Case Study 29-1 — The $168,000 That Cost $773,900: Executing an Acceleration and Auditing It Honestly

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: Northgate Outpatient Pavilion — 132,000 SF, four stories, structural steel frame, outpatient clinics and an ambulatory surgery suite for Meridian Health System. Contractor: Kestrel Construction Group, CM at Risk, GMP $47,500,000, contract time 565 calendar days, notice to proceed March 3, Year 1, substantial completion September 18, Year 2, liquidated damages $5,500 per calendar day, extended general conditions contractually fixed at $5,150 per calendar day — a combined daily exposure of $10,650. Steel subcontractor: Ironbridge Steel (project manager Hank Duffy). Structural engineer: Caldwell Structural (principal Ruth Caldwell). Kestrel: Ray Alvarez (PM), Wei Chen (project controls), Margo Deacon (general superintendent), Bea Salgado (corporate safety director), Nadia Haddad (VP Operations), Dani Okonkwo (field engineer).

The event. Submittal 019 — anchor bolts and embeds — arrived from Ironbridge on March 27, Year 1. Kestrel's plan allowed five calendar days to review and transmit; it transmitted on April 7, eleven days later. Caldwell Structural took its full contractual fourteen calendar days — Kestrel's baseline had carried ten, based on a verbal remark nobody wrote down — and returned the package approved as noted on April 21. Ironbridge's reserved mill rolling slot required release by April 18. The next opening for those shapes was May 23.

Kestrel's own conduct is the largest single contributor and the delay is non-excusable. There is no claim here. There is only a decision, an execution, and a bill.

This case study is not about how the delay happened — Chapter 14 covers that. It is about what happened after Kestrel decided to spend money, and about what an honest audit of that decision looks like eighteen months later.

What Happens

Step 1 — The time impact analysis, run overnight

Wei Chen started from Update 01, data date March 31, Year 1 — the accepted update in effect when the event occurred. Not the baseline. Not the schedule as it stood on April 22.

The procurement chain in that update carried seven calendar days of total float, and Wei could name where those seven days lived: the gap between the planned approval date of April 11 and the required mill release date of April 18.

The float absorption analysis:

Wei then built a fragnet — four activities modeling the actual sequence that now had to happen: await next mill opening (32 CD) → roll and fabricate (46 CD) → galvanize and load (14 CD) → deliver by erection zone (9 CD) — inserted it into Update 01 with correct predecessors and successors, changed nothing else, and recalculated.

Step 2 — Four options, priced by cost slope

Overnight, Wei priced five crashable activities on the post-delay controlling path with Ironbridge and Kestrel's own crews, then crashed cheapest-first, recalculating after every step.

L stopped at 4 of its available 6 because the MEP branch — which had carried four days of float — became co-critical at that point. After that, only compression upstream of the split bought anything, which is why K was next despite being more expensive per day. The eighteenth day would have required compressing both branches at a combined $26,000/CD against a day worth $10,650. That is the stopping rule, and it produced 17 days, not a negotiated number.

Direct crash cost $118,000, plus $50,000 of enabling cost — extended tower-crane and hoist hours, a working foreman and Saturday superintendent coverage, extended temporary lighting and services — for the canonical $168,000.

Step 3 — Thursday morning, April 23

Nadia: "B, C, and D are inside five thousand dollars of each other. C is the cheapest and it's the one you're not recommending. Explain."

Ray: "C gets us done September thirtieth. The lease expires October first."

Nadia: "One day."

Margo: "One day isn't margin. In September of next year we'll have a fire marshal, an elevator inspector, and a commissioning agent all standing between us and that date, and any one of them can take a week."

Nadia: "Take D. Ray, write the reason on the same page as the number. In eighteen months somebody is going to look at this and see we paid twenty-one hundred dollars more than the cheapest option, and I want them to find the sentence, not have to guess at it."

