A Step-by-Step Guide to Risk Analysis in Spider Project Software Using Monte Carlo Simulation
October 15, 2013
The first article provided a general understanding of risk analysis using Monte Carlo simulation.
This article will focus of implementing this simulation in Spider Project software and provide a detailed guide to performing risk analysis using Spider Project. Example is a sample project that goes with a free Spider Project Demo version, sp - Software Purchase.
Monte Carlo simulation is a method of risk analysis that relies on repeated random sampling to obtain numerical results by substituting a range of values, such as activities duration and volume, work load, resources productivities, activities and resources calendars, required costs. All estimates are set by project’s internal experts, and the results are calculated statistically baseâon the results of all simulations.
Risk analysis using Monte Carlo simulation will produce:
- A Monte Carlo probability distribution chart;
- A histogram with accumulated frequencies;
- A criticality index;
- Monte Carlo variance charts/diagrams.
Monte Carlo Risk Analysis in Spider Project
In the beginning of each calculation, the program will mount a baseline version, where all planned values will be substituted with basic (assessment) values set in a Monte Carlo Analysis Setup dialogue box.
A baseline version will be calculated with parameters that had been previously set in a cost setup dialogue box and a scheduling dialogue box (either resource constrained scheduling or scheduling without resource restrictions).
Each subsequent calculation will support the baseline version.
Before each next calculation, all values are randomly generated using a pseudorandom values generator based on the selected value distribution law (see Fig. 1)
Fig. 1. Selecting value distribution
- Triangular (triangle) – Values around most likely are more likely to occur. However, both optimistic and pessimistic values can also occur.
- Beta – Just like the triangular distribution, most likely value and values around the most likely (with a pessimistic skew) are more likely to occur. The density curve will be fairly smooth.
- Log-normal – Just like the beta distribution, most likely value and values around the most likely with a pessimistic skew are more likely to occur, but the density curve will be more vertically stretched.
If the three basic estimates (optimistic, expected / most probable, and pessimistic) will match the planned value in the project, than the planned value will not be changed.
The amended project are be recalculated, the results for all selected activities and phases are saved, and Monte Carlo Analysis proceeds to the next iteration.
Monte Carlo Analysis will calculate a criticality index for each activity within the project. A critical index is a percentage showing how often a particular task was on the critical Path during the analysis. Say, if simulation was repeated 1,000 times and a certain activity joined the critical path 212 times, than criticality index will be 21.2%.
Below is the step-by-step guide to performing risk analysis in Spider Project Software using Monte Carlo simulation:
Step 1. The first thing to do is set the distribution of parameters (optimistic, expected / most probable, and pessimistic values). To do this, in a Select Field dialogue box, check the Monte Carlo checkbox and select the fields for which values will be assigned. We will use the Duration field as an example (see Fig. 1)
Fig. 1. Selecting a field to set the distribution of
Gantt diagram or activities table will show thee columns of the samefield with prefixes opt-, mp- and pes-. The planned and the expected values of this field must be between the optimistic and the pessimistic values (see Fig. 2)
Fig. 2. Setting optimistic, most probable and pessimistic
If there are 3 previously modeled project versions to be analyzed using the three scenarios method, values may be imported from them by selecting Risk Analysis –> Monte Carlo Method –> Import Monte Carlo data from Three scenarios in Activity Gantt menu values (see Fig. 3)
Fig. 3. Importing data from Three scenarios
Step 2. Select activities and phases for which calculation results will be available. To select, check the Monte Carlo Analysis checkbox in Phase Properties or Activity Properties dialogue box. This option can also be enabled in Activity Gantt diagram. In order to do this, highlight the Monte Carlo Analysis column and select Yes where needed (see Fig. 4). If nothing is selected, no Monte Carlo analysis will be done.
Fig. 4. Selecting phases and activities to be calculated
Step 3. Select values and set the necessary parameters in Monte Carlo Method options dialogue box (see Fig. 5)
Fig. 5. Monte Carlo risk analysis setup
Fig. 5.1. Monte Carlo risk analysis setup
Step 3. To start calculation, in Activity Gantt menu select Risk Analysis –> Monte Carlo Method –> Analyze Risks (see Fig. 6)
Fig. 6. Starting Monte Carlo risk analysis
Step 4. To show a Monte Carlo probability distribution chart, right-click the grey area on the left of the activity name and select Probability Distribution –> Monte Carlo Method (see Fig. 7)
Fig. 7. Opening a probability distribution chart
The probability distribution chart for selected values will appear (see Figures 8 to 8.2).
Fig. 8. Project finish probability distribution chart
Fig. 8.1. Project duration probability distribution chart
Fig. 8.2. Project total cost probability distribution chart
The above charts show the connection between indicator values and probabilities. Let us take the Total Cost field as an example. Risk analysis using Monte Carlo simulation shows that the probability that the project will cost 36.406 currency units is 82.2%. With the red line moving, indicator values and probabilities will change, enabling the user to make the assessment of the accuracy of total project cost predictions.
Further, these indicators can be used to:
We have discussed step by step how to simulate risks using Monte Carlo method in Spider Project software package. As you can see, this method enables the user to obtain a fairly accurate estimate the probability of constraints under the project. Moreover, its use is not time-consuming, especially if you already have three versions of the project (optimistic, pessimistic and expected/most probable).
Summing up, I wish that this added functionality of Spider Project brings more success to your projects!