You Have Heard of Monte Carlo Scheduling. Here is What is Means

Schedules are the way you estimate the work, resources and durations for your project. One way to manage schedule risk for large projects is through the Monte Carlo modeling. Here is an example to demonstrate Monte Carlo. The example is simple, but it requires some focus for it to make sense.

First, you need to assign multiple durations to activities, along with the probabilities of hitting each estimate. You could assign many durations and probabilities for each activity, but let's keep it simple. Let's say you provide three estimates that represent the best case, most likely case and the worst case. For each of these cases, you also assign a probability of each instance occurring. 

For instance, let's use the following estimates:

• 20% change of hitting the best case estimate
• 60% chance of hitting the most likely estimate
• 20% chance of hitting the worst case estimate

You have to estimate these three numbers for each of the major work activities in your schedule (or all activities if it makes sense). For example, you may estimate an activity to most likely take 10 days, with a best case of 5 days and a worst case of 20 days.

Monte Carlo then looks at every activity in your schedule. The simulation uses a random number generator to select an estimate for each activity - best case, worst case or most likely. After the simulation has run for each activity, the entire schedule duration is calculated.

So far, so good. If we ended there, Monte Carlo would not be so interesting. However, the schedule simulation is then re-run. When the random number generator runs for each activity, differing durations will be assigned to each activity (best, worst, most likely), therefore calculating a different end-date.

The schedule model is run hundreds or thousands of times so that the percentages have a chance to play out. For instance, in the example above, if the simulation was run 1000 times, you would expect that each individual activity would be assigned the best case estimate 200 times (20%), the worst case estimate 200 times (20%) and the most likely estimate 600 times (60%).

As the modeling tool randomly picks estimated values based on probabilities, many different project end dates occur. Some show the project completing earlier since many best case estimates are randomly chosen. Some schedules show the project completing later since more worst case are randomly chosen. However, after running the project model 1000 times, a pattern emerges that allows you to estimate the chances of hitting any end date. 

Now instead of saying "we will complete the project by May 31" you would say "there is an 80% likelihood we will complete the project by May 31". If your sponsor wants the project completed by April 30 instead, you can look at your simulation and state "there is only a 45% likelihood we can complete the project by April 30".

Monte Carlo gives you the range of possible end dates and the probability of achieving them.