Saturday, June 27, 2015

Primavera Risk Sensitivity Analysis with Torando Diagram




In my interactions with management professionals across various industries, questions come in - what risk sensitivity analysis is and how to conduct them. Professionals cutting across industries - specifically in construction, manufacturing and IT – are taking a keen interest in risk management as I have seen, since last year. 

When I introduce Torando diagram - a special bar chart used in sensitivity analysis for comparing relative importance of activities (or activity costs) as compared to the project's objectives, they get a certain idea, but do not fully understand.

In this post, I’ll explain about risk sensitivity analysis along with Torando Diagram in a practical way with the help of Oracle Primavera Risk Analysis software. This also will be of helpful for aspiring professionals who are taking the Risk Management Professional (RMP) exam or Project Management Professional (PMP) exam as both in Standard for Risk Management and PMBOK Guide, torando diagram has been mentioned to be risk sensitivity analysis technique.

Risk and Risk Sensitivity Analysis:
First, a few lines on risk:
  • It can be uncertain event or condition
  • It is in the future and has not occurred when identified and analyzed
  • If it occurs, it will have an impact or effect
  • It impacts positively or negatively the project’s objective. (Project objective can be schedule objective, cost objective, scope objective and so on)

Now, what is risk sensitivity analysis? Why conduct it?
As per PMBOK Guide, 5th Edition:
Risk Sensitivity analysis helps to determine the risks that will have most potential impact on the project. It informs how the variations in project’s objectives correlate with variations in different uncertainties. Conversely, it helps to find out the extent to which uncertainty of each project element affects the objective being studied, when all other uncertain elements are held at their baseline value. 
So, the key items that we are trying to know by Sensitivity Analysis are:
  1. Which risks will have the highest potential impact on the project?
  2. What is the correlation of uncertainty of each project element on the project objective?
  3. What is the correlation of variations in project objectives with variations in different uncertainties?
Risk sensitivity analysis is a quantitative risk analysis technique. Torando diagram is used to display the data for post the sensitivity analysis. I'll take Item 2 (correlation of uncertainty of each project element on project objective), to explain more in this post. This will be easy and simple to understand.

In sensitivity analysis, you are checking the impact on the project objective, e.g., schedule objective, due to change in duration of a work package or an activity. Also, you can check impact that one activity can have on other activities. As with schedule objective, you can check on the project’s cost objective, too. It means you can check on the impact on the project's cost, if the cost of an activity or cost of a work package changes. 


Tornado Diagram:
Torando diagram/chart  is a horizontal bar chart with high impact activity/work package duration items (or cost items) listed on top and low impact activity/work package duration item (or cost items) listed below in a gradual manner. Because the shape of the sensitivity analysis graph looks like a tornado, it is called torando diagram. A tornado, if presented in a graphical view will look as shown below.

Image Credit - http://www.clker.com/
The sensitivity analysis diagrams also look like torandos. In it, the Y-axis has the uncertainty (duration items or cost items considered) at base value and the X-axis  has the correlation (or spread) of the uncertainty as compared to a project objective. 

Example:
I have used Primavera P6 project-portfolio management software to create a simple project plan. This plan is then imported to Primavera Risk Analysis (PRA), earlier known as Pertmaster. One can create the plan in PRA. However, as I have seen, most use Primavera or MS Project (which also can be used) to create a plan. Hence, taking this route.

The test project has 6 activities (Activity 1 to 6) and 2 milestones (Start and End milestones). There are 3 resources – all labor resources – assigned to these tasks. I have not taken any work package, to keep it minimal and simple. The Gantt Chartt, from Oracle Primavera P6 software is as follows.

Gantt Chart in Primavera P6
Resource 1 and Resource 2 have the following rates. Resource 3 is deliberately not having any standard rate, impact of which we will see in cost sensitivity torando.

