Case Study 1 — Fifty-Nine Modules: The Northgate Corridor Prefabrication Program

All people, companies, and projects in this book are illustrative composites. Every dollar figure is teaching data, consistent within the book and not drawn from any actual project.

Setup

The project. Northgate Outpatient Pavilion. Meridian Health System, owner. Kestrel Construction Group, construction manager at risk with a $47,500,000 guaranteed maximum price. Four stories, 132,000 gross square feet. Contract time 565 calendar days from a March 3, Year 1 notice to proceed. Liquidated damages $5,500 per calendar day, extended general conditions $5,150 per calendar day, combined exposure $10,650/CD.

The scope. The main corridor spines on levels 2 and 3 — 1,180 linear feet of overhead. Thirty modules on level 2, twenty-nine on level 3; 59 modules averaging 20 LF, each one a welded steel frame carrying supply and return duct, medical gas piping, domestic water, hydronic supply and return, a sprinkler main, cable tray, and a conduit rack. Insulated in the shop. Pressure-tested in the shop. Delivered on a flatbed and set with a telehandler. Each finished module weighs roughly two thousand pounds.

The people.

• Ray Alvarez — Kestrel senior project manager
• Grace Lindqvist — Kestrel VDC/BIM manager; owns the model and defended this program to get it funded
• Margo Deacon — Kestrel general superintendent
• Sofia Marchetti — project manager, Cardinal Mechanical (HVAC, plumbing, medical gas); the shop is hers
• Devlin Achebe — project manager, Halcyon Electric; sends a crew into somebody else's shop for four months
• Nadia Haddad — Kestrel vice president of operations; the executive sponsor, which mattered more than anything technical
• Pri Sethi — owner's representative, Meridian Health System; the person who had to make a hospital hold still

The constraint that made it possible. Two things were already true when this started, and if either had been false the program loses money. First, Grace had a coordinated model of those corridors that had already been through clash detection with Cardinal and Halcyon. Second, Cardinal Mechanical had shop capacity and wanted the work. Neither of those is a technology.

The constraint that made it hard. To fabricate a corridor, you have to know what is in that corridor months before anyone would normally have to decide. Kestrel needed a model freeze on October 6, Year 1 — roughly twenty weeks earlier than the same information would have been needed under a stick-built sequence.

What Happens

The freeze

Pri Sethi did not like the freeze date and said so.

"You're asking me to tell six clinical departments that the corridor is finished in October. Two of them haven't finished arguing about where the clean supply is going."

"I am," Ray said. "And I'll tell you what you get for it, in your terms. If the corridor freezes October sixth, we set the overhead in twenty-seven working days instead of forty-six, and about six thousand eight hundred man-hours of ductwork and pipe come off lifts in your building. If it doesn't freeze, we stick-build it, I have no complaint, and you keep the flexibility. That's a real trade and I'm not going to pretend it isn't."

"What happens if I miss the date?"

"Then whatever changes after October sixth costs more than it would have, because a shop has already cut steel to a dimension. I'd rather tell you that now than send you an invoice for it in February."

She took it to the hospital operations committee and got the freeze. Not because Ray was persuasive — because somebody with authority asked six departments a specific question with a date on it. That is the entire mechanism. The design freeze is not a technical requirement; it is an owner decision that somebody has to go get.

What the coordination actually cost

The model that had passed clash detection was not the model the shop could cut from. Clash detection asks do these two objects occupy the same space. Fabrication asks what is the exact length of this piece of pipe and where exactly is the hole in this steel frame. Taking the corridor model from coordination-level development up to fabrication-level development was seven weeks of Grace's technician, Cardinal's detailer, and Halcyon's detailer working in the same file, with a standing Thursday session that Sofia and Devlin both had to attend.

Devlin Achebe put the awkward part on the table in week two.

"I want to be straight about this. My detailer is spending three days a week in Sofia's model. Sofia's shop is getting a fabrication contract out of it. What is Halcyon getting?"

Nobody had a good answer for about ten minutes. What Ray eventually did was write it into the subcontract: Halcyon's detailing hours were paid as a defined line, and Halcyon's conduit and tray were fabricated in Cardinal's shop by Halcyon's own people, on Halcyon's payroll, in a bay Cardinal made available. The party doing the work got paid for the work. That conversation, and not any modeling technique, is why the program survived past November.

The bill for all of that upstream discipline was real and it is in the table below. It is also the line that most prefabrication business cases forget entirely.

