Case Study 36-2 — Certified and Underperforming

The Cypress Grove Municipal Services Building, City of Rivermont. Year one of occupancy.

All people, companies, and projects in this book are illustrative composites.

Setup

Kestrel Construction Group built the Cypress Grove Municipal Services Building for the City of Rivermont two years before Northgate broke ground: 78,000 SF, three stories, permit counter, city offices, three community meeting rooms, a public-works dispatch center, and a council chamber. It certified at a high level under its rating system. There is a plaque in the lobby. The mayor stood in front of it.

The design team's energy model predicted an energy use intensity (EUI) of 38 kBtu per square foot per year. The City built its operating budget on that number and budgeted $148,000 per year for utilities.

Year one came in at 71 kBtu/SF·yr and $271,000 — an $123,000 overrun, and an 87% miss against the model.

The City's facilities director called the architect. The architect called the mechanical engineer. The mechanical engineer said the building was built wrong. Kestrel got a letter.

Nadia Haddad, Kestrel's VP of Operations, made the decision I want you to notice: rather than answering the letter with a letter, she offered to split the cost of a retro-commissioning investigation with the City and to give them Ray Alvarez for two weeks. The City engaged Amara Boateng.

What happens

The wrong first question

The first meeting opened the way these meetings always open.

Facilities director: "The building doesn't perform. It's certified. How is that possible?"

Amara Boateng: "Those two sentences are less related than everybody thinks. Certification means the project documented a defined set of design and construction requirements. It does not mean the finished building will use a predicted quantity of energy, and no rating system in common use promises that. So let's stop asking how it's possible, because it's entirely possible, and start asking where the thirty-three points went."

She then said the sentence that reframed the whole engagement:

Amara: "One more thing, and it matters more than anything else I'll say today. The energy model was a compliance model, not a prediction. It was built to demonstrate that this design uses less energy than a code-defined reference building, using standardized schedules and standardized assumptions. That is what the performance compliance path requires and that is what it did, correctly. Nobody on that design team ever promised you 38. Somebody on your side wrote 38 into a budget."

The diagnosis

Eleven weeks: trend data from the building automation system, a functional retest of a sample of terminal units, meter readings, interviews with two facilities technicians, and the original commissioning record.

Bucket 1 — design and program assumptions.

The compliance model assumed a 55-hour operating week. Cypress Grove actually runs 96 hours a week: an evening permit counter two nights, community meeting rooms booked most evenings, weekend recreation programming, and a dispatch center that never closes. None of that was hidden — it was simply not what a standardized compliance schedule looks like.

Worse, a 24/7 dispatch and IT equipment room was added to the program after the model was run, during design development, and the model was never rerun. That room is a continuous cooling load in a building otherwise designed to coast at night.

Bucket 2 — construction execution.

Two real defects, both invisible without instrumentation.

• Fourteen of forty-two variable-air-volume terminal units had minimum airflow setpoints left at the balancing contractor's default value rather than the design values on the schedule. Result: those zones over-ventilate continuously and then reheat the excess air. The building heats and cools the same air at the same time, every hour it runs.
• Two economizer damper linkages were installed reversed. The units bring in outside air when they should recirculate and recirculate when free cooling is available. Both units were started up in a season in which the error would not have shown, and nobody exercised them across a seasonal transition.

The envelope, notably, was not a problem. Cypress Grove passed its whole-building air-leakage test on the first attempt, and infrared spot checks in year one found nothing significant. Kestrel had learned that lesson on a previous job and it held here.

Bucket 3 — commissioning scope.

The project bought fundamental commissioning only. The scope covered functional testing of equipment at start-up. It did not cover:

• control sequences tested across seasonal modes,
• verification of terminal-unit setpoints against the design schedule,
• a post-occupancy review at ten months,
• any measurement and verification of actual consumption.

The building had meters. Nobody read them for fourteen months.

Bucket 4 — operations and occupant behavior.

• Corridor lighting occupancy sensors were disabled by facilities staff after evening cleaning crews complained about lights going out on them.
• Night setback was turned off three weeks after occupancy, following a single Monday-morning complaint that the building was cold at 7:30 a.m. It was never turned back on.
• Space temperature setpoints were tightened from a 70–75°F band to 71–73°F in response to comfort complaints — complaints that were themselves largely produced by the over-ventilating VAV boxes in Bucket 2.
• Roughly a dozen personal space heaters were in use, plugged in under desks in the same zones.

