Case Study 1 — Eleven Disagreements

Four days, one wall assembly, and $386,900 moved out of the dark

Setup

Project: Northgate Outpatient Pavilion — 132,000 SF, four stories, CM at Risk with a $47,500,000 GMP for Meridian Health System. Where in the job: Week 6. Notice to proceed was March 3, Year 1. Mass excavation is running. The GMP is set, but the enclosure trades are not bought out. Who: Dani Okonkwo, field engineer, nine working days into their first job. Margo Deacon, general superintendent. Ray Alvarez, senior PM. Dale Whitcomb, project architect at Halvorsen + Pike.

(All people, companies, and projects here are illustrative composites.)

The assignment, verbatim: "Four hundred and eighty sheets. Nineteen hundred pages of project manual. Northgate exterior wall assembly. I want every place the drawings and the specifications disagree about it. On my desk Thursday."

The constraint that made it urgent: the architectural precast and the curtain wall are the next two subcontracts to be bought. Once a scope sheet is signed and a subcontract is executed, every ambiguity in it becomes a negotiation instead of a clarification. Kestrel had about nine working days of leverage left.

What happened

Monday: the wrong method

Dani started at sheet A-101 and read forward. By 3 p.m. they were on A-118, had read 18 sheets of floor plans, and had found nothing, because floor plans do not describe wall assemblies.

Ray stopped it. "You're reading the set like a book. Read it like an assembly. Pick the exterior wall. Now go get every document on this job that touches the exterior wall and put them in one pile."

That pile turned out to be: sheets A-011 (wall types), A-201 through A-206 (exterior elevations), A-311 (exterior wall sections), A-511 and A-512 (exterior wall details), A-521 (curtain-wall details), A-601 (door schedule) and A-603 (window and glazing schedule), G-002 (code summary) and G-006 (energy compliance narrative), S-201 and S-301 (edge-of-slab framing and spandrel conditions), and specification sections 03 45 00, 05 40 00, 07 21 00, 07 27 00, 07 92 00, 08 44 00, 09 21 16, plus Division 01.

Twenty-two documents. One wall.

Tuesday and Wednesday: the six passes

Dani ran the method. Division 01 first — which is where item 7 came from, before a single drawing was opened. Then drawings against specs by assembly. Then continuity: follow the air barrier through every detail and confirm it never stops.

That continuity pass produced the finding Dani almost talked themself out of. Detail 4/A-511 showed the fluid-applied air-and-water-resistive barrier applied over the face of the continuous mineral wool insulation. Detail 7/A-512, one sheet later, at the curtain-wall perimeter on the same elevation, showed it applied directly to the sheathing, under the insulation.

Dani's first instinct: "I must be misreading one of them."

Their second instinct — the correct one — was to log it anyway and let the architect say which it was. That is the discipline the whole method depends on. You are not qualified to decide that a conflict is a misunderstanding. You are qualified to write it down.

Thursday: the log

The two that were worth real money

Item 1 — one inch, $96,070

Wall section 3/A-311 showed 2" of continuous insulation. Section 07 21 00 required the assembly to achieve R-15 continuous. Two inches of the specified mineral wool does not reach R-15; three does.

The cost was never the insulation:

Only 40% of that number is insulation. The rest is what one inch of added assembly depth drags behind it: the clip length, the embed geometry, the window returns, and three weeks of shop-drawing rework.

Resolution. Dale Whitcomb's RFI response confirmed the specification governed: R-15 continuous was an energy-code compliance requirement carried in H+P's permit submission, and the 2" dimension on A-311 was a leftover from an earlier design iteration. H+P issued a revised A-311 and A-511 by bulletin. Because it landed in week 8 rather than week 58, the precast fabricator absorbed the connection redesign into its shop-drawing cycle at no premium, and the entire item was priced into the precast and insulation subcontracts before award. Meridian funded the delta as a design clarification against the GMP contingency.

Actual cost to Kestrel: $0. Actual cost to the schedule: 0 days.

Item 5 — the gap nobody wrote

The drawings said "6" metal studs @ 16" o.c." and did not state a gauge, which is normal and correct: gauge is a means-and-methods determination. Section 05 40 00 set a deflection limit of L/600 for framing supporting a rigid veneer. The level-1 floor-to-floor height is 16'-2".