Labor Resources - Resources screen in Primavera P6
I have created this plan (Test Project.XER) in Oracle Primavera P6 R8.4 and exported it. Next, I imported this file into Oracle Primavera Risk Analysis software (R8.7). Now, applying a probabilistic distribution for the activities, I'll get the the following chart.

Gantt Chart with Distribution values in Primavera Risk Analysis (Pertmaster)
I have applied triangular distribution, but other distributions such as Beta, Uniform etc can also be applied. In our case, we have the PERT estimate of Minimum duration, maximum duration and most likely duration.

The formula used in triangular distribution is as described below:
  • Minimum duration is 75% of remaining duration
  • Most likely (or simply likely) is 100% of remaining duration
  • Maximum is 125% of remaining duration
It can be understood by looking at few activities. 
1) For “Activity 4”, remaining duration is 4 days; 75% of it will be 3 days. 125% will be 5 days.
2) For “Activity 2”, remaining duration is  3 days; 75% of it will be 2.25 days (~2days) and 125 % will be 3.75 days (~4days)

No baseline has been performed, as I am analyzing the sensitivity of the activities and their associated costs, before doing any baseline.

Now let us check on the analysis using the Tornado diagram. I ran through 1000 iterations by changing the values of activities (1 to 6) in the project, i.e., applied the Latin Hyper-cube Simulation (LHS) technique. One can also use Monte Carlo simulation techniques. But by default in the tool, LHS is enabled. 


Duration Sensitivity:
The Duration Sensitivity of a task/activity is a measure of correlation between its duration the duration of the project. It also can be with respect to another key task/activity or summary task/activity.

Here, we are  checking the correlation between the duration of the activity with that of the project. This is how the duration sensitivity is represented in a torando diagram. Simulation used is latin hypercube.

Duration Sensitivity Torando Diagram in Primavera Risk Analysis
Interpretation
  • Activity 3 and 4 are on critical path and will of course have the most impact on project's duration, i.e., at 67%
  • Milestone Start and Finish have the least impact – at 0%
  • Activity 2 and Activity 6 - have low as well high impact 
  • Looking differently, activities having total float and free float as zero will have high impact, whereas activities with certain total float and/or free float values, will have less impact on the project's duration

Cost Sensitivity:
The cost sensitivity of a task/activity is a measure of correlation between its cost and the cost of the project. It also can be with respect to cost of another key task/activity or summary task/activity.

Here, we are  checking the correlation between the cost of the task with the project. The cost sensitivity is shown with the below torando diagram. Simulation used is latin hypercube.

Cost Sensitivity Torando Diagram in Primavera Risk Analysis
Interpretation: 
  • Activity 6 and 2 have the highest impact on project's cost, i.e., at 59% and 58%, respectively 
  • Activity 1 and Activity 5 have lesser impacts 
  • Activity 3 and 4 have no impact - as we do not have any cost associated with these activities! (Remember for Resource 3, I had not provided any standard rate deliberately)

Conclusion:
Estimates are uncertain. Hence, we apply the probabilistic distribution to get a range and iterate enough to get a set of possible data with the help of Latin Hypercube Simulation technique (or Monte Carlo Simulation technique). With uncertainties in estimates and hence uncertainties in duration of activities or work packages - there will be variations on the project's objectives, such as schedule objective, cost objective. This in turn asks for risk analysis and management. As we saw in this post, torando diagram in risk sensitivity analysis helps to know the impact these uncertainties will have on the project's objectives.



2 comments:

  1. Hi
    Nice article.
    Isn't risk analysis possible in P6?

    ReplyDelete
  2. Can be done partially in P6, not fully. Hence the new course 'Practical RMP with Primavera Risk Analysis'.

    Check the FAQ in here: http://managementyogi.blogspot.in/p/practical-rmp-with-oracle-primavera.html . Have addressed this question.

    ReplyDelete

Any comment is welcome - comments, review or criticism. But off-topic, abusive, defamatory comments will be moderated or may be removed.