What it cost

On direct cost it was a wash. Kestrel traded expensive field labor for cheaper shop labor and then spent almost the whole difference on coordination, engineering, transport, and rigging. If you sell this program on the promise that the cost goes down, you will be wrong, and you will not get a second program.

What it returned

Fabrication ran November 4, Year 1 through February 14, Year 2. Setting ran January 12 through February 20, Year 2 — twenty-seven working days for fifty-nine modules, in corridors that had been enclosed by area ahead of the March 28 dry-in milestone.

Two things in that table need saying plainly.

The 6,800 man-hours were not saved. They were moved. They moved from a lift in a corridor to a bench in a shop, and Kestrel paid for them in both places. The honest productivity number is the 2,500 man-hours, a 22% improvement, and that is the number a defensible program is built on.

Nineteen work days is not nineteen days on the contract. MEP rough-in was on the critical path after the steel delay pushed erection from August 4 to August 27, Year 1. But in-wall rough-in, above-ceiling inspections, and the enclosure sequence had their own logic. When Wei Chen ran the compressed durations through the network, the nineteen work days in those corridors produced nine calendar days at substantial completion. The rest was absorbed.

The safety accounting, which is the part I would lead with

Look at what those 6,800 man-hours were, physically. They were sheet metal workers, pipefitters, sprinkler fitters, and electricians standing on scissor lifts and rolling scaffolds at nine to twelve feet, working overhead, hanging and joining heavy material above their heads, in a corridor with other trades trying to get past them.

In the shop, the same work happens with the rack lying on a jig at waist height, in light, indoors, with the material within reach and a fitter and an inspector standing next to it.

That is not a hazard control. That is hazard elimination — the top of the hierarchy in Chapter 24 — applied to about a third of the corridor's total exposure. It is also why the welds were better, which is the same fact from a quality angle.

And the honest other half. The program created two exposures it did not have before. Setting a two-thousand-pound assembled module in a partially built corridor is a suspended-load and struck-by operation with a crush zone underneath it, fifty-nine times. Offloading fourteen flatbeds on a tight urban site with a north property line twenty feet from an active clinic is its own problem. Kestrel wrote a rigging plan for the sets, restricted the corridor below and beyond each pick, and Margo Deacon personally walked the first six sets. Bea Salgado's condition for supporting the program was that she got to review Cardinal's shop safety program before eleven hundred man-hours of Kestrel's schedule went into it. You have not removed risk if you have only relocated it into somebody else's building.

The Two Things That Went Wrong

Neither was a fabrication failure. Cardinal Mechanical did not build one bad rack in fifty-nine. Both failures were upstream, and both were ours.

Problem 1 — the camber assumption, $6,800

Back in the fall, Grace had scanned the level 1 imaging suite bay after steel and deck and found a beam bottom flange 1¾ inches lower than the coordinated model. It was not an erection error; it was a camber and deflection assumption in the shop detail that never got carried back into the coordination model. She corrected it, Cardinal's detailer re-routed 34 LF of duct on paper, and everybody moved on feeling good about a $2,100 fix.

Nobody asked where else that assumption lived.

On February 3, Year 2, the first two level 3 modules went up and the hanger drops were an inch and three-quarters too long in both bays. Same condition, same cause, a floor and a half away, on racks a shop had already built to the model.

The supplemental scan found the same condition in one more bay. That rack's drops were adjusted on the floor before it was lifted, at bench height, for essentially nothing — which is the entire lesson, arriving one bay too late.

The mechanism: a stick-built system absorbs dimensional error at every joint. A prefabricated one concentrates all of the tolerance at the interfaces, where it has to be right the first time. When you shift to prefabrication you are not only building differently, you are changing where error is allowed to live, and you have to move your verification to match. Kestrel treated a systemic model defect as a local one.

Problem 2 — the change after the freeze, $18,400

On December 9, Year 1 — nine weeks after the October 6 model freeze, with 41 of 59 modules already built — Meridian's clinical operations group added a second negative-pressure isolation exam suite off the level 3 corridor. It needed a medical-air branch and a relocated zone valve box in a corridor segment that existed as steel and insulated pipe on a shop floor two hundred miles away.

Pri Sethi did not fight it. She had signed the freeze, the cost was documented contemporaneously the same week, and it went through as a straightforward change order — which is the only reason this line is $18,400 and not an eight-week argument. Compare that to CO #14, where the same organization made a late equipment decision, gave a verbal go-ahead, built it before it was priced, and left Kestrel $43,650 short. Same owner, same kind of decision, opposite paperwork, wildly different outcome.

