Schedule Logic

Schedule Logic

Things to look for in driving paths

When viewing the driving paths there are a number of things that can provide insights as follows:

1. Is the deterministic driving path different from that of the probabilistic driving path, above we found that the deterministic driving path was far less critical than that projected by nPlan.
2. Are there bottlenecks in the schedule, i.e., areas where many links merge at a single point? These areas circled in red show accumulate risk and concentrate the uncertainty of the predecessors with every activity before it needing the be finished before it can start. We find in real life that often at least a few of these predecessor links are broken.
3. High influence nodes (the opposite of bottlenecks), the node circled in blue is an example of a high influence node with many successors dependent on the nodes successful completion. Similar to bottlenecks, we often find that if this node were to be delayed that in real life many of the successor would start without waiting for its completion, breaking the logic.
4. Overly linked schedules - check to see that links haven’t been added for “convenience logic,” review the driving paths and highly parallel areas of work. Make sure that the logic is driving to the correct milestones and isn’t connected without reason. The higher the number of links that greater the complexity and hence the possibility for patterns to emerge where uncertainty is great and hence, delays are more likely to occur.

Merge Bias - thought experiment

Merge bias is not something unique to nPlan, it happens in all QSRA’s and is a result of parallel paths converging at one point, the more paths, the lesser the likelihood of meeting the planned date. A thought experiment is to image each path leading to a bottleneck was a coin and the chance of meeting the end date is the chance of flipping and having all coins landing on heads. Initially with one coin its 50/50, then with two its 25/75 and then with four its 7/93, hence the more coins the less likelihood we have of them all landing on heads at once.

Merge Bias - planning example

When only a single activity drives the completion milestone, the result is as expected based on the 80/100/120 range applied to Activity 1, the P50 result is 100 days, and the original completion duration has a 50% chance of occurring:

When two activities drive the completion milestone, the P50 result is 104 days, and the original completion duration has a 25% chance of occurring:

When four activities drive the completion milestone, the P50 result is 108 days, and the original completion duration has an approximately 7% chance of occurring:

Death by a thousand cuts - what does this mean and what type of projects does it occur on?

Explanation:

We see this metaphor play out on projects where there are many risky activities with a small impact on the end date and hence it is not initially possible to attribute the delays to specific areas of the project without using themes or grouping to uncover the key types of risky activities. These scenarios often play out on complex schedule schedules which may be over-linked and have lots of parallel paths resulting in lots of low criticality, low impact activities. Mitigating delays on these types of projects requires a more holistic view of both the schedule structure itself to ensure it reflects reality and the high level (WBS level 1-2) groupings.