Case Study 36-1 — The Building That Failed Its Test

Northgate Outpatient Pavilion, Meridian Health System. March through June, Year 2.

All people, companies, and projects in this book are illustrative composites.

Setup

Northgate is on the performance compliance path for the adopted energy code. Trellis Engineering's energy model traded a glassier building than the prescriptive tables would have allowed against higher-efficiency mechanical equipment and a tighter envelope. To make the trade balance, the model assumed a whole-building air-leakage rate at or below a stated maximum, and Halvorsen + Pike wrote that maximum into Division 07 along with a required whole-building air-leakage test.

The enclosure is 38,500 SF of unitized aluminum-and-glass curtain wall by Vantage Facade Systems, 21,000 SF of architectural precast panels, and 34,000 SF of TPO membrane roof. Perimeter interior framing and gypsum board are Summit Interior Systems. Perimeter fire-safing and smoke seal at the slab edge are Bulwark Firestopping. The commissioning agent, Amara Boateng, is engaged directly by Meridian, not through Kestrel.

The dried-in milestone is March 28, Year 2. Substantial completion is September 18, Year 2, and every calendar day of slip past it costs Kestrel $10,650 — $5,150 of extended general conditions plus $5,500 of liquidated damages.

The test is scheduled for Friday, March 7, Year 2.

What happens

The test

Four trailer-mounted fans in four ground-floor doorways. Every intentional opening — louvers, relief dampers, plumbing vents, elevator shaft vents — sealed the night before by a two-person crew working from a 34-item checklist. Three curtain-wall units on the west elevation were not yet set and were closed with gasketed, taped plywood blanks, which Amara documented as a test deviation.

The building came in at 2.4 times the specified maximum leakage rate — with roughly a hundred square feet of enclosure temporarily performing better than the finished construction ever would.

Amara Boateng: "The number is the number. But a number without a location is just an argument. Give me four days with a camera."

The infrared survey

Four days, building held under pressure, thermal camera plus smoke pencil confirmation. Three findings.

Findings 1 and 2 were accessible and cost roughly $34,000 combined to correct. Finding 3 was behind Summit's perimeter framing on three floors.

Dani Okonkwo: "How did nobody catch this? We built a curtain-wall mockup."

They did. Eleven mockups on the job. The curtain-wall mockup was two units wide, it was water-tested, and Dale Whitcomb approved it in month seven. It stopped at the bottom of the units. The slab edge was not in it.

The repair, priced

The retest, run June 2, Year 2, passed with margin. The dried-in milestone held. Substantial completion was never threatened, because the work was absorbed with premium time rather than duration — which is why the $21,100 line exists.

What it would have cost at the mockup

$236,710 ÷ $8,600 = 27.5. Same defect. Nine months of difference. Twenty-seven and a half times the price.

The argument over who pays

Eleven weeks. Both sides had a real position, which is exactly why it was expensive.

Kestrel's position (Ray Alvarez, in the April OAC meeting): "Detail 7 on A-521 shows the safing slot with a hatch pattern and a note reading 'air seal per specification.' There is no product, no dimension, no sequence, and no indication of which trade installs it. The specification requires a continuous air barrier but does not detail this transition. We priced the drawings. There is no scope in any of our subcontracts for an air seal at the slab edge, because there is no scope in the documents."

Halvorsen + Pike's position (Dale Whitcomb): "The specification requires a continuous air barrier across the entire thermal envelope. The word is continuous. Achieving continuity across a transition between two systems is means and methods, which is the contractor's. We are not required to draw every square inch of the building. And Kestrel's own submittal for the air barrier system was approved with a manufacturer's standard detail for exactly this condition in the package."

That last sentence was the hard one, and it was true. The approved air barrier submittal included the manufacturer's typical slab-edge detail. Nobody had extracted it, assigned it, or bought it.

Priyanka Sethi, for Meridian: "I need to be direct. Meridian is not paying twice for a continuous air barrier that the contract already requires. But I also understand that nobody put this in a scope sheet, and I would rather spend money on a building that works than on a year of letters."

The settlement, reached in late May:

Kestrel's $121,510 was charged to the GMP construction contingency of $1,320,000. That is worth understanding precisely, because it is how a CM at Risk contract actually distributes this kind of loss. Drawing the contingency did not raise Meridian's price — the GMP is still $47,500,000. What it did was consume $121,510 of the money that would otherwise have been returned as unused contingency at the end of the job and split 75% to Meridian, 25% to Kestrel. So the economic loss lands as $91,133 of forgone savings to Meridian and $30,377 to Kestrel — and Kestrel additionally spent eleven weeks of senior staff time it will never bill.

