Using earned value management to predict buffer penetration in critical chain project management
Delays in projects are well known to most organizations. Critical Chain Project Management is a new method that adds a project “buffer” to secure the final completion date. The buffer protects against delays in tasks, including those arising from feature creep. However, if buffer penetration is excessive, action is taken to protect the critical chain so that the overall project schedule will not be compromised. To help predict such outcomes, we employ Earned Value Management techniques and show how measures of earned value such as Cost Performance Index (CPI) and Schedule Performance Index (SPI) can be used to analyze buffer penetration. Individual judgments of remaining task duration are replaced with recognized principles offering uniformity, traceability, and measurability. Our results indicate that Earned Value Management provides an early warning of feature creep and consequential buffer penetration.
Critical Chain and Earned Value Management are both established methods of project management. Critical Chain removes “buffers” between activities in the critical chain and replaces them with a single project buffer. Earned Value Management predates Critical Chain and measures project performance based on the cost of work performed against the earned value of that work.
There is nothing mutually exclusive about the two processes with the possible exception that one may risk driving conflicting behaviors if both systems are used for the same purposes. Success to date in merging the two approaches has typically been accomplished by treating earned value reporting as simply a necessary condition of the contract, and using the data gathering of earned value to satisfy that need. Operational decisions and assessments of the health of their projects—the actual management of the project—are driven by Critical Chains buffer management process. As time has gone on, the trend (where both have been assessed) seems to have been to replace EV's predictive tools (and other traditional methods of tracking projects) with buffer management. One problem that arises with buffer management is how to realize early what the buffer penetration will be for a task, and whether the buffer will therefore be insufficient protection against delays in that task. The presentation considers a method based on Earned Value Management and a model of project life cycles that provides a mechanism for predicting buffer penetration in a timely manner. We claim that extension of the early requirements gathering phase in the project life cycle often results in extension of the overall duration of the project as a whole because the extended requirements gathering implies extended requirements (“feature creep”). By monitoring such extensions in the requirements phase and thus “feature creep” at an early point in project execution, delays in the project can be better predicted and contingency plans can be formulated. The paper also shows how to determine the buffer by utilizing the three point estimate technique in order to determine the probability interval for the estimate. The difference between the plan for the minimum and maximum estimate will form the buffer interval.
Solving the Problem of Delays
Late projects seem to be the most common experience among project leaders and project sponsors. Some people say that we have to live with this, because the future is by its nature unpredictable based on the uncertainty of estimates. One of the answers in Project Management (PM) has been to search for the right estimate and to develop extensive methods in order to master the estimating skill. When taking a closer look at many of these methods you will find that they contain a lot of environment factors that overrides the “scientific” calculations and that they on top add a huge uncertainty factor. It is not the intention of this paper to debate these methods. The approach in this paper is to suggest a way to use the result of the used method.
The statement of the paper is: There is not an exact right estimate, what is needed is the commitment to meet the estimate that is agreed upon.
In order to meet the agreed estimate, you need to detect delays. This paper presents two methods—Both of them addresses delays and proposes measurements in order to impose management actions. These methods are Earned Value Management or C/SCSC DoD 7000.2 and Critical Chain Project Management. The first method is mainly a follow-up and prediction method and the second is mainly a planning method. They are both presented in the end of this paper.
It would be very nice if there were a common ground where the planning method and the follow up and prediction method could meet. Because this would give us a total control for the cyclic process of planning/replanning and the follow up.
Safety Buffer in Critical Chain
In the Critical Chain method you create a safety buffer based on the following reasoning:
Exhibit 5 shows a left-skewed probability distribution as a function of the activity duration. Uncertainties exists within all projects, you cannot plan away all uncertainty. People are not blind to this fact and add contingency time during the planning phase. There are several mechanisms that make people add contingency time (e.g., estimates are given based on worst past experiences) to protect themselves from management reductions, there are little or no reward for finishing early and no one wants to be known as the person who delivers late.
In Exhibit 5, the median is shown indicating a 50% (accumulated) probability of finishing an activity earlier or at that point. But who is willing to commit to an estimate with only a 50% chance of success? Normally people can commit to an estimate of, for example, 80–90% probability. The difference between the two, the contingency time, is added safety. Recommendations on buffer sizing use statistics to develop relatively simple rules, sizing the buffers (project and feeding buffers) to one-half the buffered path task length. The method considers the statistical rule saying that the uncertainty of the sum of the events is much less than the sum of the uncertainty for each event. This makes sense since you should expect both positive and negative variations for each event.
Safety Buffer based on Earned Value Management
Since Earned Value Management is a cost based method it is very convenient to use the Three Point Estimate in order to establish a probable cost interval. When planning you need to decide to what cost level you intend to plan against.
Now inspired from Critical Chain you may use a large portion of the uncertainty to form a safety buffer. The time length will of course be dependent on the number of resources available during the period where you are working in the buffer zone.
When using the Earned Value Management approach you are provided with feedback regarding worked hours and Estimate At Completion. Hence, you can calculate the estimated performance ratio and analyze the used buffer against it.
