When a Schedule Risk Assessment (SRA) runs, the tool simulates duration uncertainty by varying the remaining duration of activities between their defined Best-Case and Worst-Case duration estimates. Rather than selecting durations completely at random, the analysis uses a probability distribution algorithm to determine how likely each value within that range is to be selected.
One way to visualize this process is to imagine a roulette wheel where each number on the wheel represents a possible duration value between a task’s Best-Case and Worst-Case duration estimates.
When the wheel is spun, the selected value becomes the activity's remaining duration for that simulation iteration. Depending on the distribution algorithm being used for the task, some duration values may appear on the wheel more frequently than others, making them more likely to be selected.
During each simulation iteration, the tool “spins the wheel” on every activity has a valid Best-Case, Worst-Case, and Most-Likely* duration estimate defined.
*Schedule Risk Assessments require a Most-Likely duration estimate to be identified for all tasks. Typically, the current Remaining Duration of tasks is used as their Most-Likely duration estimate. This value is used in conjunction with the Best-Case and Worst-Case duration estimates when applying probability distributions during each simulation iteration.
Distribution Algorithms
The SSI Schedule Risk Analysis tool supports the following probability distribution algorithms for modeling Duration Uncertainty:
Triangular Distribution – Uses the Best-Case, Most-Likely, and Worst-Case duration estimates to form a triangular-shaped probability distribution. Duration values near the Most-Likely estimate are more likely to be selected than values near the Best-Case or Worst-Case estimates. This is the most commonly used distribution algorithm for Schedule Risk Assessments.
Beta PERT Distribution – Uses the Best-Case, Most-Likely, and Worst-Case duration estimates to create a smooth probability distribution that places greater emphasis on the Most-Likely duration. Compared to the Triangular distribution, Beta PERT generally produces fewer extreme outcomes and is often considered a more realistic representation of task duration uncertainty.
Normal Distribution – Creates a bell-shaped probability distribution centered around the Most-Likely duration. Duration values near the Most-Likely estimate are significantly more likely to be selected than values near the Best-Case or Worst-Case estimates. This distribution is useful when activity durations are expected to cluster around a central value and extreme outcomes are relatively uncommon.
Uniform Distribution – Assumes that every duration value between the Best-Case and Worst-Case estimates is equally likely to occur. Unlike the other distributions, the Uniform distribution does not favor the Most-Likely duration and treats all possible duration values within the range the same.
