Case Study 9.1 — Nine Weeks in a Drawer: The Northgate Curtain-Wall Recovery

All people, companies, and projects in this book are illustrative composites. The numbers are internally consistent and typical of the situation described; they are not drawn from a specific real project.

Setup

Project: Northgate Outpatient Pavilion — 132,000 SF, four stories, CM at Risk with a $47,500,000 Guaranteed Maximum Price (GMP), 565 calendar days, substantial completion September 18, Year 2. Enclosure: 38,500 SF of unitized aluminum-and-glass curtain wall (Vantage Facade Systems, project manager Renata Vogel), 21,000 SF of architectural precast, 34,000 SF of TPO roof. The milestone at stake: building dried-in, March 28, Year 2. The exposure: $10,650 per calendar day (CD) — $5,150/CD extended general conditions plus $5,500/CD liquidated damages.

Kestrel Construction Group awarded the Vantage subcontract on April 9, Year 1 — thirty-seven days after Notice to Proceed. Ray Alvarez had pushed hard for that date, over some grumbling that the design was not settled enough to buy. The reason was arithmetic: the chain from subcontract award to the first unit hanging on the wall runs 38 weeks, curtain wall installation had to start January 5, Year 2, and 38 weeks before January 5 is April 14, Year 1. The award carried five days of float on a thirty-eight-week chain.

Five days is not a cushion. It is a rounding error.

What Happens

June 18, Year 1 — the submittal review meeting

Vantage's first shop-drawing submission arrived on schedule. It was 214 sheets. Dale Whitcomb at Halvorsen + Pike reviewed it with Ruth Caldwell's office and returned it stamped Revise and Resubmit.

Dani Okonkwo had logged the return and flagged it, and now stood at the end of the trailer table explaining what they had found.

"They submitted the typical conditions completely," Dani said. "Every typical unit, every typical stack joint, all the anchors. What they did not submit is the transition details. The precast-to-curtain-wall vertical joint on all four elevations, the two-story lobby glass at the entrance, the café's corner condition, and the roof-edge termination. Roughly forty percent of the details on the north elevation are marked 'by others.'"

Ray asked the only question that mattered. "Who is 'others'?"

"That's the point," Dani said. "Nobody. Vantage assumed Cornerstone would detail the precast side. Cornerstone's shops show the panel and stop at the panel face. Neither subcontract says who owns the joint."

Margo Deacon, who had been reading the plan set upside down from across the table, put her finger on the north elevation. "So the part of my building that faces the weather is the part nobody drew."

June 18 – August 14, Year 1 — the nine weeks

The resubmittal took Vantage four weeks, because the missing details required a scope determination between two subcontractors and a design decision from H+P about the reveal at the precast joint. The re-review took another three. A third partial round on the lobby glass took two more.

Total elapsed beyond the plan: nine weeks.

Wei Chen ran it through the schedule on August 12 and brought the result to Ray with no commentary, which is how Wei delivers bad news.

Nine weeks of fabrication slip became forty-two calendar days on the dried-in milestone after the small amount of float inside the installation sequence absorbed the difference.

Meanwhile, in the same office, in the same week, a different fire was burning. Ironbridge Steel had missed its mill rolling slot because the anchor-bolt and embed submittal sat eleven days in Kestrel's own office and then absorbed Caldwell Structural's full fourteen-day contractual review. Steel erection had slipped from August 4 to August 27 — 23 calendar days, on the critical path.

Nadia Haddad, Kestrel's Vice President of Operations, had two acceleration decisions in front of her in the same week.

August 14, Year 1 — the decision meeting

Ray laid out the enclosure options.

Option A — wait for full approval and fabricate in one release.

Option B — resequence: release fabrication by elevation.

Approve the south and east elevations — the two with no special conditions — and release them for fabrication immediately. Let the north elevation, with the lobby and the café's two-story glass, finish its review. Install in three fabrication releases instead of one. Build temporary enclosure at the north-elevation openings so the interior could still be conditioned and released by zone.

Option B saved $282,650 and 33 calendar days.

Pri Sethi, Meridian's owner's representative, asked the question owners always ask and should always ask: "If the resequence is that much cheaper, why wasn't that the plan all along?"

Ray gave her the honest answer. "Because a single fabrication release is cheaper, cleaner, and safer when your shop drawings are approved on time. We are only buying this option because we already lost nine weeks. Resequencing is not a strategy. It is a recovery."

Nadia approved both accelerations — $168,000 for steel and $148,000 for the facade — in the same meeting.

Analysis

What the money actually did

The two decisions cost $316,000 of premium in a single week, drawn against Northgate's $1,320,000 construction contingency.

$316,000 / $1,320,000 = 23.9% of the entire construction contingency
                        consumed in week 25 of a 565-day project

And the contingency is not free money. Unused GMP contingency on Northgate splits 75% owner / 25% Kestrel. Every dollar drawn costs Kestrel twenty-five cents of profit it will never see.

