# Example Risk Outputs

## 1 CONTENT

The content that follows demonstrates the typical type of risk model outputs that would accompany an Integrated Cost & Schedule Risk Analysis (IRA). Please note that some of these types of outputs are predicated on the ability to gather the necessary information from the project team. Further, some types of outputs are not possible without the full and unhindered use of the Risk Integration Management Pty Ltd IRA toolset. Outputs related to incorporation of revenue streams were not requested in the request for tender documentation for QCLNG, and as such, have not been priced into our proposal. Description of these outputs is for information only.

Where practical to do so, each graph has been accompanied by a brief explanation of how to interpret the information contained therein. The data presented in this document has on occasion been deliberately crafted to demonstrate particular processes & scenarios. In all instances, it is based on a purely fictional project where the proportion of cost and schedule uncertainties have been sometimes randomly allocated. Therefore, measured outcomes are by no means typical of normal project behaviours.

### 3.1 PROBABILISTIC ANALYSIS RESULTS back to content

#### 3.1.1 PROBABILISTIC SCHEDULE OUTCOMES back to content

The following graphic shows the probabilistic date outcomes for a particular milestone within our example plan.

As is shown in the above table, the deterministic completion of 24-Apr-14 had only a 3.5% probability of being achieved. Analysis suggests that to be 10% confident of completing the project on time, a contingent target of 02-Jun-14 should be set. To increase confidences to 50% or 90%, contingent date targets of 10-Sep-14 and 29-Dec-14 should be set respectively.

Relative to the data date of the plan (02-Apr-09), the P90 outcome of 29-Dec-14 represents a period of 2097 days. When compared to the deterministic project duration of 1848 days, the recommended P90 outcome represents a 13.47% schedule contingency requirement.

At approximately +/-5%, the spread of the results from the P10 to the P90 compared to the P50 outcome is relatively narrow.

#### 3.1.2 PROBABILISTIC COST OUTCOMES back to content

The following graphic shows the probabilistic cost outcomes for the entire project of our example plan.

As is shown in the above table, the deterministic project cost of ~$68 million had a 35% probability of being achieved. Analysis suggests that to be 10% confident of completing the project within budget, a contingent budget of ~$57 million would be appropriate. To increase confidences to 50% or 90%, contingent budgets of ~$75 million and ~$138 should be set respectively.

Relative to the deterministic price of the project (~$68 million), the P90 outcome of ~$138 million represents a required contingency of ~$70 million. This outcome represents a contingency requirement of over 103% of the original price.

As can be seen, the cost distribution is clearly bi-modal (two peaks). This indicates that there is a significant event or uncertainty that is clearly influencing the outcomes of the analysis. In the course of a normal analysis, this would be investigated for validity during the quality assurance checks prior to reporting. This skew of the data is also clearly present in the +/- variance from P50, with a much larger percentage shift to the P90 confidence interval relative to the P10.

#### 3.2.1 DURATION SENSITIVITY TORNADO back to content

Duration Sensitivity tracks the correlation between change in an activity’s duration and change in the duration from the data date to the milestone of interest. The higher the correlation, the more likely the activity is to influence the date of the milestone.

Criticality measures the percentage of Monte Carlo iterations in which a task was on the critical path to the milestone of interest. Clearly, the higher the percentage, the more likely it is that the activities on that path will influence the target milestone end date.

Although Duration Sensitivity measures the relatedness between the duration of a task and the finish date of the target milestone, it is possible for an activity to have a high duration sensitivity, but never or seldom to appear on the longest path to milestone completion, even though it may be a logical predecessor. In these circumstances, it would be incorrect to report the task as a driver of the milestone. For this reason, Duration Sensitivity is multiplied by Criticality to obtain a parameter called Cruciality. This parameter ensures that any activity with a high value must be on the critical path much of the time.

#### 3.2.4 COST SENSITIVITY TORNADO back to content

Similar to duration uncertainty, cost uncertainty drivers are determined by measuring the degree of relatedness between changes in an item’s cost, and the change in the cost of the overall project across each iteration of the analysis. Each cost is then ranked as a predictor of changes in the overall project cost and this metric becomes known as the cost sensitivity index.

### 3.3 QUANTITATIVE METRICS & ANALYSIS back to content

#### 3.3.1 CRITICALITY NETWORK DIAGRAM back to content

The following diagram shows a summarization of the logic network of activities that drove the handover of the example pipeline project. Unlike normal critical path analysis, a criticality network diagram accounts for the primary, secondary, tertiary, and subsequent critical paths based on their contribution to the schedule risk model. Criticality measures the percentage of iterations in which a task is on the critical path. Therefore, by using a criticality network diagram, we’re able to see the driving paths through a project, even if the path is only present in instances where particular risks or combinations of risks eventuate. This makes criticality network diagrams enormously more powerful and informative than traditional float analysis.

#### 3.3.2 QUANTIFIED CONTRIBUTIONS back to content

Although sensitivities are useful measures of potential drivers of project outcomes, they are not without their problems. Because sensitivities are only measures of the degree of relatedness between project variables, they don’t provide any measure of the actual extent of contribution, only the extent of contribution relative to other variables. Further, because sensitivity calculations focus only on two project variables (the dependent and independent variables), they’re incapable of accounting for the influence of undeclared independent variables such as correlation between tasks or costs. Because of this, if a very small cost uncertainty were 100% correlated with a very large cost uncertainty, both items would report similar sensitivities.

As such, we don’t rely on sensitivity calculations past the initial examination of model behaviours. To reveal and rank the model drivers, we simulate the project with and without particular variables or classes of uncertainty to measure their probabilistic dollar or time influence by difference. In this way, we’re able to fully quantify the relative contribution of different sources of uncertainty on the model for a given percentile confidence, as is shown in the diagram below:

#### 3.3.3 PROBABILISTIC CASHFLOW, NPV & IRR back to content

In projects where the estimate structure supports the breakout of labour vs. materials, the Integrated Cost & Schedule methodology often makes it possible to produce risk adjusted probabilistic cash flows for project capital expenditure. In the picture below:

** Note: Because this curve was set up to measure Net Present Value (NPV) & Internal Rate of Return (IRR), the P10 and P90 curves have been reversed.

Where the required inputs are available, expenditure can be combined with revenue to produce risk adjusted cashflows. This allows projects to determine at different confidence intervals, the expected timeframe for recouping capital expenditure from expected revenue streams:

Such analysis also lends itself to project Net Present Value (NPV) and Internal Rates of Return (IRR) type calculations: