Building a Recovery Plan

Last reviewed 9 September 20263,414 words16 min read

Realistic re-sequencing vs fiction — the plan that survives its first fortnight

🟩 In one line: A recovery plan closes a measured gap with named, costed, approved measures applied one at a time to the critical path in a Reflection — anything that gets to the date by shortening remaining durations is a wish with a Gantt chart.

👤 Who this is for: M/S, with J reading for the method. Prerequisites: schedule-compression, forecasting-honestly, slippage-and-trend-analysis, revised-baseline-vs-recovery-schedule, near-critical-paths.

e / c
First, let's be honest about why this page exists

Most recovery plans are produced in three days under pressure and consist of the same network with the remaining durations of the critical activities cut by 20–30 percent. They reach the date, they get issued, and they are behind within two weekly cycles because nothing on site changed. The second recovery plan then has less credibility than the first, and by the third the Engineer stops reading them.

A real recovery plan is harder and slower. It starts with an honest forecast, names the gap, chooses measures in a fixed order, costs each one, gets the approvals it needs, resources the result, and is tracked weekly against milestones that expose failure early. This page is that method.

🟨 The standard — what "good" looks like
SourceWhat it asks for (paraphrased)What it means for the plan
FIDIC 1999 Cl 8.6 / 2017 Cl 8.7Where progress is too slow for reasons not entitling an EOT, the Engineer may require revised methods; Contractor bears the costThe plan must show methods, not shorter bars
FIDIC 1999 Cl 8.3 / 2017 Cl 8.3Revised programme with supporting report describing methods and resourcesNarrative plus resource histogram are mandatory content
NEC3/NEC4 Cl 32.1Revised programme shows how the Contractor plans to deal with delaysThe measures are the programme's content, not an appendix
NEC3/NEC4 Cl 36Acceleration by quotation: revised programme and Prices, accepted before implementationWhere the cause is the Client's, recovery is a quotation, not a gift
SCL Protocol 2nd ed., CP 15Contractor mitigates delay; not obliged to spend significant sums unless instructedVoluntary measures vs paid acceleration must be labelled
SCL Protocol 2nd ed., CP 16Acceleration compensable only when instructed or agreedGet the instruction before the second crew arrives
AACE RP 38R-06Basis of durations documentedEvery changed duration in the recovery has a written basis
AACE RP 52R-06Impact modelled on the current update in a controlled copySame discipline for recovery as for a TIA
DCMA Metrics 12–14Critical path test, CPLI, BEIThe recovery plan is tested before issue; CPLI to the target ≥ 1.0
GAO Schedule Assessment Guide, BP 3 and 9Resources assigned and realistic; updates reflect logicA recovery that doubles manpower overnight fails BP 3
PMI Practice Standard for Scheduling, 2nd ed.Schedule compression: crashing and fast-tracking, with their trade-offsThe two lever types, in the order below

🟢 Rule: honest forecast first, gap measured, levers applied in order (constraints → soft logic → resources → shifts → method → scope agreement), each one in a Reflection with a cost and an approver, the net saving checked against the near-critical paths, and the result tracked weekly against named milestones.

How it actually works

Step 1 — Diagnose before you plan

From the trend register and the last update, write down four numbers before touching P6:

NumberSourceExample
Gap: forecast completion − contractual completionCurrent update45 days
Slippage rateTrend register8 days/month, three months running
Driving chainLongest path, first activity behind DDStructure cycle → façade → finishes → T&C
Time to next unavoidable milestoneContractStructure topping-out KD in 10 weeks

A plan that closes 45 days but ignores an 8-day-per-month slippage rate will be 16 days behind before its second monthly update. The target must cover the gap and stop the rate.

Step 2 — Fix the forecast first

Every in-progress activity gets a fresh remaining duration from achieved rate (forecasting-honestly). If the recovery is built on optimistic RDs, the negative float is already inside it and the first weekly update exposes it. Only when the forecast is honest do you know the real gap — it is often larger than the reported one.

Step 3 — Set the target and keep it separate

Target = contractual completion (or an agreed intermediate date if the full gap is unrecoverable — say so). The forecast stays the network's date. Both are reported until measures are evidenced on site.