$316,000 x 25% = $79,000 of Kestrel's savings-split upside, gone

Nobody in that meeting mentioned that number out loud, which is a habit worth breaking. The contingency draw was correct — the alternative cost $484,200 on the facade alone — but "correct" and "free" are different words.

What worked

The submittal log was back-scheduled, and somebody read it weekly. Dani had built the log with lead times counted backward from the field need date, so a returned submittal in June was immediately visible as a threat to a March milestone. Kestrel found the problem in week 15 of a 38-week chain, when there were still options. A contractor who tracks submittals monthly against a bar chart finds this in December, when the only remaining option is Option A.

The resequence was possible because the building has four elevations. That sounds obvious. It is not. The option existed because the units on the south and east elevations were genuinely typical and independently fabricable — which was a design characteristic somebody had noticed. If Vantage had detailed the system as a continuous horizontal band around the building, there would have been nothing to release early.

Somebody priced both options before choosing. Ray did not argue for the resequence; he arithmetic-ed for it. Nadia's decision took four minutes because the table was on the page.

What failed

The scope gap was created at buyout and discovered at submittal. The precast-to-curtain-wall joint belonged to nobody because neither scope sheet named it. That is a $148,000 gap that could have been closed with one sentence in one scope sheet, written in March, at a cost of about zero. Scope gaps live between subcontracts, and finding them is a buyout discipline, not a field skill.

The award had five days of float on a thirty-eight-week chain. Kestrel treated April 9 as an aggressive award and it was — but "aggressive" measured against a 38-week chain with a rigid downstream milestone is still nearly zero margin. The right conclusion is not "award faster." It is that the design-assist engagement, which would have had Vantage detailing the transitions during design instead of after award, was the item that should have been bought.

The cost nobody put on the sheet

Both accelerations landed on the same elevation at the same time.

The steel acceleration added a second erection crew and Saturday premium time and resequenced enclosure by area. The facade resequence pushed the north elevation's precast and masonry veneer work into a compressed window while curtain-wall staging was already occupying the same access. Weeks 33 through 37 on the north elevation had more trades in less space than any other stretch of the job.

In week 34, on a Tuesday morning at 7:20, a frame scaffold on the north elevation at Level 3 had been partially modified overnight by a different trade running conduit. A plank was lifted and never re-secured. A mason tender, Emiliano "Milo" Serrano, stepped onto it, the plank shifted, and he went down onto the platform and caught himself on the top rail. No injury.

Bea Salgado stopped work on the elevation for the day and ran the investigation. She found three failures: no competent-person inspection tag for that shift; a scaffold modified by a trade that had not erected it, with no re-inspection; and a crew running behind after the acceleration, with an unwritten "make it up" pressure.

The third one is the finding nobody wanted to write down, and Bea wrote it down.

That is the real lesson of this case. The acceleration table had two columns, cost and days. It did not have a third column for density — how many people would be doing how much work in how little space. Trade stacking is a real, priceable consequence of every recovery decision, and it shows up as lost productivity, rework, and near-misses long before it shows up as an incident. Schedule pressure is a hazard exactly like an unguarded edge.

Discussion Questions

1. Kestrel awarded Vantage 37 days after NTP with five days of float on a 38-week chain. Was that the right decision? What would you have done differently in preconstruction, and what would it have cost?
2. The precast-to-curtain-wall joint belonged to no subcontract. Write the sentence you would add to the curtain-wall scope sheet and the sentence you would add to the precast scope sheet so that the gap cannot recur. Who should own the joint, and why?
3. Nine weeks were lost across three submittal cycles. Assign responsibility as best you can from the facts given, and name the three documents you would want in front of you before arguing about who pays the $148,000.
4. The acceleration decisions cost 23.9% of the construction contingency in week 25 of a 565-day job. What should Ray tell Pri Sethi and Nadia Haddad about the remaining contingency, and when?
5. Bea's third finding — schedule pressure — is the one most investigations omit. Why is it uncomfortable to write down, and what changes on a project when it is written down?

Your Turn

You are the project engineer. It is June 19, Year 1 — the morning after the shop drawings came back Revise and Resubmit. Nobody has yet calculated the schedule impact.

Produce two things, in one page total:

(a) A back-scheduled critical-item alert to Ray: what date the shop drawings must be approved to hold the January 5 installation start, how many weeks of slip you are currently carrying, and what date the decision to resequence must be made in order to still be useful. Use the 38-week chain from §9.5.1 of the chapter.

(b) A three-line entry for the project risk register, in the format from Chapter 6: the named risk, probability and impact, the risk owner, the response, and the contingency amount you would earmark against it today — before you know the answer.

Then check your (b) against what actually happened. The number you earmark on June 19 with incomplete information is a better test of judgment than the number anyone can calculate on August 14.