Project Management through Dynamic Scheduling

This week, I had the opportunity to take a course called Strategic Project Analysis. Here are my key takeaways. Project management involves planning, managing, and controlling a project to ensure its successful completion—on time, within budget, and according to specifications. To effectively manage a project and make informed decisions, it is essential to rely on data-driven analysis of the project's nature.

The first step in project management is planning, specifically scheduling, which serves as a reference point throughout the project. The Critical Path Method (CPM) focuses on activities that have zero slack time, meaning that any delay in these activities will directly impact the overall project timeline. In contrast, the Program Evaluation and Review Technique (PERT) employs best-case, realistic, and worst-case estimates to calculate the probability of activity durations based on a normal distribution. This is done by using the average planned duration alongside the standard deviation of critical activities.

To make informed decisions, the probability of finishing the project within your target time using the formula can be determined: NORMDIST(Target Time, Average Duration, Variance or sqrt(sum of std^2), 1). Alternatively, it is possible to calculate a delivery date for a specific percentage of risk using: NORMINV(Target Percent, Average Duration, Variance or sqrt(sum of std^2)).

To compare the two projects and select Project 1, we will determine if the probability of Project 1 being completed before Project 2 is greater than zero by using the differences in their average durations and variances as input for probability estimation. However, it is important to note that PERT has limitations; its data relies on human input, it does not account for the correlation between activities, and it overlooks critical activities.

The second step is Reactive Scheduling (Game), which centres on activity crashing by prioritizing the most time- and cost-sensitive activities. This process involves identifying the lowest cost within a dynamic environment (risk) by reducing the duration of activities on the critical path with the smallest cost slope. Conversely, for non-critical paths, the focus should be on increasing the duration of activities that have the largest cost slope. It is crucial to consider risks (uncertainties) and complexities to make informed and effective decisions.

The third step in the process is risk analysis, which uses a plan as a point of reference. After gathering data on the risk distribution of each activity, a criticality and impact analysis can be conducted using Monte Carlo simulation. Three important parameters are involved in this analysis:

1. Criticality Index (CI): This measures the probability that an activity lies on the critical path.

2. Significance Index (SI): This measures the impact that variability in activity durations has on the total project duration.

3. Schedule Sensitivity Index (SSI): This is calculated by multiplying the standard deviations of both activity durations and project durations (which indicates impact) with the Criticality Index (probability).

The CI and SI products can be used to estimate the SSI value. Effective control can be achieved by focusing on activities with a higher SSI value. The selection of activities and control balance depends on the nature of the project. This method is known as the bottom-up approach and is typically effective for project activity flows that are parallel in nature.

The final step in project management is project control, which focuses on evaluating actual progress. One example of this is the earned value management technique. This method is referred to as the Top-down approach and is generally effective for project activity flows that are serial in nature.

First, identify the key metrics, including Actual Cost (AC), Planned Value (PV), Actual Duration (AD), Earned Schedule (ES), and Earned Value (EV). Earned Value is calculated by multiplying the percentage of tasks completed by the total budget cost (BAC).

The Earned schedule is estimated using ES = t + (EV - PVt) / (PVt+1 - PVt).

The following cost and time performance metrics can be used to assess progress, such as:

1. Cost variance: CV = EV - AC ; Interpretation: < 0: cost overrun, = 0: on budget, > 0: cost underrun
2. Cost performance index: CPI = (EV / AC); Interpretation: < 1: cost Overrun; = 1: on budget; > 1: cost underrun.
3. Schedule variance: SV(t) = ES - AD; Interpretation:< 0: project delay, = 0: on schedule, > 0: ahead of schedule.
4. Schedule Performance: SPI(t) = (ES / AD); Interpretation: < 1: project delay, = 1: on schedule, > 1: ahead of schedule

In addition, the forecasting metrics can also be used to predict the project's nature using the following estimations.

1. Expected At Completion: EAC(€) = AC + (BAC – EV) / PF; using three Performance (PF) choices, PF = 1 Future as planned, PF = CPI Future as current cost performance, PF = CPI x SPI(t) Future as current time/cost performance.
2. Expected At Completion: EAC(t) = AD + (PD – ES) / PF; using three Performance (PF) choices, PF = 1 Future as planned, PF = SPI(t) Future as current time performance, PF = SPI(t) x CPI Future as current time/cost performance.

By utilizing this data-driven approach, we can effectively assess the project's performance. However, it is important to note that the nature of the project may vary, and the measurement techniques that are effective in one situation may not be suitable in another. The following chart compares different project control approaches based on network-based activities.

The chart below illustrates how real project activities generally fit into various activity networks, as outlined by Professor Vanhoucke.

In conclusion, dynamic scheduling techniques that consider planning, risk, and control are the most effective for increasing efficiency compared to the benchmark throughout the entire project lifecycle, according to Vanhoucke's work. For serial networks, Earned Value Management (EVM) proves to be the most effective approach, while for parallel networks, scheduled risk analysis is the best option.

*******

I express my gratitude to Professor Mario Vanhoucke at Ghent University (Belgium), Vlerick School (Belgium), and University College London, UCL School of Management (UK), who is also a Partner at OR (Belgium).