Step 4 — The levers, in order

OrderLeverTypeTypical savingCostApproval neededNotes
1Remove self-inflicted constraints and preference logicFast-track3–10 daysNilInternalConstraint register audit; "unlimited resources" test on every critical link
2Re-sequence soft preference logic: FS→SS with real lag; zone/floor splits; parallel work facesFast-track5–20 daysLow (supervision, access)Internal; consultant if method statement changesThe main lever on buildings — MEP first fix trailing structure by fewer floors, façade starting lower
3Add resources to critical tradesCrash1–3 days per floor/zoneMedium–highSubcontract instructionSecond crew ≈ 60–70% of first; Gulf mobilisation 4–8 weeks (visas, gate passes, accommodation)
4Extended hours / second shiftCrash10–25% on the activityMedium–highMunicipality permit for night work; labour law50 h/week >3–4 weeks loses 10–15%; 60 h loses 20–30%; night shift = the only real summer lever on external work
5Method change: precast, prefabricated risers, modular MEP, alternative façade sequenceFast-track/crash5–30 daysHigh, sometimes cost-neutralConsultant/client approval — weeks, not daysHighest saving, longest approval lead; start the approval on day one
6Procurement expedite: air freight, alternative supplier, early releaseCrashDays to weeks on the specific itemMediumClient if specified itemOnly where the item is on the critical or near-critical path
7Scope agreement: sectional completion, phased handover, descoped itemsWhatever is agreedCommercialClient, in writingNot a schedule measure; a contract one — never assumed

Never used: negative lags; proportional RD cuts; calendar changes that add days without adding hours; deleting logic; moving Finish On or Before constraints; Expected Finish dates typed to a target.

Step 5 — Build it one change at a time

  1. Project → Create Reflection of the current update; name it "Recovery R1 — DD 01Sep".
  2. Apply lever 1. Schedule. Tools → Schedule Comparison against the source. Record days saved.
  3. Re-run the near-critical filter (float ≤ 15 days on this job). Note which path is now driving.
  4. Apply the next lever to the new driving path. Repeat.
  5. Stop when the target is reached or when the next lever's cost per day exceeds what the project will pay.

Net saving is always less than the sum of the options because the second path takes over. A plan showing 45 days saved from options totalling 45 was built without step 3.

Options table in the recovery narrative — one row per measure applied:

IDMeasureTypeChange in P6Days saved (net, in sequence)CostCost/dayRiskApproverStatus

Step 6 — Resource it

Run the resource histogram on the Reflection. Check three things: the peak against site capacity (accommodation, hoists, gate throughput); the ramp — a trade cannot go from 40 to 90 operatives in a week in the Gulf; and productivity — apply the factors from step 4 in the durations, not in a footnote. A recovery plan whose manpower curve steps vertically is fiction with a histogram.

Step 7 — Test it before issue

TestPass criterion
Negative float against the targetNone
CPLI measured to the target≥ 1.00
Longest path continuous, no constraints holding itYes
Near-critical paths (float ≤ threshold)Listed with float and driver; none with less float than the plan needs to absorb a bad week
LagsNo negative; all positive lags on the correct calendar with a reason
Every changed durationNotebook "Forecast Basis" or "Recovery Basis" entry
Every measureCost, approver, status in the options table
HistogramPeak within site capacity; ramp ≤ what the subcontractor confirmed in writing

Step 8 — Write it

Recovery narrative, in this order: gap and its cause (neutral tags, one paragraph); reservation of entitlement sentence; options table; resource curve before/after; approvals required with dates by which they are needed; residual risk (what happens if the approval is late or the crew is short); weekly recovery milestones for the next 8–12 weeks with owner; trigger points ("if structure L16 pour is not achieved by 14 Oct, R1 is not recoverable and R2 will be issued"). Forecast and target both stated.

Step 9 — Track it weekly

Recovery activities get a filter and a look-ahead of their own. PPC on recovery activities is reported every week. Two consecutive missed weekly milestones means the plan is dead; re-plan then, not at the monthly update. Each monthly update shows the forecast converging on the target — or not — in the trend register.

