Esc

↑↓ move↵ openIndex · Pagefind
Project management

Monte Carlo P80

Definition

The finish date that 80% of simulated schedule outcomes meet or beat in a Monte Carlo schedule risk analysis.

Topic: Scheduling

Updated

Monte Carlo finish-date S-curveA cumulative probability curve of simulated finish dates. The deterministic CPM finish of 30 June is met in only 22 percent of iterations. The median, P50, is 14 July, and P80 is 2 August, 33 days after the CPM date.80%50%CPM22%P8030 Jun14 Jul2 Augfinish date, 2027
Illustrative. Committing to P80 means 33 days of contingency beyond the CPM finish.

A Monte Carlo schedule risk analysis recalculates the schedule thousands of times. Each time, activity durations are drawn from a range, such as a triangular or PERT (beta) distribution between optimistic and pessimistic values. The results form a distribution of finish dates. P80 is the date that 80% of the iterations meet or beat. P50 is the median. P10 and P95 bracket the optimistic and pessimistic tails.

The simulation also shows which activities matter. The criticality index is the share of iterations in which an activity was critical. Duration sensitivity is the correlation between an activity’s sampled duration and the finish date.

How it works

  1. Start from a sound schedule. Logic should be complete, with few constraints and no open ends. A DCMA 14-point check is a common first pass.
  2. Assign ranges. Each activity, or group of activities, gets a minimum, most likely and maximum duration, as in a three-point estimate, ideally from the people doing the work.
  3. Add risk events. Discrete risks from the risk register, such as a failed inspection, can be modeled with a probability and an impact.
  4. Set correlation. Activities likely to run long together, such as work by the same crew, are correlated so the simulation doesn’t let them cancel out.
  5. Iterate. The engine samples every range, recalculates CPM dates, and records the finish and the critical path. It repeats this many times.
  6. Read the results. The finish dates form a histogram and a cumulative S-curve. Percentiles such as P50 and P80 are read off the curve.
  7. Act. Size schedule contingency, target the activities with the highest criticality and sensitivity, and rerun after mitigation.

Worked example

Illustrative: a schedule with a deterministic finish of 30 June 2027 is simulated with 1,000 seeded iterations. The results:

Measure Finish
P10 25 June 2027
Deterministic CPM finish 30 June 2027 (met or beaten in 22% of iterations)
P50 14 July 2027
P80 2 August 2027
P90 11 August 2027

Committing to the P80 date means carrying 33 calendar days of schedule contingency beyond the CPM finish (30 June to 2 August). The facade installation, critical in 64% of iterations though it has float in the deterministic schedule, is the top driver, so the team adds a second crew option and reruns the simulation.

Measure What it tells you Limitation
Deterministic CPM finish One calculated date from single durations No stated confidence
P50 Median simulated finish Half of outcomes are later
P80 Finish with 80% simulated confidence Only as good as the ranges and logic
Criticality index How often an activity drives the finish Depends on the model’s paths
Total float Slack on one deterministic path Ignores uncertainty
Longest path The driving chain in the CPM model One scenario only

Why it matters

A single deterministic CPM finish is one possible outcome. In a schedule with parallel paths it is often an optimistic one, because a slip on any of several near-critical paths moves the finish. A P80 date gives owners and lenders a finish with stated confidence, and it shows how much time contingency would buy that confidence.

Which percentile to use is a business decision, not a statistical one. If finishing late carries heavy liquidated damages, lost rent or a fixed opening date, a higher percentile such as P80 or P90 is prudent. If the date is internal and the cost of lateness is low, P50 may be enough. Whatever is chosen, state it, and keep the contingency it implies visible rather than hiding it inside activity durations.

The GAO Schedule Assessment Guide includes a schedule risk analysis among its best practices for a reliable schedule, and many public owners ask for one on large projects.

Common mistakes

  • Using generic duration ranges and presenting the result as a contingency basis.
  • Quoting a percentile without the number of iterations, the distribution or the seed.
  • Simulating a schedule with weak logic. Open ends and constraints cap the paths that can move.
  • Leaving out correlation, which narrows the distribution and makes P80 look too early.
  • Treating P80 as a promise. It describes the model, not the future.
  • Running the analysis once at baseline and never again as the job changes.

How os.construction handles it

Schedule analysis runs 100 to 20,000 seeded iterations with PERT or triangular sampling. It returns P10, P20, P50, P80, P90 and P95, an S-curve, the probability of meeting the deterministic finish, and the top drivers by criticality index and sensitivity. It states that durations use the engine’s default ranges, not estimates from the people doing the work.

FAQ

What does P80 mean in a schedule risk analysis?

P80 is the finish date that 80% of the simulated outcomes meet or beat, so there's an estimated 80% chance of finishing on or before it, given the model's assumptions. It's a percentile of the simulated finish dates, not a guarantee.

What's the difference between P50 and P80?

P50 is the median simulated finish: half the iterations finish earlier, half later. P80 is later and more conservative. The gap between them shows how wide the risk is. Owners often plan or commit to a higher percentile such as P80 when the cost of finishing late is high.

Why is the deterministic CPM date often early?

Where several paths join, the finish waits for the latest one. Even if each path alone is as likely to be early as late, the chance that all of them are on time together is lower, so the simulated distribution tends to sit later than the single CPM date. This is often called merge bias.

How many iterations are enough?

Enough that the percentiles you report stop moving noticeably when you run more. A thousand or more is common for project schedules. Report the number of iterations, the distributions used and the random seed so the result can be reproduced.

What is the criticality index?

It's the percentage of iterations in which an activity was on the critical path. An activity with a high criticality index often drives completion even if it isn't critical in the deterministic schedule, so it's a good candidate for risk response.

Sources

Open file