Overtime and Shift Arithmetic
Last reviewed 9 September 20261,811 words8 min read
Why twelve hours doesn't buy fifty percent more
🟩 In one line: Extended hours buy a fraction of the calendar time they appear to buy — put the extra hours in a calendar, put the lost efficiency in a productivity factor, and show the net gain in working days before anyone signs the overtime sheet.
👤 Who this is for: Junior and mid-level planners asked to "show what 12 hours gives us." Prerequisites: Crew-and-manhour-formulas, Productivity-adjustment-factors, Calendar-setup-and-holidays.
First, let's be honest about why this page exists
Every recovery meeting ends the same way: "Go to twelve hours." The arithmetic in the room is 72 ÷ 60 = 1.2, so a 40-day activity becomes 33 days. It doesn't. After the third week the crew is slower per hour, supervision is thinner, materials don't arrive faster, and the extra hours are spent standing in the same congested zone.
This page gives you the factors to use, where each one goes in P6, and the one worked example that shows why a night shift usually beats a long day. It is deliberately short — the argument is in Productivity-adjustment-factors and Schedule-compression.
🟨 The standard — what "good" looks like
| Source | What it says (paraphrased) | Use it for |
|---|---|---|
| AACE RP 25R-03 Estimating Lost Labor Productivity in Construction Claims | Lists overtime, shift work and overmanning as recognised productivity-loss causes; efficiency falls with the number of weeks sustained (check against your copy) | Justifying the factor you apply |
| Business Roundtable Report C-2 Scheduled Overtime Effect on Construction Projects (1980) | Sustained 50- and 60-hour weeks lose efficiency progressively; after several weeks the extra hours produce little or no extra output (check against your copy) | The classic curve everyone quotes |
| SCL Protocol 2nd ed., Guidance Part B (acceleration / disruption) | Acceleration measures should be agreed and their productivity effect recorded (check against your copy) | Why the factor must be written down before the shift starts |
| AACE RP 38R-06 Schedule Basis Memorandum | Productivity assumptions and calendars belong in the SBM | Where the factor lives |
| Hub convention | Tender rates already assume 6-day × 10 h (60 h/week). Beyond that: 66 h loses 5–10 % after 3–4 weeks; 72 h loses 10–20 %; 84 h (7 × 12) loses 20–30 % and collapses after 6–8 weeks. Second crew delivers 60–70 % of the first. Summer night shift is the only real lever on external work. | Your default numbers |
🟢 Rule: hours gained go in a calendar; efficiency lost goes in a productivity factor; the answer is reported in working days saved, never in hours added.
How it actually works
Step 1 — Know your base. In the Gulf the base is already 60 hours: six days, 10 hours, with 8.5 hours mid-June to mid-September for the midday break. Your unit rates were priced on that. Any factor you apply is relative to 60, not to a 40-hour week from a textbook.
Step 2 — Pick the regime and its factor.
| Regime | Hours/week | Factor weeks 1–3 | Factor week 4 onward | Effective hours (wk 4+) | Real gain vs base |
|---|---|---|---|---|---|
| 6 × 10 (base) | 60 | 1.00 | 1.00 | 60 | — |
| 6 × 11 | 66 | 0.95 | 0.90 | 59 | ≈ 0 % |
| 6 × 12 | 72 | 0.90 | 0.80 | 58 | ≈ 0 % (short burst +8 %) |
| 7 × 10 | 70 | 0.92 | 0.85 | 60 | ≈ 0 % (short burst +7 %) |
| 7 × 12 | 84 | 0.85 | 0.70 | 59 | ≈ 0 % (short burst +19 %) |
| Day + night shift, second crew | 60 + 60 | 1.00 + 0.65 | 1.00 + 0.65 | 99 | +65 % |
Read the right-hand columns. Extended hours are a three-week tool. Sustained, they give you back roughly the same output as the base week, at overtime rates, with a tired crew. A second shift is the only regime that keeps giving.
Step 3 — Check the law. Gulf labour codes cap daily overtime (UAE: typically 2 hours per day; KSA: an annual overtime ceiling — verify current articles). A 7 × 12 regime is frequently not lawful for more than short periods. Put the legal check in the options table before the cost.
Step 4 — Put it in P6 correctly.
| What changed | Where it goes | Never |
|---|---|---|
| More hours per day | New project calendar, e.g. "6-day 12 h — Recovery", Time Periods set to 12 h; Admin → Admin Preferences → Time Periods "Use assigned calendar…" ticked (check against your P6 version) | Editing the standard 6-day calendar that every other activity uses |
| Lost efficiency | Remaining Duration recomputed: RD = remaining manhours ÷ (crew × hours/day × factor), rounded up; basis in Notebook "Recovery Basis" | Cutting RD by 72 ÷ 60 |
| Second shift | Either double the crew in the RD formula with the 0.65 factor on the second crew, or a "6-day 20 h" two-shift calendar with RD from combined output | A 24 h calendar with the same crew size |
| Summer external work | Night-shift calendar 10 h (no midday ban applies at night); day calendar stays 8.5 h | Pretending the day crew works 10 h in July |
Do it in a Reflection, one activity group per step, Schedule Comparison after each, and compare the net days on the longest path — not the days on the activity.
