Are your schedules coming in late? The problem may be as fundamental as duration uncertainty in scheduling.
The goal of proficient project managers is strict adherence to the schedule. The duration estimates in scheduling are specific and definitive. This decisiveness of duration estimates is alluring but, in many situations, misleading. This puts the project manager in a difficult scenario. The durations and completion dates the project manager is so committed to achieving may be completely erroneous.
A common misstep in scheduling is to take the critical path method (CPM) as fact rather than projections or estimates of future events. The problem is pervasive; many hold to the false belief that logic and durations in the project schedule are known with certainty.
This article discusses uncertainty in duration estimates, a major challenge to realistic scheduling.
Uncertainty in durations may be due to estimation error. Common causes of estimating errors include gaps in knowledge related to the work to be done, resources and resource productivity, and degree of reliance on others to complete respective tasks. Estimate accuracy increases as planning and engineering proceed, and knowledge of the project grows.
Duration uncertainty is also attributed to project risks. These risks affect duration. Examples of duration risk include: use of new technology, resource multitasking and associated availability issues, mismatch between of organizational resource productivity rates and actual rates, materials from suppliers arriving late or in unacceptable condition, failing to gain timely approval from regulatory agencies, or pressure to adopt unrealistically short durations.
Forgetting or ignoring duration risk once a single point duration estimate is produced ultimately leads to schedule problems when the project meets reality.
Both estimating errors and project risk contribute to duration uncertainty. This uncertainty is best represented with a probability distribution. A probability distribution represents the relative likelihood of each duration occurring. Something as simple as a three-point estimate makes a significant improvement compared to the single point deterministic estimate.
The three-point estimate becomes a triangular probability distribution. In the three-point estimate the estimator provides the longest possible or pessimistic duration, the most likely duration, and the shortest or optimistic duration. Schedulers can take this three-point estimate or triangular probability distribution and determine its effect on activity durations using a Monte Carlo simulation. The Monte Carlo simulation combines the probability distributions of activities along a respective path. In this way schedulers have a more realistic prediction of their schedule outcome.
Traditional deterministic scheduling assumes that duration estimates are known with exact certainty. The reality is there are many factors that affect the actual duration of an activity. It is better to qualify duration estimates using probability distributions.
A triangular probability distribution from a three-point pessimistic-most likely-optimistic estimate is a significant improvement over the single-point deterministic duration estimate. Schedulers can use a three-point estimate and the Monte Carlo simulation to find a more likely schedule duration outcome.
For a more in-depth study on duration uncertainty consider “Practical Schedule Risk Analysis” by David Hulett.