Monte Carlo Simulation for Project Cost Risk
Monte Carlo simulation runs your project budget through thousands of randomised futures, producing a probability distribution of outcomes rather than a single number. AACE International recommends it for any project above moderate complexity.
The method in five steps
- Break the budget into cost elements (WBS leaves or risk-bearing line items).
- For each element, set a distribution (triangular, beta, lognormal) using three-point estimates or historical data.
- For each risk event, set probability and impact distributions.
- Run 10,000+ iterations. Each iteration samples one value from each distribution and sums the totals.
- Report the result as a cumulative distribution: P50, P80, P95 budgets.
Example output
For a hypothetical 10M USD construction project with three-point estimates per cost element, a Monte Carlo run might produce:
| Percentile | Total cost | Interpretation |
|---|---|---|
| P10 | 9.2M USD | 10% chance project comes in this cheap or cheaper |
| P50 (median) | 10.8M USD | 50/50 chance project finishes at or below |
| P80 | 12.4M USD | 80% chance project finishes at or below (typical budget target) |
| P90 | 13.5M USD | 90% chance project finishes at or below |
| P95 | 14.8M USD | tail risk; only set this if downside is catastrophic |
The point estimate of 10M USD was actually at roughly the P25 percentile. Budgeting at the point estimate means the project starts with a 75% chance of overrun. Most large public-project standards (UK ICE PR47, AACE 40R-08, US DoE 413.3B) target P80 or P85.
Sources
- AACE International (2019). Recommended Practice 57R-09: Integrated Cost and Schedule Risk Analysis Using Risk Drivers and Monte Carlo Simulation of a CPM Model.
- AACE International (2008). Recommended Practice 40R-08: Contingency Estimating - General Principles.
- Hulett D. (2011). Integrated Cost-Schedule Risk Analysis. Gower.
- US Department of Energy (2018). DOE O 413.3B: Program and Project Management for the Acquisition of Capital Assets.