Attributing the gap

Amara's estimated attribution of the 33-point gap:

The correction, and what it cost

Kestrel absorbed the two construction items — the VAV setpoints and the economizer linkages — as warranty work, roughly $22,000, without arguing about it. Ray's view, and Nadia agreed: they were defects, they were Kestrel's, and the cost of arguing exceeded the cost of fixing.

Year two came in at 55 kBtu/SF·yr and $212,000 — sixteen points recovered. The arithmetic predicted about thirteen and a half (the 41% of the gap that was actually fixable); the extra two and a half points came from staff retraining and optimized start-stop tuning that Amara threw in.

Then Amara did the piece of work that mattered most, and it was arithmetic, not engineering. She rebuilt the model with the actual operating hours and the actual program, and produced a corrected expected EUI of about 56.

Corrected expectation: 56. Year-two actual: 55.

Once the model was told the truth about how the building is actually used, the gap essentially disappeared — and the building came in just under the corrected expectation.

Analysis

The headline finding is that the biggest single cause was not construction. Fifty-five percent of the gap came from the model being asked a different question than the one the City thought it was answering. A compliance model demonstrates that a design outperforms a reference building under standardized assumptions. It is a comparison, not a forecast. Treating a code compliance model's EUI as an operating budget is a category error, and it is an extremely common one — made by owners, by facilities directors, and sometimes by the people who write the press release.

But twenty-eight percent was construction, and that part is genuinely ours. Fourteen terminal units left at a balancing default and two reversed damper linkages are ordinary, findable, unglamorous defects. They passed every inspection anyone ran, because nobody ran the one inspection that would have found them: verify installed control setpoints against the design schedule, and exercise the sequences across seasonal modes. That is not a mysterious capability. It is a line item in an enhanced commissioning scope.

The commissioning scope is the cross-cutting cause, and this is the lesson to carry. Look at the table again and notice which findings an enhanced commissioning scope would have caught:

Enhanced commissioning would have caught findings accounting for 41% of the gap — which is to say essentially all of the fixable portion of it. The incremental cost of that scope on a building this size is a fraction of the $123,000 the City overspent in year one alone — and year one is not the only year. The building will run for forty.

The distinction to take away. Certification measured what Cypress Grove's project team documented. Performance measured what the finished building does under real operation. They are different measurements of different things, and a building can score well on one and poorly on the other without anybody having lied.

The systems that measure the second thing are the measured-performance ones from §36.3: ENERGY STAR benchmarking against real utility bills, Passive House's tested envelope, the Living Building Challenge's twelve months of operating data. The City of Rivermont now benchmarks all of its buildings annually against actual consumption, which is the correct response and costs almost nothing. That is the change that came out of this, and it came from an owner, not from a contractor.

Discussion questions

1. Amara said nobody promised the City 38 kBtu/SF·yr. Is that a satisfying answer to a facilities director looking at a $123,000 budget overrun? If you were on the design team, what would you have done differently at the moment the number entered the City's budget?
2. Kestrel absorbed roughly $22,000 of warranty correction without arguing. Was that the right call? Build the argument that it was not, and then say which argument you would actually make to your VP of Operations.
3. The over-ventilating VAV boxes caused comfort complaints, which caused setpoint tightening and space heaters, which increased consumption further. Trace that chain and name two other places on a building where a construction defect produces an operational response that compounds it.
4. Enhanced commissioning would have caught most of the fixable gap. Who should pay for it — the owner, as an added service, or the contractor, as the cost of proving its own work? Argue both, then say which allocation produces better buildings and why.
5. The City now benchmarks every building annually against actual utility data. What would change about how Kestrel bids and closes out municipal work if every owner did that?

Your turn

Find a real building you can get utility data for — your campus, your employer's office, a public building whose energy data is published (many cities publish benchmarking data for municipal buildings, and ENERGY STAR's benchmarking program is the usual mechanism).

Estimate its floor area, compute its EUI in kBtu per square foot per year from twelve months of utility bills, and then answer three questions in writing:

1. How many hours a week is that building actually occupied and conditioned?
2. What continuous loads exist that a standardized schedule would not assume — a data closet, a kitchen, a 24-hour function?
3. If someone handed you a compliance model EUI for that building, by roughly what factor would you expect actual consumption to exceed it, and why?

You will not get a precise answer. That is the point of the exercise. What you will get is the instinct to ask what a model assumed before you let anybody put its output into a budget — and that instinct, deployed once in a preconstruction meeting, is worth more than any number in this case study.