It's better to burn ????????? down than to fade away ??

Introduction to ‘burndown’ charts

A burndown graph helps to predict how risk exposure will change as a project progresses. Although the actual risk exposure will be dependent on which risks impact, which treatment actions are successful and the emergence of unanticipated events, it provides a profile of what is anticipated at a particular point in time (based on the information available).

As circumstances change over the course of the project, this ‘baseline’ prediction can be compared to actual risk exposure informed by the latest updated risk register/ analysis and the profile going forward from this point is refined to reflect how the reality of the past months could play out in the newly predicted future. It is useful to note any significant events that have occurred that caused deviation from the original predictions to maintain an audit trail of significant impacts and take lessons for later stages of the project or for similar future schemes.

Having an appreciation of where significant exposure is and when it is expected to reduce will help to drive decisions around deploying available resources to where they will have the greatest influence.

How to calculate the data for a typical cost risk burndown chart

A cost risk burndown chart can be established arithmetically in a relatively simple way using the mean of the cost impacts of each risk, weighted by the probability of it happening. The sum of all the identified risks gives an appreciation of the ‘current’ cost risk exposure for the project (based on the risks identified) and this is the starting point for the burndown chart.

As a project progresses, treatment actions are delivered with the aim to address the identified risk and drive the exposure to the ‘residual’ value once they have all been delivered and the remaining risk accepted. This ‘residual’ value, as any assessment, is an estimate and may not turn out as expected. The achievement of the anticipated ‘residual’ value relies on the planned treatment action being implemented as planned and having the desired influence. Further granularity for the burndown profile can be achieved by considering the influence of individual treatment actions if an effective evaluation and likely date can be established for the expected influence of each.

There is also a point at which the risk could no longer impact the project (e.g., because the activity that the risk is associated with has been completed). This is often referred to as the ‘Expiry’ date.

The following table attempts to summarise the lifecycle of a single risk:

Table 1: Explanation for each stage in the lifecycle of a risk event

For a single risk event, a simple example may look like this:

Table 2: Example of a typical cost risk evaluation

The Mean ‘Current’ Cost Risk Exposure Value for a risk is calculated by weighting the average of the impacts by the probability of the risk occurring. The above example, adopting a triangular distribution, is calculated in this way:

25% x (￡100k+￡200k+￡600k)/3

???????????????????????????????????????????????????????????????????????????25% x ￡300k

???????????????????????????????????????????????????????????????????????????????????????￡75k

Equation 1: Current evaluation and exposure of the example risk

On completion of treatment actions, if the probability reduces to 10% but impacts remain the same, the Mean ‘Residual’ Cost Risk Exposure Value will look like this:

10% x (￡100k+￡200k+￡600k)/3

??????????????????????????????????????????????????????????????????????????? 10% x ￡300k

????????????????????????????????????????????????????????????????????????????????????????￡30k

Equation 2: Residual evaluation and exposure of the example risk

The above summarised in the lifecycle of this individual risk would look like this:

This would look like this in a chart showing just this single risk:

Figure 4: Burndown of a single-risk event

A typical burndown chart takes the sum of the cost risk exposure for all identified risks as the starting point on the current date. It then establishes what risk reduction is achieved in each period over the course of the project to show the overall risk exposure profile at each date; this represents the collective risk exposure as risks are treated and/ or retired during the lifecycle of the project. Ultimately, the profile should reach zero at the end of the project, though there may be certain risks that are carried over into the operational phase and so this may not always be the case.

A typical risk burndown chart may look something like the following:

Figure 5: Burndown of the overall cost risk exposure of a project

The line on the above chart represents the ‘Predicted’ profile. This shows the cost risk exposure burndown and is defined by the ‘current’ evaluation along with dates on which ‘residual’ and risk retirement are expected to be achieved. In addition to this, further lines may be added to track how ‘Actual’ data differs from the original prediction. It can be further annotated to explain any significant delta from what was expected in any period. A further line may be added to show, based on previous deviations from the originally predicted data, how the profile going forward is now expected to track.

