Investment decisions (CAPEX)8 min read

How to Compare CAPEX Options When Every Scenario Looks Plausible

DBR77 TeamPublished

When every CAPEX option looks plausible, stop comparing the slides and compare how each option behaves on the same bad weeks. Run every option through one fixed set of disruptions: demand drop, unfavorable mix, a slower bottleneck machine, a late supplier, a short shift. Then rank the options by output at the real bottleneck, waiting and inventory, cash timing and the cost of reversing the decision. If two options still tie, choose the one you can stage or undo.

Why every option looks good on a slide

A typical committee sees three proposals for one problem: a robot cell, a second line, or a relayout with a larger buffer. Each comes with a spreadsheet, a payback inside the target band and a confident sponsor.

They all look plausible because they share the same optimism: a smooth ramp, a stable mix, on-time suppliers and average cycle times. Under those conditions most options work, so the ranking depends on who presents best.

The factory does not run on averages. A slow station builds a queue in front of it. A shared forklift is free on average and busy exactly when the line needs it. A spreadsheet cannot show which option only works when nothing goes wrong. A simulation model that runs production over time can.

Step 1: Freeze the decision and the option set

Before any model runs, write down three things.

  1. The decision itself, in one sentence: "Which option raises output of product family B to the planned volume by Q3 next year?"
  2. The option list, with numbers and owners. Include "defer" or "do nothing" as a real option. It is the baseline the others must beat, and sometimes it wins.
  3. The kill criteria, written before the first run. Examples: the option misses the delivery target in the agreed downside case, it makes the line depend on one station nobody can back up, or it needs a supplier performance nobody will sign.

An option that appears in week six and skips the tests the others went through makes the whole comparison unfair, so close the list now.

Step 2: Run every option through one stress set

In a fair comparison every option faces the same shocks, inputs and simulated time. A standard stress set usually includes:

  • the base case, with the planned order mix for a typical week
  • demand down and demand up by a defined share of volume, with the same shift rules
  • an unfavorable mix: more of the family that hurts the investment most, for example more changeovers
  • the bottleneck machine or robot running at a defined slower cycle time
  • an inbound material delay that procurement agrees is realistic
  • a shift pattern with fewer hours than planned

Every option runs on the same fixed nominal times, and the same input always gives the same result (Averages vs simulation over time), so a difference between options comes from the option, not from noise. How much that difference depends on each assumption, and which inputs drive it, is the subject of What a Sensitivity Analysis Should Show Before Approval.

Step 3: Score more than payback

Payback belongs in the pack, but it should not decide on its own. Score every option on the same lines, in the base case and in the stress cases.

The numbers below are illustrative.

MeasureOption A: robot cellOption B: second lineOption C: relayout with buffer
Output per week, base case4,100 units4,300 units4,000 units
Output per week, key station slower3,200 units3,900 units3,700 units
Average waiting time before assembly22 min15 min18 min
Work in progress on the floorHighMediumLow
Cash out before first benefit9 months14 months4 months
Can it be staged or partly reversed?PartlyNoYes

On the base case, Option B wins. With the key station slower, B still leads, but C is close, pays back much earlier and can be stopped halfway. A committee that sees only the base case never gets to have that discussion.

Work in progress deserves its own line, because it is cash. A furniture manufacturer whose layout no longer fit its product variety used a DBR77 digital twin of production and warehouse. AI generated layout proposals, specialists chose one, and it was simulated. Working capital fell by PLN 300,000 after eliminating the buffer. Departmental costs fell by 14% and direct labor demand by 1.5 full-time positions. None of that shows up in a payback calculation that only counts the machine price.

To turn these lines into a business case without made-up numbers, see How to Build a Digital Twin Business Case.

See how plants compare investment options

Robot cell or second line, relayout or bigger buffer: the use cases show where plants compared variants like these in DBR77 Digital Twin before money was committed.

Browse use cases

Step 4: Retire weak options early, with written reasons

Committees often keep every option alive to avoid conflict, so the last meetings recycle options that should have closed weeks earlier. A simple rule works better:

  • Run every option on the list through the stress set without special tuning for anyone's favorite.
  • Hold one short meeting that applies the kill criteria from Step 1 without reopening them.
  • Archive each retired option with the scenario that killed it and a one-line reason.
  • Allow a reopen only with new evidence or a changed scope.

Check that a retirement is not a knife edge: if a small input change would bring the option back, say so. Then focus the next round on the survivors.

Step 5: Attach the right evidence to each approval gate

Most plants approve CAPEX in stages. Simulation helps most when each gate asks for one specific piece of evidence.

  • Concept gate: two or three option types compared on the base case. Are we solving the right problem?
  • Business case gate: the short list through the full stress set, with bottleneck, waiting time and inventory per option.
  • Design gate: sensitivity on the top assumptions, each with a named owner.
  • Execution gate: ramp-up and installation sequence tested, so output holds while equipment is moved or installed.
  • After approval: a rerun when scope, mix or supplier assumptions change before money is spent.

How to put the result on one page for the board is described in How to Present Simulation Results to the Board.

When two options still tie

If two options survive every stress case with similar results, decide on flexibility. Ask three questions of each:

  • Can the spend be staged, so you buy the second half only if demand arrives?
  • Can it be reversed at a reasonable cost if the product mix changes?
  • Can production keep running during installation?

Prefer the option that stays acceptable if the base case turns out wrong.

FAQ

How many options should a CAPEX comparison include?

Usually two to four, plus "defer". More options rarely add insight and make it harder to run all of them through the same stress set.

Which disruptions should be in the stress set?

Use the ones your plant has lived through: a slower bottleneck, a late delivery, a demand swing, a bad mix week, a short shift. Agree the size of each shock before the first run and keep it the same for every option.

Do we need live machine data to compare CAPEX options?

No. Most comparisons start from layouts, routings, measured cycle times and the planned order mix. Live data from DBR77 IIoT helps later, when the model is reused often.

What if the simulation ranking disagrees with the payback ranking?

Look at why. Often the best payback assumes the base case, while the simulation shows the option losing output when the key station is slower. Show both rankings and explain the difference.

Conclusion

Start before any model exists. Write the decision in one sentence, list the options with "defer" among them, and get the kill criteria signed by the people who will sit on the committee. Then agree the size of each shock in the stress set with procurement and operations, so no sponsor can bring a gentler version for their own option.

Compare your investment options on the same bad weeks

Have two or three proposals on your committee's table? The demo shows how options like these are set up against one stress set, starting from the bottleneck.

Talk to an expert

Sources

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.

Talk to an expert

Questions

Have a question about this topic?

Tell us what it is about. We will reply within one business day.

We use your data only to answer your request. See our Privacy Policy.

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.