DCMA-14 Metric 8: High Duration Activities in a Construction Schedule

The maximum activity duration in a DCMA-style review is 44 working days, about two months. Incomplete activities with a baseline duration above that, and a baseline start inside the detail planning period, are flagged as high duration. No more than 5% of incomplete tasks should exceed the limit.

High duration is metric 8 of the DCMA-14 schedule assessment. It is less about the number than about control: a long activity hides progress, float and risk inside a single bar.

What does the DCMA high duration check measure?

The pamphlet counts incomplete tasks whose baseline duration is greater than 44 working days and whose baseline start falls in the detail planning period, then divides by the incomplete tasks in the population. The result should not exceed 5% (DCMA-EA PAM 200.1).

```
High Duration % = incomplete tasks with baseline duration > 44 WD
(baseline start in detail planning period)
÷ incomplete tasks in the population
```

The pamphlet states the purpose directly: to determine whether a task can be broken into two or more discrete tasks, which improves cost and schedule insight. Mosaic Projects, in its white paper on the 14 checks, describes high durations as generally an indication that a plan is too high level (Mosaic WP1088).

Why does the check use baseline duration?

Using baseline duration keeps the test about planning quality. If remaining duration were used, a 90-day activity would pass the moment it was 60% complete, even though it was never broken down enough to track. With baseline duration, the activity stays flagged until it finishes.

This also means you cannot clear the flag by shortening remaining duration during an update. A reviewer who sees a long baseline activity with an unexplained drop in remaining duration will look for the progress evidence, not the metric.

What is the detail planning period and the rolling wave exemption?

The metric applies to tasks that start inside the detail planning period, a rolling window that moves forward with each update. Tasks beyond that window can be planned as larger work packages or planning packages, and they are not counted. This is the rolling wave principle: plan near-term work in detail and far-term work at a higher level, then break it down as it approaches (Mosaic WP1088).

For a construction schedule, this means a 120-day placeholder for "Level 5-8 interiors" that begins eight months from now can be acceptable in a baseline, provided the schedule says it is a planning package and you break it down before it enters the window. The window length itself is set by the contract or program, not by the pamphlet's 44-day figure, so check your specification.

Which long activities are legitimate in a construction schedule?

Some long bars are fine. The test is whether the duration reflects one continuous, trackable thing or a bundle of work that could be statused separately.

If you keep a long activity, say why in the narrative or activity notes. A reviewer is far more willing to accept a documented exception than an unexplained bar.

Worked example: calculating the high duration percentage

Illustrative example: assume a schedule with 400 incomplete tasks in the population (completed, LOE, summary and milestone tasks already removed). Twenty-six of them have a baseline duration over 44 working days and a baseline start in the detail planning period.

At 5%, the schedule could carry 20 such tasks. Splitting just 6 of the 26, starting with the ones on or near the critical path, brings the result to 20 ÷ 400 = 5.0%. That is the arithmetic, but the better question is which of the 26 actually hide risk. Start with long activities that carry low float or sit near the driving path, because a slip inside them is the hardest to see until the whole bar finishes.

Note that splitting changes the denominator as well. Replacing one long task with four short ones adds three tasks to the population, which lowers the percentage slightly even for the tasks you did not touch. Re-run the count after the split instead of adjusting the old numbers.

Why do long activities hide risk?

A single 60-day bar can be 50% complete and still be a week behind, and nothing in the schedule will say so until the end date moves. Percent complete on a long activity is a judgment call, so two people can report different progress on the same bar. Short activities replace that judgment with a binary event: the work is done or it is not.

Long activities also weaken float. If a long activity carries all the work of a floor, its successors cannot start until the whole bar finishes, even when the first bay is ready for the next trade. A split with realistic overlap often shows float the original bar hid, and it makes look-ahead planning more precise. The check also supports cost insight, which the pamphlet names as part of its purpose, because progress measured in smaller increments supports more accurate earned-value reporting. See the resources check for how loaded cost or hours interact with activity size.

How do you find high duration activities in P6?

Filter the activity table for original duration greater than 44 working days, excluding LOE, WBS summary and milestone activity types, and for status of not completed. Show the baseline start if you have a baseline assigned so you can check the detail planning window. Count the results and divide by the count of incomplete, non-LOE activities.

Two practical notes. First, P6 shows durations in hours or days depending on user preferences, so confirm the unit before you filter. Second, duration is measured in working days of the activity calendar, so a 7-day calendar activity with a 44-day duration covers fewer calendar weeks than the same count on a 5-day calendar. Check the calendar column together with the duration filter. Also see the P6 calendar notes in the negative float guide, since calendar mismatches cause errors in several metrics.

How do you break down long activities?

Divide by the dimension that makes progress measurable in the field. That is usually location, then crew or trade, then inspection or approval stage.

Illustrative example: assume a single activity "Structural steel erection" with a 90 working day duration. Splitting by level and zone gives:

Each piece is now short enough to status weekly, carries its own float and ties into inspections. Keep the total sequence duration consistent with the original unless the original was wrong, and if it changes, record why. Prefer finish-to-start links where possible; where you use start-to-start with a lag, the lag should reflect a real offset, as covered in lags in construction schedules.

What mistakes do reviewers flag?

• Splitting into dummy slices (for example, "Part 1 of 4") with no real scope boundary, which adds activities without adding control.
• Splitting an activity but leaving the successors tied to the first slice, which breaks the logic.
• Lowering a baseline duration after the fact to clear the flag.
• Reporting a clean metric because long activities sit just outside the planning window, with no plan to break them down.
• Ignoring the calendar, so the same scope counts differently in different tools.

A related symptom is high float on the successor activities after a split, which usually means the new pieces lack a successor.