Transferring a Gulf Coast Estimate Into a Live Unit
A refinery owner is budgeting a brownfield tie-in at an inland Midwest site. The estimator takes a 2024 Gulf Coast detailed estimate of $48M for a comparable grassroots unit - 40% labour ($19.2M), 45% materials ($21.6M), 15% other and indirects ($7.2M) - applies a location factor of 1.18 to the total, and adds three years of 3% per year escalation as 9% simple, arriving at $61.7M. Site realities: the tie-in sits inside an operating unit where hot-work permits are limited to six hours per day, the site is in a non-attainment air district requiring extra permitting, and the local craft market is competing with two large data-centre projects paying roughly 20% above prevailing. Engineering is 10% complete.
1. Critique the transfer and reconcile it to a corrected base cost. Where is the location factor misapplied?
Read the worked answer
Error 1 - the factor is applied to the wrong scope. Location factors reflect differences in labour rates and productivity, plus selected indirects. Commodity materials - structural steel, valves, pipe - trade at national or global prices and do not become 18% dearer because of the site. Applying 1.18 to all $48M overstates materials by roughly 21.6 × 0.18 = $3.9M.
Error 2 - the grassroots analogy ignores brownfield productivity. A tie-in inside a running unit, with six hours of hot-work window per day, is not a greenfield erection. A productivity penalty of 1.20–1.30 on labour is standard for this environment; here six hours of window and permit queues argue for the upper end.
Reconciliation (GC base date)
- Labour: 19.2 × 1.25 (brownfield) × 1.18 (location/wage) = $28.32M
- Materials: 21.6 × 1.00 = $21.60M
- Other/indirects: 7.2 × 1.18 = $8.50M
- Corrected base: $58.4M
Error 3 - escalation. 3% simple over three years is also 9% on a base that has not been escalated to the right date; compounded to the construction midpoint over 3.5 years, 1.03^3.5 = 1.109, giving $64.8M before contingency - already above the estimator's $61.7M with no contingency and none of the missing items below.
2. What has the estimate missed entirely, and what is the likely magnitude of each?
Read the worked answer
Missing items, roughly in order of size
- Schedule-driven indirects. Six-hour hot-work windows may stretch field duration by 25–40%, and supervision, scaffolding, equipment rental, QA/QC and site overhead all run on duration, not on installed quantity. On a 15% indirect share, a 30% duration extension adds roughly 0.15 × 0.30 = 4–5% of base - $2–3M.
- Permitting in a non-attainment district. New-source review, emission offsets and a longer approval calendar. Offsets in particular can be a real, purchasable cost, not just delay.
- Operational interference costs - unit shutdowns, temporary facilities, tie-in window penalties, hot-tap and freeze procedures, and the process safety documentation that goes with them.
- Site logistics in a live unit - laydown restrictions, escorted access, small-batch deliveries, extra material handling.
- Craft wage pressure - two data-centre projects paying 20% above prevailing is not a location factor; it is a live market condition that will show up in bid returns and in retention of the crew. This argues for a labour-market escalation curve distinct from general escalation.
- Owner costs excluded but not stated - owner's project management, pre-commissioning support, insurance, spare parts.
Judgement for the exam answer: the dominant risk is not the location factor at all. It is duration-driven cost in a hot-work-restricted, non-attainment brownfield, and that risk is invisible in a multiplicative factor chain.
3. Engineering is only 10% complete. What would you require to move this to a Class 3 estimate, and how would you present the current uncertainty?
Read the worked answer
To reach Class 3, require:
- 30–60% engineering definition, with P&IDs issued for design and a defined tie-in scope.
- A duration model built from the six-hour permit window, not a generic productivity factor - this is what converts the biggest risk into a number.
- Firm quotations for long-lead items and bulk materials at the corrected base date.
- A permitting plan with dates from the air district, not an assumption.
- Craft labour survey for the local market, including the data-centre competition.
- Site-specific indirects built up from a staffing plan tied to that duration.
Presentation now Give a range, not a point: the corrected base of ~$65M plus contingency sized by risk analysis, quoted as P10/P50/P90 with the assumptions named. Class 4 accuracy on the order of −30%/+50% is the honest label, and the largest single uncertainty is field duration, so present it as a tornado diagram showing duration sensitivity first, wage escalation second, and the location factor a distant third.
The discipline point: never present a deterministic number that a factor chain produced, without showing which factor drives the answer. Here it is time, not geography - and the estimator's method was structurally blind to time.