CP

A,C,F,G

that occurring in just a minute. Instead, let's look at the pessimistic estimate and its impact of project duration and the critical path. By inserting the pessimistic durations into the CPM matrix we get the results shown in Table 3-31.

Our duration is now extended to 33 with the same critical path. This seems long, doesn't it? Again, we will examine the probability of such an estimate in just a minute. The problem is that we do not expect all of the pessimistic estimates to occur. Instead, if we operate this as an experiment numerous times, we would expect the mean value to occur. This is shown in Table 3-32.

As we did in the budgeting, we can now represent the project duration as a normal distribution centered around the mean project duration. The question is how we calculate the standard deviation to use. The answer is that we will only use the variances of the task that are on the critical path. Examine Table 3-33.

Both of these estimates are unreasonable. To establish a better estimate of the project duration, we will use the project mean and standard deviation to establish a project duration that we are 95 percent sure of meeting. We work this the same way as we did with the budget. This time we will call

TABLE 3-32 Mean Values

Pred.

Dur. 