Budget-Constrained Joint Confidence Level Approach
As long as we have tried to understand and quantify the uncertainty associated with our cost estimates, we have understood that cost risk and schedule risk are highly interrelated. The current emphasis on Joint Confidence Level (JCL) analysis at NASA has focused a great deal of attention on this relationship.
Traditionally, cost estimators have tried to incorporate schedule risk into cost uncertainty assessments by simple means, using the results of a separate schedule risk analysis to stretch and add cost to the project at some inferred burn rate. Often, this schedule-based cost impact is added on top of CER uncertainty and technical risk penalty factors. This results in double- counting, because CER databases generally include schedule risk impacts. One might argue for simply ignoring the schedule risk assessment, however, this would neglect the unique schedule risks of the project (e.g., high degree of concurrency, small margins, dependencies on external events, etc.) in our cost uncertainty assessments.
Another solution to the double counting problem is predicated on the assumption that cost risks and schedule risks can be clearly separated by separating Time-Dependent and Time- Independent costs. The risk on Time-Dependent costs is then driven completely by schedule risk. The problem with this approach is that we lose the benefit of our historical cost databases which give us a legitimate baseline for scoping “unknown unknown” risks.
Furthermore, independent schedule analyses typically do not consider disparities between the cost estimate and the budget. This is why there are often discrepancies between cost risk assessments and schedule risk assessments. If our independent cost estimate tells us that a WBS item’s cost (i.e. effort required) will be significantly more than the budget (i.e. resources available) for that item, the overall schedule for that item will surely slip. This casts a shadow of doubt (uncertainty) on all schedule activities planned under that WBS item. The obvious way account for the discrepancy between cost and budget in the risk assessment is to stretch the schedule activities out until budget becomes available to complete the work.
This paper describes a cost and schedule risk assessment technique based on the assumption that work cannot be performed until funding is available to pay for it. In this approach, cost and schedule risk assessment is performed simultaneously. A simplified schedule network is used to capture the important interdependencies between major project elements. Budget leveling is also performed simultaneously in the risk assessment process, insuring that all model outcomes conform to budgetary reality.
Cost and schedule are two branches of the same tree and the trunk of that tree is effort. Trying to do cost risk assessments and schedule risk assessments independently and then fold the results of one into the other will always result in double-counting risk because both are driven by effort uncertainty. The only way to avoid this problem is to perform cost and schedule risk assessments simultaneously, based on one estimate of effort and effort uncertainty.
Tecolote Research, Inc.
Mr. Frederic has been with Tecolote Research since 1986. He is the Chief Scientist of Tecolote’s Santa Barbara Group, which includes offices in Santa Barbara, Ogden, Albuquerque, and Dayton. In his career at Tecolote, Mr. Frederic has logged a wide variety of experience including cost database software development, CER development, technical baseline development, schedule analysis, simulation software development, and cost estimating. His efforts have addressed many aerospace technologies including radars, missile systems, launch vehicles, launch facilities, space vehicles, and aircraft. In recent years, Mr. Frederic has performed cost risk assessments in conjunction with some of Tecolote’s most important estimating projects, including the Global Positioning System (GPS), the Airborne Laser (ABL), the International Space Station (ISS), and the Ares I launch vehicle.