Critical path: calculated and factual

What makes a path critical in the model

Logic, durations, calendars, constraints and progress. Change any one and criticality moves. Constraints are the usual culprit: a mandatory start, a deadline constraint or a stale actual date can hold a path critical long after the work stopped controlling anything.

Settings that quietly change the answer

• Retained logic against progress override, which changes how out-of-sequence progress is calculated.
• Calendars applied inconsistently across activities, so a delay is absorbed by a weekend that the site actually worked.
• Open ends, where activities have no successor and therefore never drive anything.
• Long lags substituting for real activities, hiding what the project was actually waiting for.
• Level of detail so coarse that the controlling work is buried inside a summary bar.

Testing the model against the project

1. What were site managers prioritising, and what do the minutes say they were waiting for?
2. Which missing design, access, approval or material constrained the next operation?
3. What do installed quantities and resource allocations show about where the work actually was?
4. Do the look-aheads issued at the time identify the same priority as the model?
5. Did the apparent constraint genuinely govern, or is it an artefact of a setting?

Where the two disagree, the disagreement is the finding. An analysis that reports a calculated path without testing it against the record has described its own model rather than the project.

Near-critical work

Paths with small float become critical the moment anything slips, and on a compressed project several paths sit within days of each other. Reporting only the single longest path in that situation overstates certainty. Identify the near-critical paths in each period and say how close they were.