Case Study 20-1 — Ninety-Two and Sixty-One

How Kestrel's concrete crew lost a third of its output in forty-eight hours, and how five hours of paperwork found out why

Setup

Project: Northgate Outpatient Pavilion — 132,000 SF, four stories, CM at Risk with a $47,500,000 guaranteed maximum price. Owner: Meridian Health System.

Week: 18. Foundations are in their last three weeks; contract milestone for foundations complete is July 3, Year 1. Kestrel self-performs the concrete.

The people:

• Jamal Foster — self-perform concrete superintendent. Twenty-six years placing concrete.
• Ray Alvarez — senior project manager, narrating.
• Dani Okonkwo — field engineer, four months out of school, about to do their first productivity investigation.
• Margo Deacon — general superintendent.
• Lorena Vasquez — project accountant, who owns the cost codes.

The activity: cost code 033000, structural concrete placement — footings, foundation walls, and grade beams. Canonical Northgate quantities: 148 footings / 1,240 CY, plus 620 CY of foundation walls and grade beams.

The budgeted unit rate: 1.15 man-hours per cubic yard, placed and finished. This is not a published figure. It is Kestrel's own number, built from nine years of cost history on similar work, and Tomás Reyes reviews it every January.

The crew: ten people. One foreman, four journeyman carpenters and cement masons, two apprentices, three laborers. A pump and an operator on a day rate.

The problem, on two pieces of paper:

Same crew. Same mix design. Same supplier. Same pump, same operator. Fifty-eight degrees and dry both days.

At $54.00 an hour burdened, Thursday earned 61 × 1.15 = 70 MH and spent 100. Thirty man-hours of variance — $1,612 — in a single day.

What Happens

I gave the file to Dani on Friday morning with one instruction: do not talk to the crew yet. Read the paper first, form your candidates, then go ask. Ask first and you get the first plausible story, and the first plausible story on a construction site is almost always "the crew was slow" — which is not a cause, it is a restatement of the number.

Dani came back at noon with five candidates. This is the right way to do it: name every cause the data could support, then eliminate.

Eliminating four of them

Candidate 1 — the crew. Sign-in sheets: same ten names both days. Time cards: same classifications, all ten hours coded to 033000 on both days, no split coding. Eliminated in four minutes.

Candidate 2 — concrete supply. The batch tickets are the single most underused document on a concrete job. Every load carries arrival time, discharge start, discharge complete, and quantity. Dani built this:

Interesting — the trucks were waiting longer on Thursday. But note the direction. Trucks waiting on a crew is a symptom of a slow crew, not a cause of one. If the supplier had been late, the crew would have been waiting on trucks. The wait ran the wrong way to be the cause. Eliminated, and upgraded to evidence for something else.

Candidate 3 — the pump. The operator's log showed zero downtime, zero blockages, one line washout Tuesday and two Thursday, and one setup Tuesday against two Thursday, with a relocation between them logged at 10:40 a.m. Also not a cause — but the second setup is now the loudest thing in the file.

Candidate 4 — the material. Field test reports: average slump 4.75 inches Tuesday, 4.5 inches Thursday, both inside the specified range; concrete temperature 62°F and 61°F. Same mix design, same plant. Eliminated.

Isolating candidate 5

Dani laid the pour plan next to the daily quantity report and found the "Notes" block Jamal's foreman had written at the bottom:

 Notes: Pump relocated to NW footing group at 10:40 -- east ramp blocked,
        backfill not complete. Crew split 5/5 for balance of pour.
        Second washout required. Lost approx. 1.5 hr of placement time.

The pour plan for Thursday had a single pump setup on the east side, reaching the east grade beams and then swinging north to a group of eleven footings at the northwest corner. That reach depended on the pump being able to drive up the east haul ramp. The excavation subcontractor had told us the ramp backfill would be complete Wednesday. It was not.

So the crew poured the east grade beams from the planned setup, tore down, moved the pump around the building, set up again, and poured the northwest footings from a second position — with the crew split into two groups of five for the second half of the day so that one group could strip and clean the completed east work while the other placed.

The arithmetic that closes the case

This is the part I want you to be able to do, because a diagnosis without arithmetic is an opinion.

Step 1 — separate lost time from lost rate.

Step 2 — decompose the 31-cubic-yard gap.

Loss from lost placement time = (8.0 − 6.4) hr × 11.5 CY/hr = 18.4 CY Loss from the reduced rate = 6.4 hr × (11.5 − 9.5) CY/hr = 12.8 CY Total explained = 31.2 CY

Actual gap: 92 − 61 = 31 CY. The two mechanisms account for the entire loss.

Step 3 — what each mechanism actually is.

• The 18.4 CY of lost time is the pump relocation: setup, teardown, travel, re-staging, second washout. It is pure access.
• The 12.8 CY of lost rate is the split crew. Jamal said it in one sentence at the gang box: ten people on one pour is a machine — two hoses, two vibrators, a screed gang, a finisher, a truck man, a cleanup man, everybody inside forty feet of everybody else. Split it into two fives and you get two half-machines, each of which still needs a full set of tools, a full set of eyes, and its own start-up and clean-up. That is the crew-size curve from §20.8 running backwards: below the optimum, you lose specialization and rhythm faster than you lose headcount.

Step 4 — the money, including the part nobody counts.

