Using the Delay Simulator safely
How to model a delay without changing the plan, why compound delays take the maximum rather than the sum, and how to reset.
Running a simulation
Open the simulator controls on the Gantt, choose a task, and set the delay with the slider, up to 30 days. The chart, the Impact Matrix, and every task's float all recompute immediately.
You can delay more than one task at a time. Each task you have delayed shows its current simulated delay next to its name in the picker, so a compound scenario does not lose track of itself.
Nothing is saved
Simulated delays are held separately from the plan of record. They are passed into the scheduling calculation as an overlay, so you see the full downstream consequence, but nothing is written to the project.
This is the point of the feature. If what-if analysis means editing real dates and undoing them afterwards, people stop trusting the plan, because they can never be sure whether what they are looking at is real. Keeping simulation separate is what keeps the plan meaningful.
Compound delays take the maximum, not the sum
When several delayed tasks feed into one task, that task inherits the largest incoming shift, not the total of them.
Two chains running three and five days late push the task they converge on by five days, not eight, because the task was already waiting on both -- the three-day chain finishes while the five-day one is still running. This is correct scheduling behaviour, and it is the case people most often get wrong by hand.
Reading the result and clearing it
Every affected task shows how many days it moved and which upstream task caused the move. Revised dates are computed alongside the planned ones rather than replacing them, so the original plan stays on screen for comparison.
Reset the simulation when you are done. If the decision is to accept the delay, enter it as a real delay on the task rather than leaving the simulation running.
Which plans have the simulator
The Delay Simulator is available on Standard and above.