Case Study 10-2 — Six Weeks Saved: What Skipping Coordination Cost Curtis Boone

All people, companies, and projects in this case are Tier-3 illustrative composites.

Setup

Project: Rivermont Elementary School #12. $22,400,000, hard-bid design-bid-build lump sum, public owner, prevailing wage, 100% payment and performance bonds. 540 calendar days. Liquidated damages $2,750 per calendar day; Kestrel's extended general conditions on this job run $3,100 per calendar day — a combined daily exposure of $5,850/CD.

The building: 78,000 SF, two stories, masonry and steel, with a gymnasium, a cafeteria and full commercial kitchen, a media center, twenty-eight classrooms, and a music suite. MEP is roughly 34% of the direct cost of work — lower than a healthcare building, but the kitchen, the gym, and the music suite each carry a concentrated coordination problem.

The project manager: Curtis Boone. Fifteen years with Kestrel. He wins work by buying it low and running lean. He is a genuinely good builder, he is well liked by his subs, and his field instincts are better than mine were at his age. He is also wrong about one specific thing, and this case is about that thing.

The decision. The specification requires "coordination drawings" for MEP. The word model does not appear in it. Curtis reads that as permitting the older method: each trade produces a 2D overlay of its own work on a common background, the drawings are stacked on a light table or layered in a PDF, and conflicts are resolved by discussion. He does not run a federated 3D model, does not hire a VDC coordinator, and does not schedule clash rounds.

His stated reasoning, in a preconstruction meeting I attended:

"Full BIM coordination on a school is a luxury. It's six weeks of preconstruction and about seventy-eight thousand dollars in detailing I don't have in this number. We hard-bid this job. My subs have all built schools. They'll work it out in the field like they always have."

Cost avoided: $78,000. Preconstruction time avoided: 6 weeks. Those are the two numbers he reported, and both are true.

What Happens

Months 1–4: nothing goes wrong

This is the part that makes the mistake so durable. For four months, Curtis looks right. The overlays get produced, three coordination meetings happen, a handful of obvious conflicts get resolved, and the job proceeds. Rough-in starts on the second floor classrooms — repetitive, simple, generous ceiling space — and it goes fine.

Repetitive classroom ceilings are the easiest MEP condition in commercial construction. They are also the last place a coordination problem will reveal itself.

Month 5: the RFI curve turns

The trades reach the gym, the kitchen, the media center, and the main corridors — the four locations where systems concentrate — and the RFI log changes shape.

Two hundred fourteen MEP RFIs in six months against Northgate's sixty-one over a comparable phase. And the content is the tell. Northgate's MEP RFIs are mostly genuine design questions: a specification conflict, a missing detail, an equipment substitution. Rivermont's are overwhelmingly of one type:

RFI 148: The 48"×14" supply main at grid F/7 conflicts with the 6" sprinkler main and the structural bracing. Please advise routing.

That is not a design question. That is a coordination question being routed through the architect because nobody resolved it in a model. The architect, who is not the coordinator and does not have a 3D model either, answers it in an average of 19 calendar days — slower than the contractual 10 working days, because the volume has buried him.

Months 5–9: what the field actually did

The gym. The gym has a long-span joist structure and a high ceiling. The supply mains, a sprinkler grid over a two-story volume, and gym lighting with fixture guards all converge under the low chord. Nobody had checked the clear height under the deepest joist against the required ceiling. The resolution: lower the gym ceiling 6 inches across the whole space, revise the lighting layout, re-detail the sprinkler drops, and re-do 2,100 SF of soffit at the perimeter.

The kitchen. The commercial hood's grease exhaust duct has code-driven clearance-to-combustibles requirements and cannot be routed casually. It was drawn on the mechanical plan as a single line. In the field it would not clear the makeup-air unit's supply duct or the structural framing. Resolution: a dropped soffit through the kitchen and a relocated makeup-air unit, which required a new roof curb, new structural framing, and a roofing revision — on a roof that was already installed.

The corridors. The main first-floor corridor carries the feeder to the second-floor electrical room in cable tray, plus the gym and kitchen ducts, plus the sprinkler main. 1,900 linear feet of duct was hung, discovered to conflict, cut down, and re-hung.

The inspections. Three above-ceiling rough-in inspections were failed in five months: one for duct hung from the metal deck rather than from the structure, one for firestop that did not match a tested system at the corridor rated wall, and one for missing seismic bracing on a re-hung main where the original delegated bracing calculation no longer applied to the new route. Each failure cost the re-inspection slot — four to five working days each — plus the corrective work.

Month 11: the damage, totaled

Every dollar Curtis "saved" cost $8.21. ($640,100 ÷ $78,000 = 8.21.)

The schedule outcome: 41 calendar days of slip, of which 19 were bought back through acceleration and 22 were absorbed. The school opened, but the district moved its teacher-orientation week and ran two weeks of the school year out of a neighboring building.

Month 14: the claim that failed

Curtis filed a delay claim for the full 41 days, on the theory that the architect's 19-day average RFI response time — against a contractual 10 working days — caused the delay.

The theory is not crazy. The architect was late. But the claim failed on causation, and here is the mechanism.