The lesson is not "owners change things." Owners always change things. The lesson is that a fabrication program converts a normal late change into an expensive one, and the only defense is that everyone agreed in advance what the freeze meant and what breaking it costs.

Analysis: Did It Pay?

The arithmetic.

Schedule value: 9 CD × $10,650/CD = $95,850

Direct cost saving: $25,000

Less the two failures: −$6,800 − $18,400 = −$25,200

Net: $95,850 + $25,000 − $25,200 = $95,650

Ninety-five thousand six hundred fifty dollars on a $47,500,000 project. About two-tenths of one percent. Plus 6,800 man-hours of overhead work performed standing on a floor instead of on a lift. Plus a corridor that is demonstrably better built.

And it consumed four months of concentrated attention from Grace Lindqvist, Sofia Marchetti, and Devlin Achebe, plus an executive sponsor and an owner willing to make her clinical departments answer a question early.

So the verdict: it paid, and it paid on schedule and safety rather than on cost. Anyone who tells you multi-trade prefabrication is a cost play has either not measured one or is selling you something. On the four failure modes from §39.1 it scores badly on adoption surface — it needed the owner, two competing subcontractors, and Kestrel's own VDC group to change behavior in a coordinated way — and it scores badly on value concentration, because the benefit split three ways while Kestrel paid the coordination. Those are commercial and organizational weaknesses, and they are precisely what an executive sponsor exists to solve.

What it scores perfectly on is work displacement. It removes work. It does not move a burden to a foreman so that somebody in an office can read a chart. It takes a third of a corridor's labor and performs it somewhere better. That is why it is the only thing in Chapter 39 that changes how the building actually gets built, and it is why Grace fought for it against a proposal that looked far better on a screen.

One number to carry forward. Northgate's prefabricated corridors ran at 3.90 site man-hours per linear foot against 9.66 stick-built. When Kestrel designed the pilot for its next multi-trade program, the pre-registered success criterion was 4.5 MH/LF — set deliberately above what Northgate achieved, because a criterion you set at your own best-ever result is not a fair test, it is a trap you built for yourself.

Discussion Questions

1. Ray told Pri Sethi the trade honestly: nineteen work days and 6,800 man-hours off lifts, in exchange for a design freeze she did not want. Rewrite that conversation for a design-bid-build public owner with no preconstruction relationship and a project manager who cannot bind the user departments. Is the program still available to you? What, specifically, is missing?
2. Devlin Achebe asked what Halcyon got out of spending three days a week of detailing in Cardinal's model. Ray answered by paying for the hours and giving Halcyon a bay. Name two other structures that could have solved the same misalignment, and say what each one costs Kestrel.
3. Problem 1 cost $6,800 because a systemic model defect was treated as a local correction. Design the specific procedural change — one sentence, in the BIM execution plan from Chapter 35 — that would have caught it. Then say what that change costs on a job where the assumption turns out to be local after all.
4. The program's net return was $95,650, of which $95,850 was schedule value that only exists because nine calendar days moved at substantial completion. Suppose Wei Chen's analysis had shown three calendar days instead of nine. Recompute the net. Would you run the program again? Would you tell your executive the truth about why?
5. Bea Salgado made her support conditional on reviewing Cardinal's shop safety program. Some contractors would consider that overreach into a subcontractor's business. Argue both sides, then say what you would actually do and what you would put in the subcontract.

Your Turn

Kestrel is offered a second multi-trade opportunity on the same job: the level 4 corridor spine, 460 LF, in a wing where the design is not frozen and Meridian's surgical services group is still deciding on two pieces of owner-furnished equipment.

Using only the Northgate rates established above — 9.66 MH/LF stick-built site labor, 3.90 MH/LF prefabricated site labor, 3.64 MH/LF shop labor, and the same roughly break-even direct cost — produce a one-page recommendation containing:

1. The site man-hours removed and the net labor hours saved, both computed and shown.
2. The one gate question from §39.9.1 that this opportunity fails, quoted, with your answer to it in a single sentence.
3. Your recommendation — proceed, decline, or proceed on a condition — in under 100 words, addressed to Nadia Haddad, who has ninety minutes and two other items on the agenda.

Then write the one sentence you would say to Pri Sethi if she asks you to do it anyway.