And the exposure if nothing changes: this was not a freak event. It was the third time in six weeks that a pour plan had assumed an access condition nobody had confirmed. There were roughly eighteen pour days remaining on the structural concrete and about 1,800 cubic yards still to place across grade beams, slab on grade, elevated deck topping, and pads.

Analysis

What failed was not the crew and not the pour. It was the constraint check.

The pour plan was a good plan. It became a bad plan when a condition it depended on — a backfilled haul ramp — did not exist, and nobody checked. That check would have taken one person four minutes on Wednesday afternoon.

Three mechanisms are worth extracting.

1. The loss was created by a different trade, on a different part of the site, on a different day. The excavation subcontractor did not do anything wrong that anyone would recognize as wrong; they were a day behind on a backfill nobody had told them was a predecessor to a concrete pour. It was not on their look-ahead as a constraint because nobody had put it there. This is theme 6 in a very concrete form: the work is coupled even when the contracts are not.

2. The diagnostic path ran entirely through documents that already existed. Sign-in sheets, time cards, batch tickets, the pump operator's log, field test reports, the pour plan, and the daily quantity report. Kestrel did not create a single new form to solve this. What it created was the habit of reading them together. If any one of those documents had been missing — and the one that most often is missing is the quantity report with a working Notes field — the investigation stalls at "the crew was slow."

3. The cost code made it visible and the note made it explainable. A cost code with a unit of measure attached gives you a productivity factor. A note field gives you a cause. You need both, and most jobs have neither. See Chapter 28 for the cost-code discipline that makes this possible.

The fix

Five changes, agreed in a twenty-minute meeting between Jamal, Margo, and me.

1. A pour-day readiness checklist, signed off by the superintendent the afternoon before — access route confirmed by walking it, pump position confirmed with reach checked against the actual setup point, formwork inspected, rebar inspected and released, embeds verified, washout available, crew and pump confirmed, weather checked. Constraints get removed the day before or the pour moves. This is the constraint-log discipline from Chapter 27, applied to one operation.
2. Backfill and haul-road work added to the six-week look-ahead as a predecessor to every affected pour, with the excavation subcontractor's foreman in the Tuesday coordination meeting when pours are discussed.
3. Pump positions drawn on the pour plan with reach circles, and an alternate position drawn for every pour. If the alternate is needed, it is a decision, not a scramble.
4. A crew-splitting rule: no split pour with fewer than eight people per position unless the superintendent writes down why. The rule is not "never split." The rule is "splitting is a priced decision, so price it."
5. The Notes field on the daily quantity report is mandatory, and I read it the next morning and ask about anything in it. That last clause is the whole rule. A report nobody reads becomes fiction inside a month.

What it was worth

Over the balance of the structural concrete — roughly 1,800 CY across grade beams, slab, topping, and pads — Kestrel's concrete cost codes moved from a year-to-date productivity factor of 0.94 at the time of the investigation to 1.02 at completion.

Budget MH at a blended 1.05 MH/CY on 1,800 CY = 1,890 MH At PF 0.94: 1,890 ÷ 0.94 = 2,011 MH At PF 1.02: 1,890 ÷ 1.02 = 1,853 MH Saving: 158 MH × $54.00 = $8,532

Plus roughly one avoided make-up pour a month at about $4,300 each. Call the whole thing $20,000 to $22,000 on a job whose projected net profit is under a million dollars — for a checklist and a habit.

But the number is not the point. The point is that Dani now knows how to do this, and the readiness checklist got copied onto the deck pours, the curtain wall installation, and eventually the MEP rough-in. The tool generalized. The savings did not stay in the concrete code.

Discussion Questions

1. Dani was told not to talk to the crew until the documents had been read. Make the strongest argument against that instruction — when would talking to the crew first be the better move, and what would you have to do to keep the first conversation from anchoring the whole investigation?

2. The truck-wait data (6 minutes Tuesday, 21 minutes Thursday) looked at first like evidence of a supply problem and turned out to be evidence of a crew running slow. Describe another construction metric where the causal direction is easy to get backwards, and how you would test which way it runs.

3. The excavation subcontractor was a day late on a backfill that nobody had identified as a predecessor to a concrete pour. Whose failure is that, and what does your answer imply about how the six-week look-ahead should be built? (Compare Chapter 19 on coordination structure and Chapter 27 on constraint removal.)

4. The crew-splitting rule says "write down why," not "don't." Why is that the better rule? What would go wrong with a flat prohibition, and what would go wrong with no rule at all?

5. Suppose the concrete had been placed by a subcontractor rather than self-performed. Which of the seven documents Dani used would you still have had access to, which would you have had to ask for, and which would you probably never have seen? What does that tell you about what to require in a subcontract?

Your Turn

You are the project engineer. Your masonry subcontractor's cost code for CMU at the loading dock shows the following at the end of week 12:

Do four things.

1. Compute the productivity factor and forecast the man-hours at completion.
2. Write down five candidate causes, in the order you would investigate them, using the structure in the table above — one column for what would confirm each, one for what would eliminate it.
3. For each candidate, name the specific document or observation you would ask for. Mark which of them you have a contractual right to and which you would have to ask for as a favor.
4. Write the three-sentence email you would send to the masonry subcontractor's project manager. It should ask for data, not blame — and it should be an email you would be comfortable seeing attached to a claim eighteen months from now.