Traffic Light Indicators
The traffic-light system provides a simple intuitive way to evaluate the status of a single project or an entire portfolio of projects. The lights show the status based on real data. It combines both the earned value method and the critical chain method.
Exhibit 6. EVM Estimate
1. The Management Reserve is the remaining allocated project budget compared to the estimated cost at completion.
2. Milestone Planning is the deviation from the ideal when estimated activities overrun completion dates or when milestones are missing.
3. The planned ratio of the remaining estimated time.
4. The amount of reported time is compared to planned effort for the last four weeks.
5. The risk exposure is lowered when the current activities are successfully finalized and when the current work is concentrated to a few activities.
6. The buffer indicator shows the penetration of the buffer and if the project has difficulties to manage green light for resource planning.
Exhibit 9. Ackugram With Properties
The relation between Earned Value Management, Critical Chain and the Traffic-light Indicators can be found in Exhibit 9.
Critical Chain (CC): Theory of Constraints Applied on Project Management
A number of management philosophies have been developed during the last decade, trying to define necessary actions that companies have to take in order to stay competitive in the new global economy. Critical Chain is based on one of them, Theory of Constraints (TOC), originally developed for the production industry (Goldratt, 1990). One of the underlying thoughts of TOC is not to initiate many local improvements that requires attention, time and money and in the end does not contribute to a global improvement. Instead, the focus should be put on the true constraint of the system, the weakest link, as the way to achieve considerable lasting improvements.
CC is a mixture of new ways of thinking combined with traditional project management methods, system thinking and human behavior theories. CC has the expectations of consequently achieving project goals.
Its focus is:
• Handling uncertainty
• Removing the undesired affects of human behavior
• Reducing the level of multitasking (probably the main cause to wasted lead-time).
In practice, single project:
• Removing built-in safety time included in each single activity estimate
• Identifying the bottleneck of the single project, the critical chain, taking both logical and resource dependencies into account
• Placing a part of this safety in a buffer at the end of the critical chain protecting what is important, the project end date
• Protect the CC from variances in noncritical paths by adding feeding buffers at each integration point
• Plan the project as late as possible to reduce the level of multitasking and improve cash-flow
• Implement Relay Runner attitude
• Follow-up using “remaining time” as input from project members to produce forecast reports, e.g., scatter plots. Move away from follow-up against dates.
In practice, multiproject:
• Identifying the bottleneck of the multiproject organization, e.g., a certain type of resource, a department, a machine or even a policy. This bottleneck dictates the pace for the organization to achieve its goals
• Prioritize possible business opportunities
• Create a project portfolio plan (synchronization plan) and introduce new projects into the system based on the availability of the bottleneck (constraint). This will eliminate some of the resource conflicts. Resource owners use the penetration of buffers as a tool for prioritizing resources, solving remaining conflicts.
Overview of Earned value
The earned value system requires as initial input a project plan containing a finite number of tasks. Each task should last approximately one reporting period, i.e., have a definite planned start and end date, although as we will see this is not absolutely mandatory; for each task, the planned effort per reporting period must be indicated.
During project execution, the following tracking data are required as input:
Each task must be marked as either Not Started, Under Work, or Completed.
For each task, the spent effort for each reporting period must be entered.
Estimations: Do an estimate for the “percentage effort remaining” per task as an input. The problem is that it is very difficult to provide good estimates of work remaining, particularly early on during a task execution. That is why it is standard practice to estimate a task as follows:
• Not Started counts for 0% complete
• Under Work counts for 50% complete
• Completed counts for 100% complete.
Although these values may surprise some, it turns out that they are good enough to produce very accurate measures and predictions. The accuracy is enhanced if task duration is kept to more or less one reporting period.
The earned value system produces several outputs as described next.
BCWS: Budgeted Cost of Work Scheduled. This represents the cumulative planned spending on a project at a given time. At the end of plan, this value indicates the project's planned total cost.
ACWP: Actual Cost of Work Performed. This is an actual value, i.e., is a result of the tracking project spending. It is the cumulative cost of the work performed up to a given time. At the end of project execution this value represents the project's total cost.
BCWP: Budgeted Cost of Work Performed. This cumulative value indicates how much the work that has been done should have cost according to the original plan. This value can be compared with ACWP. If BCWP is greater than ACWP, the project is producing more than planned for cost. If BCWP is less than ACWP, then the work costs more than planned.
CPI: Cost Performance Index: This is the ratio of BCWP/ACWP.
If the CPI is greater than 1, the project is ahead of budget, if CPI equals 1, it is on budget, if CPI is less than 1, it is over budget. It is worth mentioning that statistically, project CPI will never significantly increase once the 15% spending mark is reached. Unfortunately, CPI can always decrease.
EAC: Estimate At Completion. This value predicts the total cost of the project based on current knowledge. It is an increasingly accurate indicator of total project cost.
SPI: Schedule Performance Index: This is the ratio of BCWP/BCWS.
If the SPI is greater than 1, the project is spending faster than scheduled, if CPI equals 1, it is spending on schedule, if CPI is less than 1, it is spending less than scheduled.
Proceedings of the Project Management Institute Annual Seminars & Symposium
October 3–10, 2002 • San Antonio, Texas, USA