Kestrel accelerated. Ray's sentence, on the decision memo, was one line: "Option D is $2,100 above Option C and delivers six additional days of margin against Meridian's October 1 clinic lease expiry, which has patients scheduled behind it."

Step 4 — Execution

The acceleration ran from August 27 through late December, Year 1. It worked. Ironbridge put a second raising gang on sequences 2 through 4, ran six-day weeks for nine weeks, and topped the building out. The deck crew ran two pumps and Saturdays. The enclosure was released by elevation instead of by level, so the north and east faces could start while the deck-edge work on the south was still running.

That last change was the free one. It had no line on the crash table.

It put the deck-edge detail crew, the precast erector, the curtain-wall crew, and the fireproofing crew on the same elevations in the same weeks. Four trades in a space the plan had sized for two.

Week 34, a Tuesday, north elevation, level 3, 7:20 a.m. A frame scaffold had been partially modified overnight by a different trade to run conduit; a plank was lifted and not re-secured. Milo Serrano, a mason tender, stepped onto it. The plank shifted. 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.

Her investigation found three failures:

1. No competent-person inspection tag for that shift — the tag was two days stale.
2. A scaffold modified by a trade that did not erect it, with no re-inspection.
3. A crew running behind after the steel acceleration, with an unwritten "make it up" pressure.

Near-miss reports across the job spiked in weeks 34 through 36 and then fell back to baseline.

Analysis

The honest audit, eighteen months later

At closeout, Ray and Wei ran the acceleration through a lessons-learned accounting. Not the decision page — the whole bill.

Compare that to Option A — do nothing, absorb 23 days — at $244,950.

Fully accounted, the acceleration cost $528,950 more than not accelerating.

Three things about that number, in order of importance.

First, the apportionment is a judgment and it is stated as one. The curtain-wall package forecast $820,426 over budget. Wei attributed roughly $430,000 of it to deck-edge geometry and field modification, and the balance to a supplier problem and an owner-requested corner-detail change that had nothing to do with acceleration. Reasonable people could argue that split. What is not arguable is the sign. Everybody on that job could have predicted, in April, that compressing deck-edge detailing and stacking four trades on one elevation would produce rework. Nobody wrote a number next to it, and a consequence with no number next to it loses every argument to a column of dollars.

Second, Kestrel would make the decision again. The alternative was Meridian holding clinics in a building that did not exist, with a lease that had expired and patients on a schedule. Kestrel's $528,950 bought Meridian something Kestrel's cost report cannot price. That is a legitimate reason to spend money and it is why the decision was Nadia's rather than the spreadsheet's.

Third — and this is the finding that matters — Kestrel would not execute it the same way.

The cheapest day on the crash table was the most expensive day on the job

Look at activity L: deck edge, embeds, stair and hoistway steel — released by elevation instead of by level. Cost slope $7,000 per calendar day, the second cheapest on the page. It bought 4 days for $28,000.

Those four days produced $64,000 of deck-edge rework and roughly $430,000 of curtain-wall field modification.

   L — compress deck-edge detailing and release enclosure by elevation

      What the crash table said it cost         $ 28,000   for 4 CD
      What it actually cost
          deck-edge rework          $ 64,000
          curtain-wall shimming
            and field modification  $430,000
                                    ────────
                                    $494,000
      Ratio of realized cost to priced cost:    17.6 to 1

The cost slope prices the compression. It does not price the consequence. L's compression touched the one interface in the building where two independently fabricated systems have to meet at a tolerance — a unitized curtain-wall panel landing on a cast-in embed — and it compressed the detailing and layout half of that interface, which is precisely the half where an error stays invisible for five months and then appears 38,500 square feet at a time.

The right execution was available and cost almost nothing to find: take the 4 days out of activity M instead. M — precast and curtain-wall installation — had 14 days of compression available at $15,000/CD. Four days there would have cost $60,000 instead of $28,000, a $32,000 premium, and it compresses installation rather than detailing, which is a far more forgiving place to lose tolerance.