Analysis

What actually failed was not workmanship. It was a scope gap dressed as a detail.

Every one of the three subcontractors who touched that joint was correct from inside their own contract:

• Vantage installed the units and their perimeter seals per the approved shop drawings. Their air barrier plane stops at the edge of their system.
• Bulwark Firestopping installed a fire-rated safing and smoke seal assembly that meets its listing. Fire-safing is not an air barrier and is not tested as one.
• Summit Interior Systems started at the floor track, which is where their scope starts.

The air seal was in nobody's scope because it was in nobody's scope sheet, and it was in nobody's scope sheet because at buyout nobody traced the air barrier across the transition and asked who owned it. This is the Chapter 16 lesson in its purest form: scope gaps live between subcontracts, and an air barrier is nothing but a list of places where two subcontracts meet.

Why the mockup did not catch it. Because a mockup tests what is in the mockup. The curtain-wall mockup was designed to prove the curtain wall, and it did — the units performed. The failure was at the boundary of the system, and boundaries are precisely what a single-system mockup excludes. That is a design flaw in the mockup program, not in the curtain wall.

Why the settlement landed where it did. Notice what each party paid for. Meridian paid for work — the air seal itself, which somebody would always have had to install and which was never in the GMP. Kestrel paid for timing — the demolition, the reframing, and the retest, all of which exist only because the problem was found in March instead of month seven. That is a defensible allocation and it is worth internalizing, because it is how these settlements usually resolve when both parties are acting reasonably: the owner pays for the scope, the contractor pays for the discovery timing.

What Kestrel actually lost. $30,377 of forgone savings share — the direct dollar hit is smaller than the $121,510 headline, because the contingency structure spread it. But add eleven weeks of Ray's and Dani's attention during the busiest quarter of the job, a construction contingency now $121,510 thinner with eight months of job left to run, and a measurable amount of standing with Pri Sethi at exactly the moment Kestrel most needed it. The thinner contingency is the one that should worry you: it is the reserve for every risk still ahead, and it just paid for a risk that was behind.

The two process changes Kestrel adopted

Nadia Haddad required both on every subsequent project with an air barrier requirement. They are cheap, and neither one is clever.

Change 1 — the air barrier continuity plan, produced at buyout. Before the first enclosure subcontract is awarded, somebody traces the air barrier on a wall section set with a colored pen, from the bottom of the foundation, up the wall, across every transition, over the roof, and back down — without lifting the pen. Every place the pen must lift becomes a row in a register with four columns: the detail reference, the product, the named subcontractor, and the hold-point inspection. The register is an attachment to every affected subcontract. On Northgate the exercise would have taken about forty minutes and found the slab edge.

Change 2 — mockups must include their transitions, and something must be air-tested early. Two rules. First, no system mockup is approved unless it includes the transition above and the transition below — the mockup proves a boundary, not a field condition, because the field condition is not where anything fails. Second, on any project with a whole-building air-leakage requirement, an early sectional test is scheduled: a mockup chamber, or a first-floor area isolated with temporary partitions, tested at the first opportunity after the enclosure system is in place. Cost on a job like Northgate is on the order of $14,000. It buys the difference between a detail revision and a demolition.

Discussion questions

1. Kestrel absorbed the cost of discovery timing and Meridian paid for scope. Argue the opposite allocation as persuasively as you can. Which argument would you rather make in front of a mediator, and why?
2. The approved air barrier submittal contained the manufacturer's typical slab-edge detail. Who had a duty to extract it and assign it — Kestrel's project engineer reviewing the submittal, the architect reviewing it, or the air barrier manufacturer's representative? What does your answer imply about how you review submittals in Chapter 25?
3. The retest passed with margin, the milestone held, and the building performs. In what sense was this still a failure? Name three costs that never appeared on a cost report.
4. Change 2 requires an early sectional test costing about $14,000 on every project with an air-leakage requirement. On a job where the test would have found nothing, that $14,000 is pure loss. How do you defend the policy to a project executive looking at four such jobs in a row?
5. Fire-safing is a fire-rated assembly and is not an air barrier. Name two other places on a building where a code-required assembly is routinely mistaken for a control layer it does not provide.

Your turn

Take a wall section from any project you have access to — a real set, a set from a class, or the described Willow Street assemblies in Appendix K. Trace the air barrier with a pen from the footing to the roof and back down, without lifting it.

Every place you have to lift the pen, write one row: detail reference · product · which subcontractor installs it · what inspection confirms it. Bring the register to the next person who will look at it with you and ask a single question about each row: "Is that in somebody's scope sheet?"

Most readers find between four and nine lifts on their first attempt. If you find zero, you are tracing the drawing you wish you had rather than the one you were given.