Case Study 29-2 — Four Months of Silence: The Delay Curtis Boone Never Saw

The projects, companies, and people in this case study are illustrative composites created for teaching. The arithmetic is real; the names are not.

Setup

Project: Rivermont Elementary School #12 — 78,000 SF, single-story with a two-story classroom wing, a commercial kitchen, and a regulation gymnasium on a 9.2-acre greenfield site at the edge of town. Owner: Rivermont Unified School District. Public. Prevailing wage, certified payroll, 100% payment and performance bonds. Delivery: hard-bid design-bid-build lump sum, $22,400,000, 610 calendar days, notice to proceed December 13, Year 1, substantial completion August 15, Year 3. Two summers, one school year, and an opening date that is not a date — it is a school bus schedule. Liquidated damages: $2,500 per calendar day. Contractor: Kestrel Construction Group. Project manager Curtis Boone.

Curtis's staffing, from his bid: a project manager at 50 percent, one superintendent, a part-time project accountant, and no project engineer. That is not carelessness. It is the arithmetic of how he won: Curtis shaved $465,000 out of his bid in the final review to beat five other contractors, and one of the lines he shaved was the field-office staff.

Curtis's general-conditions rates, and this matters at the end: he bid extended general conditions at $2,400.00 per calendar day. He actually burns $2,642.62 per calendar day.

Curtis's schedule. He submitted a compliant CPM baseline at notice to proceed. It was accepted. It is a genuinely competent schedule — Curtis has been doing this for nineteen years and he knows how a school goes together. The gymnasium is the critical path, and it is the critical path for a reason anyone who has built a school will recognize: the gym is CMU bearing walls carrying 96-foot long-span steel joists, and until those walls and bond beams are up, nothing about the gym roof, the gym overhead MEP, the gym ceiling, or the gym floor can start. Everything else on the job has float.

Between the baseline and the month-9 update, Curtis issued no schedule updates at all.

Not for months 5, 6, 7, or 8. The district's specification required monthly updates. Nobody enforced it, nobody withheld payment over it, and Curtis genuinely believed the baseline was still the plan.

What Happens

The slip nobody could see

Curtis bought the masonry package by taking a $1,684,000 number to a fourth bidder and getting $1,540,000. The subcontractor who took it was thin. The gym schedule needed nine masons and three tenders to hold the CMU wall production Curtis had modeled. From month 5 forward, the subcontractor manned it with four masons and two tenders, because that is who he had.

Production ran at roughly 60 percent of plan, continuously, for four months.

Here is why nobody caught it, and none of the reasons involve anybody being lazy.

• The daily reports recorded manpower correctly. Every day, by trade. Nobody ever totaled them, or compared the total to a plan, because nobody's job description said to.
• The gym looked like it was moving. It was. Every week there was more wall than the week before. Walking the job tells you the direction of travel and never tells you the rate.
• The baseline still said what it always said. With no update, the schedule was arithmetically incapable of reporting a problem. It was not wrong. It was mute.
• Curtis manages by walking and by relationship, and he is good at it. That method has a measurement blind spot exactly the width of a production rate, and it is invisible from the inside.
• A second problem arrived in month 7 and hid inside the first. RFI-062 asked how the gym bond beam was to be reinforced at the long-span joist bearings, where the drawings and the structural details disagreed. The architect's contractual review period was 14 calendar days. The answer came back in 24. Masonry stopped at the bond-beam course for ten of those days.

Nobody gave notice on RFI-062. Not because anybody decided not to — because the person whose job it would have been to notice that a 14-day clock had run to 24 did not exist.

The month-9 update, and why it happened

In month 9 the district's construction manager — a retired school-facilities director who had read the specification — withheld progress payment #9 pending a schedule update, as the contract allowed. Curtis had a consultant run one.

The gym chain came back at minus thirty-four calendar days of total float.

Reconstruction: what each update would have shown

After the job, Kestrel's project executive had the same consultant reconstruct the updates that were never run, from the daily reports, the delivery tickets, and the masonry subcontractor's own certified payrolls. Every input needed for these updates existed at the time. Nobody had to invent anything.

Read the float column down the page. It falls 9, then 11, then 10, then 9, then 6. That is not a random walk. That is a production system running at a stable, wrong rate, reporting itself faithfully, month after month, to nobody.

   Gym chain total float, calendar days

   +11 ██████████                         U-04   "two points behind — rain"
    +2 ██                                 U-05   ◄── THE SIGNAL. $14,000 fixes it.
    -9      ████████                      U-06   ◄── still cheap. $41,000.
   -19      ██████████████████            U-07
   -28      ███████████████████████████   U-08
   -34      █████████████████████████████ U-09   ◄── the first update anyone ran
        ────┼───────────────────────────
            0

Why month 9 was too late, specifically

Recovery was not impossible in month 9 because Curtis lacked money or will. It was impossible because two of the three things he would have needed had left.

At U-06, the gym chain was at −9 and every one of those three resources was still available and uncommitted. A recovery plan is a purchase, and what you are purchasing is other people's future capacity. That capacity is on sale early and off the market late. This is the mechanism behind an observation that sounds like a platitude until you have lived it: a schedule problem found early is a management problem; the same problem found late is an arithmetic problem with no solution in it.

The bill

Rivermont Elementary reached substantial completion 34 calendar days late.

Against an anticipated fee on this job of $784,000, thirty-four days of silence cost 21.9 percent of the fee.