Tracking achievement of opportunities

Risks are considered to include both ‘Threats’ and ‘Opportunities’; so, consideration needs to be given to whether these influences should be taken collectively, or a separate chart created for each. There is no reason that the overall cost risk exposure could not be taken as a single total risk exposure curve where threats and opportunities are combined but I typically keep these separate and individually track the achievement of opportunities and the progress in reducing exposure to threats. Clearly, if taken on its own, an opportunity chart would not burn down and so is more of an opportunity ‘burn-up’ chart.

Manhattan exposure chart

Another similar approach which helps to highlight the hotspots of risk exposure during a project is a ‘Manhattan’ profile. This style of graph is so-called because it resembles the Manhattan skyline. Instead of being a ‘Descending Cumulative’ distribution like a burndown, this is a histogram which shows the specific risk exposure for each individual time period and helps to focus attention on those periods where the most exposure is and so potentially the greatest influence can be gained from effective risk management. It is developed in much the same way as the burndown chart, but consideration is also taken of the ‘Trigger date’ of each risk.

A typical ‘Manhattan’ profile looks like the following chart:

Figure 8: Example of a Manhattan profile

Establishing a Burndown chart based on Probabilistic Cash Flow

Schedule modelling software generally has the functionality to produce a Probabilistic Cash Flow (PCF) chart which provides useful insight into the spend profile of a project which accounts for uncertainty applied to the model. Where the approach described above merely looks at the mean value of the uncertainty applied in terms of risks and uncertainty applied to an estimate, using the PCF data to construct a burndown chart enables a view to be gained of various percentiles in the same way as the overall output of the model (e.g., a P80 burndown can be established).

The PCF typically looks like the following chart and combines the cumulative profile of the cash flow at certain percentiles overlaid on a Manhattan-type histogram which shows the spend in each period. A burndown can be established using this data by starting with the current exposure and deducting the reduction in each period as described by the PCF data.

Figure 6: Probabilistic Cash Flow Chart

Where this output has previously only been available for fully integrated cost and schedule models based on a fully resource-loaded schedule, Safran now provides functionality which enables a cost model and schedule model to be linked without the need to resource-load the schedule. This has previously represented a significant barrier to constructing an integrated cost and schedule model.

Schedule Burndown Chart

I have recently investigated for a client how a schedule risk burndown could be produced based on the outputs of a schedule risk analysis. Although the principles are similar to a cost burndown, a schedule burndown cannot be derived in quite the same way as cost since the schedule risk exposure of an individual risk or duration uncertainty is of little value without considering the influence of float and logic. A significant schedule risk mapped to a task with a large amount of float may have less of an influence than a less significant risk mapped to a task on the critical path.

In order to understand the contribution of individual risks on the output of a schedule analysis, I previously needed to re-run the analysis multiple times, excluding a single risk on each run. To streamline this process, I used the tornado chart to inform those risks that were having the greatest influence and stopping at the point where re-running the model without a specific risk resulted in minimal difference.

To determine the ‘expiry’ of each risk, I simply took the completion dates of the latest task that each risk was mapped to. To appreciate the ‘residual’ influence of each risk, I repeated the exercise for a post-treatment model.

This is a relatively crude approach but provides an insight into how the schedule risk exposure may reduce as the project progresses and was a useful graphic to inform the decision-makers where the greatest risk exposure was during the delivery of the project and where the greatest influence could be achieved from implementing controls to improve project timescales and achieve greater schedule certainly.

One of the really useful features of Safran Risk software is that I no longer need to run the model multiple times excluding a risk each time to understand the influence of each individual risk. The ‘Sensitivity Analysis’ functionality does this automatically which saves a significant amount of time.

Displayed in a tornado-style chart, this shows how much time each risk or duration uncertainty is adding to the output of the analysis, as can be seen below:

Figure 7: Example of a Safran ‘Sensitivity Analysis’ Tornado chart

A risk burndown chart provides useful insight during the delivery of a project and helps decision makers to understand at what point the risk exposure is likely to reduce in response to completion of the more uncertain aspects of delivery. Furthermore, by gaining this insight at an early stage, plans can be developed and implemented to address this uncertainty with a view to achieving greater predictability of the costs and timescales for completion.