Run those three facts together and the light-gauge stud that every bidder priced does not satisfy the criterion. The heavier gauge across 21,000 SF of backup, plus additional deflection track and bridging, came to $62,000.

Neither document was wrong. The drawing was silent where the specification was governing, and the specification was silent about the span that made it bite. No single reader of either document would have found it. Only the side-by-side found it.

Resolution. This one Kestrel largely owned, and correctly so — determining stud gauge to meet a specified deflection criterion is the contractor's obligation. But finding it in week 6 meant it went into the scope sheet before the framing subcontract was bid, so every bidder priced the same thing and Kestrel bought it at market instead of negotiating a change against a signed subcontract later. Ray's estimate of the difference between those two outcomes: roughly $40,000, plus a fight.

Analysis: why the method worked

It read by assembly, not by sheet. Twenty-two documents from six disciplines and two document families, all describing one wall. Nine of the eleven items were invisible from any single document.

It ran Division 01 first. Item 7 — a required, water-tested envelope mockup that appeared on no drawing, in no logistics plan, and in nobody's schedule — was found before a drawing was opened. It carried $47,000 and a five-week lead time, which in week 6 is an inconvenience and in week 40 is a schedule event.

It logged uncertainty instead of resolving it. Dani did not decide anything. Every item went to the architect or into a buyout scope sheet. That matters legally as well as practically: most general conditions impose a duty to report discovered discrepancies, and a discrepancy you find and do not report can become yours.

It happened while Kestrel still had leverage. Every one of these eleven items was resolved before the relevant subcontract was executed. After execution, each one becomes a change order negotiation with a subcontractor who has already been awarded the work and has no competitive pressure.

What it did not do: the method found conflicts in the exterior wall only. Nobody had yet run it on the roof, the interior partitions, the imaging suite, or the MEP systems. Margo's next assignment for Dani was the roof. It produced fourteen items.

The honest accounting

Not every item was a win. Item 3 (precast panel thickness) came back from H+P confirming the 5" panel and revising the reveal depth in the specification — Kestrel's $28,000 exposure evaporated, which is exactly as good an outcome as being paid. Item 6 (perimeter fire containment with no specification section) turned into a three-week argument about whose scope it was and was ultimately split between the drywall and firestopping packages at $74,000, funded by Meridian as a design omission. Item 11 (the vapor retarder) produced an RFI response that read, in essence, "no vapor retarder is required in this climate zone for this assembly" — worth zero dollars and enormous peace of mind, because Kestrel now had that answer in writing from the architect rather than as an assumption.

Total time invested: four days of a field engineer. Fully burdened, roughly $2,400.

Discussion questions

1. Ray stopped Dani from reading the set sheet by sheet on Monday. What specifically made "read by assembly" the higher-yield method, and what kinds of discrepancy would a sheet-by-sheet read find that an assembly read might miss?
2. Item 5 was arguably always Kestrel's cost to bear. Why was finding it in week 6 still worth roughly $40,000?
3. Items 9 and 11 carried a dollar exposure of $0. Make the case that they were among the most valuable findings on the list.
4. Kestrel had "about nine working days of leverage" before the enclosure subcontracts were executed. Describe how the value of an identical discrepancy changes at week 6, week 20, and week 58 — and name the specific mechanism that changes at each point.
5. Suppose Dani had found item 1 during the bid phase on a hard-bid job, said nothing, priced the 2" insulation, and planned to claim the difference after award. Walk through what happens to that strategy — commercially, contractually, and to the company's bonding relationship.

Your turn

Take the roof assembly on a project you can get documents for — your own job, a school district's published bid set, or the Willow Street package in Appendix K.

Build the pile first: every drawing sheet and every specification section that touches the roof. Expect twelve to twenty documents. Then run passes 2, 5, and 6 from §7.10 of the chapter — drawings against specs by assembly, details against sections for continuity of the membrane and the vapor retarder, and the seams (roof-to-wall, roof-to-curb, roof-to-equipment, roof-to-parapet, and the boundary between the roofer's scope and the sheet-metal scope).

Log at least six discrepancies in the ten-field format from §7.10, with a proposed resolution and a dollar figure for each. Then do the thing that turns this from an exercise into a habit: pick the one you are least sure about and write the RFI anyway.