$32,000 against $494,000. That is the whole lesson, and it is not a lesson about scheduling. It is a lesson about which activities you are allowed to squeeze.

Rule: never compress the layout, detailing, or interface-coordination half of a two-system interface. Compress installation instead, even when the cost slope says otherwise. The crash table cannot see tolerance.

What the schedule got right

It is worth being fair to the instrument. The TIA was correct: 23 days, computed in one night, from a contemporaneous update, using a method any consultant on either side of a dispute would recognize. The crashing analysis was correct: 17 days at $118,000 of direct cost, with a stopping rule derived from arithmetic rather than negotiation. The forecast was correct: the accelerated package landed the projected completion six days past the contract date, which put occupancy well ahead of Meridian's lease expiry — and the residual six days were themselves recovered in closeout, so substantial completion was achieved on September 18, Year 2.

Everything that failed, failed outside the schedule. The schedule told the truth about time and was silent about tolerance, trade density, and fatigue — and silence, on a decision page, reads as zero.

What changed at Kestrel

Three practices, adopted after this job.

1. A safety impact review is a required attachment to any acceleration decision, performed by the person who owns safety, before the decision. Six questions. The sixth is what is the stop condition — the threshold at which the acceleration is suspended — because Northgate's plan had a start date and no end.
2. Any activity whose downstream chain terminates at an external constraint you cannot buy back — a mill slot, a permit, a utility cut-in, an owner-furnished delivery, a seasonal restriction — is managed as critical regardless of its float.
3. Interface activities are flagged in the schedule and are not crashable without a written quality review. Wei added an attribute to the P6 activity codes for it. It took an afternoon.

Discussion Questions

1. Wei's apportionment assigned $430,000 of an $820,426 curtain-wall overrun to the acceleration. Build the argument for a smaller number and the argument for a larger one. What evidence would settle it, and does that evidence exist eighteen months later?
2. Option C was $2,100 cheaper than Option D and finished one day before the lease expired. Kestrel chose D. Construct the memo you would write if the board later asked why the company spent $2,100 more than the cheapest available option — and then construct the memo you would have had to write if Kestrel had chosen C and finished on October 2.
3. The "free" lever — resequencing the enclosure by elevation — turned out to be the most expensive item on the page. Does that mean resequencing is a bad tool? What would you change about how a resequencing proposal gets reviewed, without giving up a genuinely no-cost compression method?
4. Finding 3 of the scaffold investigation — "a crew running behind, with an unwritten 'make it up' pressure" — has no author. Nobody instructed anybody to skip a scaffold tag. Explain the mechanism by which an office decision in April produces a lifted plank in week 34, and name three specific controls that interrupt it.
5. Kestrel's delay was non-excusable, so there was no claim and no owner contribution. How would every number in this case study change if the 23 days had been owner-caused — and which of Kestrel's decisions would you expect to change as a result?

Your Turn

You are Wei Chen on the morning of April 23, Year 1, and Nadia has asked for one thing before she signs: a one-page safety and quality impact review of Option D, attached to the decision memo.

Produce it. Six sections, one page:

1. New hazards created by the second raising gang, the sixth day, and the second swing stage.
2. Trade stacking — which trades, which elevations, how many weeks, and the peak number of trades per elevation against the site logistics plan.
3. Overtime exposure — how many consecutive weeks, and what ends it. Write the end condition as a date or a measurable trigger, not as "when we catch up."
4. Interfaces at risk — name the two activities in Option D where compression touches a tolerance between two independently fabricated systems, and state what you would do differently.
5. Added controls — the inspection or hold points that change, with who performs them.
6. The stop condition — the specific threshold at which the acceleration is suspended, and who has the authority to invoke it without asking the project manager first.

Then write the single sentence that goes at the bottom of the decision page, next to the $168,000, summarizing what this acceleration costs that is not denominated in dollars. One sentence. If it takes two, you have not finished thinking.