Risk Analysis

 The Risk Analysis in CS Project is particularly viable since the scheduling engine is extremely fast. This means that you can run many iterations in a short space of time (e.g. a Risk Analysis of 500 iterations done on a Pentium II 266 MHz running Windows 2000 takes less than 45 seconds for a project of 400 activities and 800 assignments, and less than 15 seconds for a project with 400 activities and no assignments).

 The Risk Analysis in CS Project is designed to calculate a probability distribution for the project finish date as well as the overall project cost. In fact two curves are drawn - a normal distribution displaying the number of times that the project finish date (or cost) falls within a certain period, and the probability curve. What this allows you to determine is the most likely project finish date (peak or mode of the normal distribution) as well as the probability of achieving a certain project finish date (or cost). Conversely, if you want to be e.g. 90% certain of achieving your project finish date, you can determine what this date is. The mean value that is calculated is the arithmetic mean, whereas the mode is the peak of the distribution curve. What CS Project does not do, is give you a plan that will finish on this desired date; the plan is still the original plan with the estimated durations, i.e. its finish date is different (usually earlier). Statistically you will finish on the desired date (with the relevant probability) when working according to the given plan.

The following parameters need to be entered before the Risk Analysis is started:

1. Duration - this is the most likely duration of the activity and is the same duration that is entered for normal scheduling.

2. Duration (HI Range%) and Duration (Low Range %) - these fields represent the pessimistic and optimistic durations as a percentage of the most likely duration. They can only be entered in the activity table (spreadsheet).

3. Cost (Hi Range %) and Cost (Low Range %) - these fields represent the pessimistic and optimistic costs as a percentage of the most likely cost (both activity other costs and resource unit costs). They can only be entered in the activity table (spreadsheet) and resource table (spreadsheet).

When you start the Risk Analysis from the Tools menu, you can set the following parameters:

1. Number of Trials - this is the number of times that the Monte Carlo simulation will be run, i.e. it is the number of times that the entire schedule will be re-calculated.

2. Date interval - this can be any time unit. It determines the scale on which the probability curve is drawn, hence a finer resolution will give a more accurate curve. If you set e.g. a daily scale, then the distributions are calculated on a daily basis. This is similar to changing the date scale on the histogram.

3. Cost interval - this is similar to the time interval.

 The Risk Analysis will recalculate the schedule as often as you have specified. For every iteration it changes the duration and costs of every single activity and resource. These durations (and costs) for each activity and resource are selected at random in between the limits (low and high ranges) that you have specified using a triangular distribution. This means that most duration estimates will fall on or close to the estimated duration and very few will be on or close to the limiting values. The more iterations you run, the more accurate the analysis will be, especially if you select a small time unit. However, there is a cut-off point beyond which you gain little in accuracy and waste a lot of computing time. This cut-off point will depend on the size of the project, its network logic etc. (e.g. if you had one activity, then 50 iterations would produce a good result, whereas with other projects you might need to run 1000 or even 5000 iterations).