🟥 Where people go wrong
- Dividing the duration by the hours ratio. 40 days × 60 ÷ 72 = 33 days appears in recovery plans every month. It assumes a crew works as fast in hour twelve as in hour two, and the Business Roundtable data says they don't.
- Applying the factor and the calendar to the same thing. Hours lost go in the calendar; efficiency lost goes in the factor. If you shorten the calendar for the summer break and factor the rate for heat, you have penalised the same hours twice.
- Leaving the recovery calendar on after the recovery ends. Six months later the same activities still show 12-hour days and the forecast is fiction. Put an end date for the regime in the options table and switch the calendar back at the next update.
- Forgetting that materials, cranes and inspections don't work overtime. A 12-hour blockwork crew fed by a 10-hour hoist and an 8-hour inspector produces 10 hours of work. Check the feeding chain before the labour.
- Ignoring the 8.5-hour summer day. Between mid-June and mid-September the "12-hour day" on external work is legally impossible. The lever there is a night shift, priced as one.
- Selling the burst gain as a sustained gain. Three weeks at 7 × 12 genuinely gives 15–20 %. Twelve weeks at 7 × 12 gives nothing and a safety record. State the duration of the regime in the same sentence as the gain.
⚖️ When you're challenged
"If we work 20 % more hours, why aren't we 20 % faster?" Because output per hour falls after the third week. The industry data shows a sustained 72-hour week ends up producing roughly the same as a 60-hour week — we'd be paying overtime to stand still. A second shift is the regime that actually adds output.
"Where did you get 0.80 from?" It's the hub convention drawn from AACE 25R-03 and the Business Roundtable study, applied relative to our 60-hour tender base. If you want to test it, the factor and the crew size are in the Notebook on every affected activity — change one number and reschedule.
"Just put the whole tower on 12 hours." I'll model it, but the answer will be near zero after week four, and the UAE daily overtime cap makes it unlawful to sustain. Let me show you the night-shift option alongside it with cost per day saved.
📄 Related pages
- Productivity Adjustment Factors — the full factor library and how factors stack
- Crew and Manhour Formulas — the duration formula the RD recomputation uses
- Schedule compression — where overtime sits in the lever order (it's late in the list)
- Building a Recovery Plan — options table format, cost per day, pre-issue tests
- Calendar Setup and Holidays — creating the recovery calendar and the summer hours
- Mitigation, acceleration and constructive acceleration — when the Employer pays for the night shift
✏️ Worked example — Dubai 22-storey tower, blockwork Levels 8–15
Position at DD 25-Jan: Blockwork L8–L15 remaining 4,800 manhours, crew 12, base calendar 6-day 10 h, factor 1.00. RD = 4,800 ÷ (12 × 10) = 40 working days. It is on the longest path.
| Option | Output per day | RD | Days saved | Notes |
|---|---|---|---|---|
| A — Stay at 6 × 10 | 120 mh | 40 | — | Base |
| B — 6 × 12 sustained | 12 × 12 × 0.80 = 115 mh (after wk 3) | 40 | 0 | Weeks 1–3 at 0.90 = 130 mh; net over 40 days rounds to 40 |
| C — 6 × 12 for 3 weeks, then base | 18 days × 130 mh = 2,340 mh; remaining 2,460 ÷ 120 = 20.5 → 21 | 39 | 1 | Burst gain only |
| D — Day crew 12 + night crew 10 | 120 + (10 × 10 × 0.65) = 185 mh | 4,800 ÷ 185 = 25.9 → 26 | 14 | Night-shift premium and supervision |
| E — Second day crew (14 total in zone) | Congestion factor 0.85 on both: 24 × 10 × 0.85 = 204 mh | 24 | 16 | Only if the zone will hold 24 people and the hoist feeds them |
Reported to the PM: Option D saves 14 working days on the longest path at a cost of AED X per day saved; Option B costs more and saves nothing. Option E is checked against the hoist capacity in Site-logistics-in-the-schedule before it goes in the options table.
📖 References
- AACE International RP 25R-03, Estimating Lost Labor Productivity in Construction Claims (check the edition in your contract)
- Business Roundtable, Report C-2, Scheduled Overtime Effect on Construction Projects, 1980 (check against your copy)
- SCL Delay and Disruption Protocol, 2nd ed. (2017), Guidance Part B — acceleration and disruption (check against your copy)
- AACE International RP 38R-06, Documenting the Schedule Basis
- Oracle Primavera P6 Professional User Guide — Calendars; Admin Preferences, Time Periods (check against your P6 version)
Illustrative ranges, for sanity-checking only. Not a substitute for your own as-built data, and not for use as evidence in a contractual claim.
From the field
Experience from working planners. Unreviewed — read it as experience, not guidance.
Add what you know about overtime and shift arithmetic. What worked, what the consultant pushed back on, what you would do differently next time. A paragraph is plenty.
Contributors get their name and one link on the site — your own templates, course or consultancy. We take nothing and hold nothing.
Add your experience