📜 Whose cost, and how the plan is worded
Cause of the gapContract positionHow the recovery narrative is framed
Contractor's ownFIDIC Cl 8.6 / 8.7: revised methods at Contractor's cost; NEC Cl 32: revised programme; SCL CP 15"Mitigation measures at the Contractor's cost"
Employer/Client's (EOT due)Recovery is acceleration — FIDIC by agreed variation; NEC Cl 36 quotation; SCL CP 16"Proposed acceleration, subject to instruction under … ; without prejudice to entitlement under Cl 8.4 / Cl 60.1" — and do not mobilise the second crew until the instruction exists
MixedSeparate the days in the options table by causeTwo sub-totals; measures attributed

Spending the money before the instruction, then claiming it, is the constructive-acceleration argument — expensive to run and often lost. The reservation sentence and the split table are what keep the door open.

🟥 Where people go wrong
  1. Cutting remaining durations by a percentage. The oldest fiction. Nothing on site changed, so the first weekly update is behind. Every duration change needs a resource or method change written next to it.
  1. Closing the gap but not the rate. 45 days recovered on paper while the site keeps losing 8 a month. The target must cover both, or state honestly that only the gap is being addressed and the rate is being fixed separately.
  1. Reporting the target as the forecast. The narrative shows 30 April because the recovery says so. Two dates, two bases, until the measures are on site and in the achieved rates.
  1. Unresourced. The Gantt chart reaches the date; the histogram doubles in a week; the second crew's productivity is assumed at 100%; nobody asked about accommodation. Run the histogram before the Engineer does.
  1. Uncosted, unapproved. Measures that need consultant approval of a method change or a municipality night-work permit are shown as if approved. Put the approval lead time on the critical path of the recovery itself.
  1. No weekly tracking. Issued at the monthly update, next examined at the following one, by which time it has failed silently for four weeks. Weekly milestones with owners, PPC, trigger points.
  1. Issued as a revised baseline. The slippage disappears from the reports and the accepted baseline is at risk. Recovery is a Reflection or named copy; the baseline register does not move (revised-baseline-vs-recovery-schedule).
⚖️ When you're challenged

"Just show me a plan that gets to the date. I don't care how." I can produce that in an hour and it will fail in two weeks, and the next one will be believed less. The plan I'm proposing gets to 5 May — five days short — with measures we can actually mobilise, and it tells you what the last five days would cost if you want them. That's a plan you can take to the client.

"Why should we pay for a second formwork set? You're the one who's late." On the structure, yes, and those measures are at our cost — it says so in the table. The façade re-sequence and the exhaust unit expedite are recovering the transformer delay, which is the Employer's, and those are shown separately as proposed acceleration pending instruction. The table splits the days and the money by cause.

"Can't we just work Fridays and longer hours?" Fridays adds 15% calendar days on paper and about 8–10% real output after fatigue. Sixty-hour weeks for three months lose 20–30% per hour. Both are in the options table with their real savings and costs. They're lever four for a reason — the re-sequence in lever two saves more for less, and it goes first.

"Your recovery says 30 April, your forecast says 14 June. Which one do I put in the client report?" Both, with one line each. Forecast is today's rates with today's resources. Target is with the measures in the table, three of which need approvals listed with dates. As those land, the forecast moves. Putting only the target in the report is how we end up explaining a surprise in June.

📄 Related pages
✏️ Worked example

A 22-storey residential tower in Dubai, 6-day calendar, FIDIC 1999. Update 08, data date 1 September. Contract completion 30 April. Honest forecast after RD refresh: 14 June — a gap of 45 working days, of which 38 are the Contractor's structure slippage and 7 are an Employer-caused MEP design change (claimed). Slippage rate 8 days/month for three months. Structure at Level 14 of 22, achieved cycle 9 days against 7 planned. Critical path: structure → façade unitised install (held by preference logic until structure at L18) → internal finishes → T&C → handover.

