What is the standard approach to planning work that might or might not be needed based on certain circumstances?

Question

I am wondering what the typical approach would be here:

In a project, we are renewing old machines and therefore the new ones will need to be integrated into an older system. That might or might not cause problems, and we know the time needed to fix these problems. However, our management requires planning and data, and I struggle with how to reflect the situation:

• If all goes well, the project will take 100 MDs
• If there are problems in all (let's say 10 areas), then the project will take 300 MDs.

Now, if I plan the project for the worst possible scenario, 300 MDs, then - if there is less problems - the progress and effort spent will not really match. Our management looks at % Done and also effort spent.

Answer 1

You know your best case estimate and your worst case estimate. To get an estimate for the most likely time it will take to complete the project, you look at each possible problem and factor in how likely that problem is to occur and what time it costs to resolve if it does occur. The time you should put in your estimate to account for that problem is the likelihood times the effort.

So, if each of the 10 problems costs you 20 MD to solve and they have a 75% chance of happening, then it is likely that those 10 problems together will cost you 10 * 75% * 20 MD = 150 MD. This gives you a total likely effort of 250 MD (by adding the 100 MD it takes if there are no problems at all).

You then communicate to management that the project will take at least 100 MD, is likely to be finished in 250 MD but is not expected to take longer than 300 MD.

As long as the effort spent in relation to the reported % Done walls within that window of 100 to 300 MD total effort, your project should be considered to be on target.

Answer 2

This is standard risk management.

1. First and most important never quote a number to management; express the estimate as an estimate with a confidence interval - it will take from 100 to 300 MD. (BartVanIngenSchenau describes a good way to model this; I was in the middle of composing a similar answer, but I defer to that excellent summary.) Consider updating management with the standard Planned/Projected/Actual (Planned is the initial plan, Projected is the current plan adapted for any changes/risks/issues, Actual is the performance % done).
   NOTE: the key here is not that management needs to be fully trained in project management assumptions & methodologies - the key is that they have confidence that you are actively managing the project and confidence that you will deliver the project at/under the projected schedule/cost. You earn that confidence by convincing them that you've planned for these eventualities.

2. Include reserve time in your project plan. This deliverable will take 250(+/-150) days; the estimated delivery date for the deliverable is start +250, but the project includes an additional 150 days of management reserve that can be allocated to these risk if they become issues.

3. Include these items in your risk register. Plan in advance, inform management, and request management intervention when appropriate to avoid or mitigate these risks. ("Integration of machine 1 involved complications that we had planned for, but had hoped wouldn't occur. Machines 2, 4 and 7 are similar and may have similar issues; if so, we can either use some of the time in management reserve, or we can bring on a short term expert surge resources to reduce the time/effort. The impact on cost and schedule is $X and Y MD. While I recommend using reserve schedule, it is your decision.....")