Schedule Risk Assessment and Parametric Modeling
Schedule Risk Assessment: During a SCRAM Review a schedule health check is performed to evaluate the quality of a schedule to determine its suitability for running a Monte Carlo simulation. The health check examines the construction and logic of the schedule under review and includes an analysis of the schedule work breakdown structure, logic, dependencies, constraints and schedule float.
A Monte Carlo analysis is then performed on the critical path and near critical path tasks and work packages in the schedule; an example of the output of this type of analysis is shown in Figure 2. Tasks are allocated three-point estimates based on the assessed level of risk. During a SCRAM assessment, risks and problems identified from each of the RCASS categories discussed above provide input into these probability estimates. The three-point estimate (pessimistic, optimistic, most likely) can be applied with either a generic risk multiplier (derived from past experience) across all like tasks or a risk factor based on a task-by-task risk assessment.
The result of the Monte Carlo analysis is a distribution showing the percentage probability of achievement for any planned delivery date. If the planned program delivery is on the left side of the program completion distribution curve, there is cause for concern, depending on the degree of risk the stakeholders are prepared to accept. Projects should use the results of the analysis to develop mitigation plans to ensure that the risks don’t become reality.
Figure 2. Monte Carlo Schedule Analysis
Another consideration of a SCRAM schedule health check is the allocation of schedule contingency. Some contingency is recommended for the inevitable rework. It is important to have some schedule contingency distributed throughout the schedule accompanying the higher risk tasks instead of a cumulative buffer at the end of the schedule before delivery or held as management reserve. This will allow some slippage to occur during development without disrupting subsequent successor task(s) scheduling.
Software Parametric Modeling: SCRAM can be applied at any point during the system engineering or project lifecycle. For the software development elements of a program, a schedule forecast tool is used to assess existing schedule estimates. SCRAM includes this forecasting activity because software is a common schedule driver for complex systems and software durations are almost always optimistic. While SCRAM is not dedicated to a specific forecasting tool, the preference is to use a tool that uses objective software metric data ‘actuals’ that reflect the development organization’s current performance or productivity.
The inputs to the model are size (usually estimated source lines of code and actual code complete to date), major milestones planned and completed, staffing planned and actual, and defects discovered.
Figure 3 below shows a typical output from a modeling tool.
Figure 3. Parametric Modeling