An 80% chance on each of two independent long-lead paths is a 64% chance that both arrive when qualification needs them.
That is merge bias, and a qualitative risk register cannot show it. The FMEA register lists what might go wrong, scores it and gives each risk an owner. It suits week-to-week risk management. It cannot add risks up along the program and tell you how likely the finish date is.
A quantitative schedule risk analysis does that. Each activity on the critical and near-critical paths gets a three-point duration estimate. A Monte Carlo simulation runs the program thousands of times and returns a distribution: the date with a 50% chance of being met, the date at 80%, and which activities drive the spread.
Most GMP projects do not need it at every stage. It earns its cost when:
- Several long-lead paths (AHUs, clean utility skids, MCCs) converge on one qualification start.
- The finish date is tied to something that will not move, such as a booked inspection or a registration filing.
- Liquidated damages or a product launch sit behind the date.
The inputs are the hard part. Three-point estimates have to come from the people who own the work, not from the scheduler alone, or the simulation reproduces the plan's optimism with more decimals.
In our EPCM work the FMEA register is reviewed weekly and reported monthly beside schedule, cost and earned value.
Which date does your board see, P50 or P80?
Industries




