DCMA-14 Metric 6 — High Float: What Excessive Float Reveals

High float in a construction schedule is total float greater than 44 working days (two months) on an incomplete activity. The DCMA 14-point assessment says that if more than 5% of incomplete tasks are in that state, the network "may be unstable and may not be logic-driven." Most often the cause is missing logic, constraints or calendar errors, so high float works better as a symptom than as a defect in itself.

This is metric 6 in the series. The sections below cover how to compute it, what usually causes it, and where the fixed cutoff gets criticized.

How does DCMA define the high float metric?

The formula is:

High Float % = incomplete tasks with total float > 44 working days ÷ incomplete tasks

Per the DCMA pamphlet, the threshold is 5%. Total float is defined as late finish minus early finish, and the 44 days are working days on the activity's or project's calendar. The population excludes completed tasks, level-of-effort tasks, summary tasks and milestones.

Mosaic Projects' white paper describes a broader population that includes milestones, summaries and completed work, which contradicts the pamphlet. That is another reason to state which population you used.

Illustrative example: with 250 incomplete detail tasks, 5% allows 12 tasks over 44 days (4.8%). A result of 20 tasks is 8% and flagged.

Total float versus free float

Total float is the time an activity can slip without delaying the project finish. Oracle defines float as the time an activity can be delayed before the project finish is at risk, and negative float as the project already being behind. Free float, per Oracle's data dictionary, is the delay possible "before delaying the start date of any successor."

Free float is never greater than total float. A task can carry 60 days of total float because a long chain of successors shares it, but only a few days of free float, because the next activity is waiting. That is why some reviewers check free float to separate shared float from isolated slack.

What causes high float in a construction schedule?

Pamphlet text and practice point to a short list.

The pamphlet says directly that high float "may be a result of missing predecessors and/or successors." The logic metric is therefore the first place to look. Missing successors also give you float on paper only: they let the activity finish at any date, so it looks safe while nothing prevents the delay from reaching the end.

Padding is harder to see. A scheduler may stretch durations so that an activity finishes with float to spare, or add lags to the same effect. See lags in a construction schedule for how DCMA treats lags used to manipulate float.

Worked example: a missing successor

Illustrative example with assumed durations. The driving path runs 90 days to completion.

L400 finishes on day 15 at the earliest. With no successor, its late finish defaults to the project end at day 90, so total float is 75 days, well above 44, and the task counts toward the metric.

Now add the missing tie. The conduit must be in before structure starts, so link L400 to S120 with FS. S120 cannot start before day 40, so L400's late finish becomes day 40 and its float drops to 25 days. The task leaves the high-float count, and the number now means something: a 25-day slip in the conduit work would begin to delay structure.

The schedule gained information, not just a better score. That is the sense in which high float is a symptom of missing logic.

Is the 44-working-day cutoff fair?

Not everyone thinks so. SmartPM's DCMA 14 checks article (April 2022) calls the 44-day limit controversial because it is the same for a 6-month and a 2-year project, and suggests looking at free float. The pamphlet itself offers no derivation for the number beyond equating it to two months.

The criticism has force in both directions:

• Short projects. On a 6-month project of about 130 working days, 44 days of float on a task is a third of the schedule. A sensible reviewer would want that explained.
• Long projects. On a 2-year project, 44 days is under 10% of the duration. Early-stage activities can legitimately hold more float than that, and the metric will flag them.

Treat the number as a screening tool, then judge the flagged tasks. Do not substitute your own threshold casually; a percentage-of-duration rule seen on one third-party site has no support in the pamphlet and should not appear in a submittal as DCMA's.

GAO's Schedule Assessment Guide takes the question from another angle with its "reasonable total float" practice, asking whether float values make sense given the logic instead of applying one cutoff.

How do I find and fix high float in P6?

1. Add a total float column and filter to incomplete detail activities with float above 44 working days. Check the float unit setting in your project so you compare days, not hours.
2. Group by WBS to see whether the flagged tasks cluster in a few packages. Clusters point to pasted-in schedules or missing ties.
3. For each flagged task, classify the cause using the table above.
4. Fix in order of likelihood: add missing predecessors and successors, review constraints, correct calendars, then challenge padding.
5. Reschedule and recount. Keep a before-and-after count in the narrative.

A repair of missing logic often clears more than this metric. It can also bring back the critical path test and reduce hard constraints that were added to cover for undefined sequence.

What will a reviewer ask about high float?

Reviewers tend to ask four things. What percentage did you get, with which population and tool? Where do the flagged tasks sit in the WBS? Which of them are open ends, and which are real independence? And has the count moved since the last update?

A rising count between updates is a signal on its own. New scope added without links, activities detached when predecessors were deleted, and a calendar changed on the wrong activities all show up as float appearing on tasks that had little before. Compare the flagged lists across updates by activity ID, not only the percentage.

How does high float relate to near-critical work?

Float is also how you rank near-critical activities, and distorted float misleads that ranking. If padding or open ends inflate float, truly near-critical activities look safe, and a delay may reach the critical path with no warning. Fix the distortion before using float to prioritize recovery work.

When should I leave high float alone?

Some float is real. Landscaping, signage and certain closeout items can run late without moving completion, and seasonal work may wait on a window. If the float is legitimate, document it: one line per task or per package, saying what the task is independent of and why. Reviewers asking questions about a flag are looking for that sentence. If the same tasks appear on every update with the same note, say so once in the narrative and keep the list current, instead of rewriting the explanation each month.