The owner's consultant reviewed all 214 MEP RFIs and categorized them. 156 of the 214 — 72.9% — were coordination questions that a federated model would have resolved without ever generating an RFI. Only 58 were genuine design questions. The consultant's argument was straightforward: the architect's response time was slow because Kestrel generated four times the normal RFI volume by declining to coordinate, and a contractor cannot recover for a bottleneck it manufactured.

Curtis could not rebut it, because he had no way to separate his coordination-caused RFIs from design-caused ones. He had never categorized the log. He had never tracked which RFIs were on the critical path. He had no contemporaneous schedule updates showing the delay's progression — the CPM was updated twice in fourteen months, both times after the fact.

Settlement: 9 calendar days of non-compensable time extension. $0 in compensable delay. The remaining 32 days stayed on Kestrel's books.

Nadia Haddad, Kestrel's VP of Operations, wrote one sentence in the lessons-learned file that Ray has since put on the wall of every job trailer he runs:

"We were right that the architect was slow. We could not prove that it mattered, because we had made ourselves the reason he was slow."

Analysis

The mechanism, not the moral. Curtis is not lazy and he is not stupid. His error is a specific, learnable one: he priced coordination as a cost and not as a schedule input. In his estimate, $78,000 of detailing was a line item competing with other line items. In reality, coordination is the predecessor of fabrication, which is the predecessor of rough-in, which is the predecessor of two inspection hold points that gate every finish trade in the building. Deleting a predecessor does not delete its duration; it relocates the duration into the field, where it is performed at field rates, by people on lifts, in the presence of an inspector.

Why classrooms hid the problem for four months. Repetitive, simple, generously spaced ceilings are the easiest MEP condition there is. The gym, kitchen, media center, and corridors are the hard ones — and they are always later in the sequence, because they are bigger, more complex, and more dependent on equipment submittals. A coordination method validates itself on the easy work and fails on the hard work. That is the trap, and it is why the absence of early problems is not evidence.

The 2D overlay method is not fake — it is just not sufficient here. Layered 2D drawings worked for decades and still work on simple buildings with generous cavities. What they cannot do is resolve the third dimension. A 2D overlay shows that a duct and a pipe occupy the same plan location; it cannot tell you whether they occupy the same elevation, and in a ceiling cavity elevation is the whole question. On a warehouse with a 6-foot cavity that does not matter. On a school corridor with 26 inches, it is the only thing that matters. The method is not the error. Choosing the method without running the cavity arithmetic is the error.

The claim failure is the most expensive lesson in the case. Curtis lost $562,100 on the coordination decision. He then lost the ability to recover any part of it because his own conduct destroyed causation. This is the direct connection to Chapter 33: entitlement, causation, and damages are three separate proofs, and you must win all three. He arguably had entitlement — the architect was late against the contract. He had damages. He could not prove causation, and two out of three is zero.

What Northgate did differently, and what it cost. Kestrel spent roughly $210,000 on Northgate's VDC coordination — Grace's time, the trades' detailing hours, the software, and the meeting time across a 12-week window. It resolved several thousand conflicts in a model. Using the conservative $2,000-per-field-conflict figure from Chapter 10 §10.7, the avoided cost is measured in millions. That is not a marketing claim; it is the same arithmetic Curtis ran, with the sign corrected.

Discussion Questions

1. Curtis's specification required "coordination drawings" and did not say "model." He complied with the letter of the specification. Was he wrong? Separate two questions in your answer: whether he breached the contract, and whether he made a bad management decision. Can the answers differ?
2. Run the cavity arithmetic Curtis never ran. For a school corridor with 12'-0" floor-to-floor, an 18-inch structural depth, and a 9'-0" required ceiling, compute the available cavity. Then list what would have to fit, and state at what point you would insist on a 3D model regardless of what the specification says.
3. The owner's consultant categorized 156 of 214 RFIs as coordination-caused. Suppose you are Curtis's attorney. What contemporaneous records, if they existed, would have let you rebut that categorization — and at what point in the project would you have had to start keeping them?
4. Kestrel spent $164,000 to accelerate and recover 19 days, leaving 22 days at $5,850/CD = $128,700. Do the arithmetic on whether the acceleration was worth it in pure dollars, then name two non-monetary factors that could make the answer different. (Compare the reasoning Kestrel used on the Northgate steel delay.)
5. Curtis is likable, skilled, and well regarded by his subcontractors. His crews would follow him anywhere. How does a construction company prevent a manager like this from making this decision again — without turning coordination into a checkbox that gets performed badly?

Your Turn

You are the incoming project manager on a $14,000,000 municipal library with a 20-inch typical ceiling cavity, a two-story reading room, a café, and a small commercial kitchen. Your estimate has $0 in it for MEP coordination and your specification, like Curtis's, says only "coordination drawings."

Write a one-page memo to your VP of Operations requesting a coordination budget. It must contain:

• The cavity arithmetic for your worst location, showing available versus required.
• The three locations you predict will fail, and why.
• A dollar request with a build-up (detailing hours, coordination management, software, meeting time).
• The cost of not doing it, expressed both in field rework at the order-of-magnitude framing and in days at your daily exposure rate — pick a realistic LD rate for a $14M municipal job and state it.
• One sentence on what you will do if the answer is no.

That last bullet is not a throwaway. Half of management is knowing what your fallback is before you ask for something.