And there is a second, quieter number in there. Curtis shaved his general-conditions line on bid day and carried extended GC at $2,400.00/CD in his bid, against an actual burn of $2,642.62/CD. Had he ever established entitlement to compensable time, his recoverable daily rate would have been argued from his own bid documents. The difference — $242.62 per calendar day — is money he could never recover from anyone, on any theory, because he priced it away himself in a conference room eighteen months earlier. Over 34 days that is $8,249.

What notice in month two of the slip would have been worth

The slip began in month 5. Month two of the slip is month 6, when the gym chain read −9.

The swing is roughly $154,000. The instrument that produces it is a monthly schedule update: two hours to collect actual starts, actual finishes, and remaining durations from the superintendent and the foremen; one hour to run it; one hour to write the narrative. Call it six hours a month.

Six hours a month for twenty months is 120 hours. At a fully burdened project-engineer rate of $62/hour that is $7,440 of labor. Curtis shaved the project engineer out of his bid to save somewhere around $110,000 across the job.

The part that does not have a number

In Chapter 33, Curtis sits in a conference room eleven months after substantial completion with a delay consultant who tells him: "I believe you. I can't prove you."

This case study is why. Of the five recognized methods of delay analysis, four require contemporaneous schedule updates. Time impact analysis needs the update that was in effect when the event occurred. Windows analysis needs a regular series of them. Collapsed as-built needs a defensible as-built with logic, which in practice comes from updates. Curtis has a baseline and an as-built date. He is left with the weakest method in the catalogue and no way to show that RFI-062 — a real, documented, ten-day breach of a contractual review period — ever touched the critical path.

The ten days were real. The entitlement was real. The proof never existed, and it stopped being creatable at the end of month seven.

Analysis

The failure is not that Curtis is careless. He is not. He is skilled, likable, experienced, and wrong about the model. He believes a schedule is a plan you make once and then execute by force of personality. That belief works — genuinely works — right up until the job contains a rate you cannot see by looking at it, and every job contains one.

Three specific mechanisms did the damage, and each has a cheap counter.

And the structural point underneath all three. Curtis did not decide to skip schedule updates. He decided, on bid day, to win the job by $465,000 — and one of the things inside that $465,000 was the person whose job the updates would have been. The staffing decision and the schedule-control failure are the same decision, separated by eighteen months and a bid opening. That is worth sitting with before you shave a field-office line to make a number work, and it is the honest reason this book keeps putting Curtis next to Ray: they are not a good contractor and a bad one. They are two coherent business models, and one of them has a hole in it that only shows up under load.

One more thing, in fairness to Curtis. Kestrel found the gym problem on Northgate's sister projects because Wei Chen exists, and Wei Chen exists because Northgate is a $47.5 million negotiated job that can carry a full-time project controls manager. A $22.4 million hard-bid school cannot. The answer is not "hire Wei Chen on every job." It is that the smaller the job, the more the schedule update has to be a simple, disciplined habit performed by whoever is there — the superintendent, the PM, a shared corporate scheduler for four hours a month — rather than a sophisticated deliverable produced by a specialist. A crude update run every month beats an elegant one run never, by about $154,000.

Discussion Questions

1. At U-05 the gym chain still had 2 days of float and the fix cost about $14,000. At U-06 it was at −9 and the fix cost about $41,000. Explain, mechanically, why the cost of recovery rises faster than the number of days lost. Use the three-resource table above in your answer.
2. Curtis's contract required monthly updates and the district did not enforce it until month 9. Whose failure is that, and what does an owner actually buy by enforcing a scheduling specification? Write the two-sentence policy you would give a school district's facilities director.
3. Reconstruct the notice Curtis should have sent in month 7 on RFI-062. Then list, specifically, the documents that notice would have needed to survive — and identify which of them existed on the job at the time and which did not.
4. Curtis carried extended general conditions at $2,400.00/CD in his bid and burned $2,642.62/CD. Explain to a young estimator why shaving the general-conditions line on bid day is a decision about claims recovery as well as a decision about price, and what you would do instead to find $150,000 on bid day.
5. Suppose you inherit this job at the beginning of month 8, at −28 days, with no updates in the file. What are the first three things you do, in order, on your first day — and which of them is aimed at the schedule, which at the record, and which at the people?

Your Turn

You are the project engineer Curtis did not hire, starting on Rivermont Elementary at the end of month 4, when the gym chain still has 11 days of float.

Produce the one-page monthly schedule report you would put on Curtis's desk on the fifth working day of every month. It must fit on one page and it must be producible in under two hours. Include:

• (a) The data date, and the projected substantial completion against the contract date.
• (b) A float trend table: three paths — make one of them the gym chain and one of them a procurement chain — with total float at each of the last four data dates and a change-per-month column.
• (c) A planned-versus-actual manpower table for the three trades on or near the controlling path, by area, with a variance column. This is the table that would have caught the masonry problem in month 5.
• (d) A ball-in-court list of open RFIs and submittals, sorted by days aging, with a column naming what each one gates and the date it is needed.
• (e) One paragraph — no more than six sentences — that names the single worst thing on the page, says what it costs at $5,142.62 per calendar day, states what you propose to change, and gives the date by which you will know whether it worked.

Then write the sentence at the top of the page that a busy, skeptical, nineteen-year veteran will actually read. You get one sentence. Make it a number and a consequence.