Model the target state. The final layout, with the new area, machines and routes.
- Model each intermediate stage. Which machines stand where, which aisles are closed, which routes are in use, which machines are down.
- Add the shared constraints. Crane and technician time, shift patterns, deliveries that must continue.
- Run each stage with the order mix planned for that week. Compare output and buffer levels against the target.
- Test other sequences. Change the order of moves and look for moves that can run in parallel without hurting output.
- Set hold points. Agree what must be back to normal before the next stage starts.
| Stage | Physical state | What the model checks |
|---|---|---|
| Baseline | Current layout, all machines running | Output, waiting time and buffer levels today |
| Stage 1 | First machines moved, one aisle closed | Longer routes, forklift load, buffer in front of the next process |
| Stage 2 | Some processes in the new hall, some in the old | Transport between halls, waiting for pickup, shifts in two places |
| Stage 3 | Final layout, ramp-up | Output against target, where parts still wait |
| After go-live | Model updated to the real state | Ready for the next change |
For the final layout itself, compare several variants with the same measures, as described in How to Compare Layout Variants Without Guesswork.
Planning a rebuild in a running plant?
Which machines have to move, and which orders cannot slip while they do? With that list we can suggest which stage of the rebuild to model first.
Real case: a new hall in a running car lamp plant
A car lamp manufacturer in southern Poland, with 270 people on two shifts, added a new production hall. The plant runs injection molding, painting, electroplating, assembly, labeling, quality control and packaging. Machines had to be relocated and capacity expanded without interrupting production. The team also had to check intermediate storage capacity and keep intralogistics deliveries running during the move.
DBR77 built a digital twin that included the new area, the material flows and the processing time of each task. Process animations were used to test relocation options and the order of the moves. What the project found:
- The result was an expansion plan designed to keep output at the target level while machines move.
- Buffer storage was the main planning challenge: how big the buffers between processes should be and where they should stand.
- Extended transport routes were the main source of production losses. Longer trips cost more output than anything else in the plan.
- The final machine layout differed from the initial assumption, because running the flow showed a better arrangement than the first drawing.
- The best relocation sequence let some processes run in parallel, so some moves did not have to wait for the previous one.
Management reviewed the final design in VR. That helped them choose the sequence of transferred processes faster.
Start with the data you have
Brownfield plants rarely have clean digital data, but a first model can use the CAD layout, routing sheets, order lists and cycle times measured with a stopwatch. Nominal times are enough to start: enter them by hand, or bring them in by import (import scope to be confirmed). The simulation is deterministic (Averages vs simulation over time), so the same input always gives the same result. The first runs of the current plant will rarely match the floor exactly, and that gap shows where the real constraint is before the rebuild starts.
If the modernization also adds machine data, the two projects support each other. Live data from DBR77 IIoT is a later stage, once the model is used regularly and you want it to follow real cycle times as they change. See What the First 30 Days of IoT Should Look Like in a Brownfield Factory for that side of the work. If this is your first model, How to Run Your First Simulation Project covers the steps, roles and timing.
After the move: keep the model
A brownfield plant is never finished. After go-live, update the model to the real state: where machines stand now, the real cycle times after ramp-up, the routes people use. The next change, whether a robot, a new product or another hall, then starts from a model that already matches the plant. Without that update, the model stays a one-time study.
FAQ
Can we build a digital twin of an old plant without digital data?
Yes. Most first models start from a CAD layout or a PDF drawing of the hall, routing sheets, order lists and measured cycle times. Missing values can be measured on the floor with a stopwatch.
Should we move everything at once or in stages?
Model both. A single cutover can be faster on paper, but it puts all moves on the same crane, technicians and buffers at the same time. Compare output and buffer levels week by week as well as at the end date.
Do we need to model the whole plant?
No. Model the area affected by the change plus the processes that feed it or depend on it. In the car lamp case, the twin included the new area, its material flows and the processing time of each task.
How do we get management to agree on the sequence?
Show them each stage along with the final layout. In the car lamp case, management reviewed the final design in VR, which helped them choose the order of moves faster.
Conclusion
List every machine that has to move, how many days each will be down, and which processes feed it or depend on it. The moves with the longest downtime next to the busiest processes are the stages to model first. Agree a hold point for each of them before the dates go into the project plan.
Test the order of your moves first
DBR77 Digital Twin can run a relocation one stage at a time. In a demo you can see different move sequences compared on output and buffer levels for each week of the rebuild.
Sources
- DBR77, Case study: Plant expansion plan (summary on this site)
- DBR77 Digital Twin, Averages vs simulation over time
- DBR77 IIoT, What the First 30 Days of IoT Should Look Like in a Brownfield Factory
Want to test a decision from your plant?
Describe the decision to a DBR77 expert and we will show how DBR77 Digital Twin compares the options on output, waiting time and transport load.
Questions
Have a question about this topic?
Tell us what it is about. We will reply within one business day.
Digital Twin newsletter
Short notes on testing layout and flow decisions on a model, and news from DBR77 Digital Twin. You can unsubscribe at any time.