Measures applied in sequence in "Recovery R1 — DD 01Sep":

IDMeasureTypeChange in P6Days saved (net)CostCost/dayRiskApproverCause recovered
R1-01Remove Start On or After on façade L3 (no document basis)Fast-trackConstraint deleted4NilLowInternalContractor
R1-02Façade install starts at structure L12 not L18 (soft preference logic)Fast-trackPredecessor changed from L18 slab to L12 slab, FS + 5d lag on 6-day calendar (edge protection)12AED 180k (extra crane hooks, access)15kMedium — crane availabilityInternal + consultant method statementContractor
R1-03MEP first fix trails structure by 4 floors not 7Fast-trackSS links re-pointed floor by floor6AED 60k supervision10kLowInternalContractor
R1-04Second formwork set + additional crew, structure L16–L22; cycle 9 → 7 daysCrashRDs L16–L22 revised from 9 to 7 with basis note; productivity 0.65 on second crew reflected10 (7 floors × 2 days, less ramp-up)AED 620k62kMedium — 5 weeks to mobilise via visas; set arrives L16Subcontract instructionContractor
R1-05Air-freight replacement MEP switchgearCrashProcurement RD 42 → 21 days7AED 210k30kLowEmployer (specified item) — proposed accelerationEmployer
R1-0610-hour to 11-hour day, internal finishes, Feb–AprCrashCalendar hours 10 → 11 on finishes calendar; productivity factor 0.92 applied3AED 140k47kLowInternalContractor
Total42 (options sum 48; near-critical takeover after R1-02 and R1-04)AED 1.21m

After each measure the near-critical filter (float ≤ 15) was re-run. After R1-02 the lift installation path became the driver for 6 days; after R1-04 the finishes path took over. That is why 48 days of options net to 42.

Result: recovery target 5 May (5 working days short of contract). Options to close the last 5 days (night shift on external works, cost/day AED 85k) listed but not adopted — decision for the project director. Forecast remains 14 June in Update 08; the narrative shows both dates.

Weekly recovery milestones for the first 8 weeks:

Week endingMilestoneOwnerTrigger if missed
7 SepFaçade method statement revision submittedFaçade subR1-02 at risk
14 SepSecond crew visa applications lodged (38 operatives)Structure subR1-04 slips one floor per week of delay
21 SepL15 slab pouredStructure subTrend not yet arrested
28 SepFaçade L3–L5 brackets setFaçade subR1-02 saving reduces 2 days/week
5 OctSecond formwork set on siteStructure subR1-04 saving reduces 2 days/floor
12 OctL16 slab poured at 7-day cycleStructure subIf 9 days: R1-04 has failed; R2 to be issued
19 OctSwitchgear instruction from EngineerEmployer/EngineerR1-05 lapses; 7 days return to Employer's account
26 OctMEP first fix L11 completeMEP subR1-03 saving at risk

By Update 10 (DD 1 November): L16 and L17 poured at 7 and 8 days, façade started at L12 on 23 September, switchgear instruction received 17 October. Forecast moved from 14 June to 22 May. Target held at 5 May. Trend register shows the first month with a negative movement in six.

📖 References
  • FIDIC Conditions of Contract for Construction, 1999, Cl 8.3, 8.4, 8.6; 2017, Cl 8.3, 8.5, 8.7 (check the edition in your contract)
  • NEC3 / NEC4 Engineering and Construction Contract, Cl 32.1, 36 (check the edition in your contract)
  • SCL Delay and Disruption Protocol, 2nd ed. (2017), Core Principles 15, 16
  • AACE International RP 38R-06, Documenting the Schedule Basis
  • AACE International RP 52R-06, Time Impact Analysis – As Applied in Construction
  • DCMA 14-Point Schedule Assessment, Metrics 12, 13, 14
  • GAO Schedule Assessment Guide (GAO-16-89G), Best Practices 3, 9
  • PMI Practice Standard for Scheduling, 2nd ed., schedule compression
  • Oracle Primavera P6 Professional User Guide — Reflections; Schedule Comparison; Resource Usage Profile; Multiple Float Paths

From the field

Experience from working planners. Unreviewed — read it as experience, not guidance.

Add what you know about building a recovery plan. 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