CSU Chancellor’s Office AHU Renewal

The building was too cold. Staff supporting all 23 CSU campuses — running the administrative headquarters for the largest four-year public university system in the country from a 155,000-square-foot office on the Long Beach waterfront — were showing up to desks that felt like walk-in freezers.

The diagnosis was already done by the time the RFP hit our desk. The fix wasn’t.

How this project started.

CSU brought in Taylor Engineers to figure out why the building was overcooling. Their trend review, dated June 2023, pulled a full year of one-minute BAS data across the chiller plant, air handler, and zones and worked backward from the symptoms.

The AHU chilled water valve was staying pinned at 100% open instead of modulating. That drove supply air ten degrees below setpoint and forced zone reheat coils to burn energy compensating. The enthalpy economizer was misbehaving, sometimes closing the outside air damper when free cooling was sitting right outside. The chillers were short-cycling. And the supply and return fans hadn’t risen above 80% speed for more than six hours in the entire calendar year. They were oversized for the load the building actually carried.

Taylor’s recommendations: reprogram sequences to ASHRAE Guideline 36, replace faulty temperature sensors, lock out the chillers against a variable supply-air setpoint, drop the enthalpy economizer for dry-bulb (California doesn’t need enthalpy control), and right-size the fans to roughly 80% of the original design airflow.

That’s the report that shaped the RFP.

The RFP — Collaborative Design-Build.

CSU issued the AHU Renewal Project bid solicitation as a Collaborative Design-Build. That’s CSU’s variation on progressive design-build: two contracts, one team. Phase 1 covers design and preconstruction and ends when the team commits to a Guaranteed Maximum Price. Phase 2 is the construction contract, executed only if the GMP checks out. The Trustees are not obligated to proceed to Phase 2 with the Phase 1 team.

CSU’s selection preferences told you what they cared about. They gave scoring preference to teams proposing the largest amount of self-performed work. And because this project is fundamentally mechanical, they required a licensed mechanical prime contractor holding the C-20 license. General contractors bidding with a subcontracted mechanical scope were not eligible.

Athena was prequalified, invited to submit qualifications in July 2023, shortlisted, and awarded the contract in December 2023.

The building.

401 Golden Shore, on the Long Beach waterfront. Next to the Aquarium of the Pacific and the Catalina Express terminal. Original construction late 1990s (design drawings dated 1997). 155,000 square feet across the office floors, with a mechanical penthouse on the roof.

The penthouse is a run of small rooms: fan room, machine room, chiller room, vestibule. All sized to the equipment installed with the building. Enough room to walk between the fans. Not much more.

The design phase.

We ran design in the standard four steps: 30% schematic, 60% design development, 90% construction documents, 100% CDs. At each milestone we ran a fresh estimate against the working budget and reviewed with the CSU project team before proceeding. No surprises at the 90% pricing round. That’s the whole point of progressive DB.

100% CDs finalized June 2025. California State Fire Marshal approval July 2025. CSU construction permit August 2025.

The space problem.

Access to the mechanical penthouse was the same access anyone else used: passenger elevator to the top floor, then flights of stairs, then a 7-foot door into the mechanical room. Every rigging path had to be engineered around every corner, every ceiling drop, every existing service.

Cranes staged from the north parking lot. Materials laid down on the south lot. Rooftop temporary laydown protected the existing roof membrane with plywood. Nothing set directly on the roof.

Supply and return fans were phased as consecutive mobilizations. Never concurrent. One fan system down at a time. The building stayed occupied and operational through the entire retrofit. No cold offices, no closed floors, no reroutes.

Why FanWall — and why right-sizing mattered.

Taylor’s trend data showed the original supply and return fans had never worked harder than 80% capacity in the entire calendar year of 2022. The story wasn’t aging equipment. Those fans were oversized to begin with, sized for a building the designers projected in 1997 that turned out to run at lower loads once it was actually occupied. A like-for-like replacement would have installed the same problem again.

Nortek FanWall arrays let us right-size properly. Instead of one large centrifugal fan, an array of small individually-controlled fan cubes stages fans on and off with demand. You get part-load efficiency the original single-fan design physically could not achieve. Three specific wins for this retrofit:

• Right-sizing. Design airflow dropped to ~80% of the original, matching what the building actually needs. Fans run in their efficient range instead of throttling against a damper.
• Redundancy. One cube fails, the array keeps moving air. On a single fan, one failure meant no supply.
• Rigging math. Each cube fits through the existing 7-foot mechanical room door. No structural modifications, no rooftop crane through the roof deck.

Two new variable frequency drives, one per array. A third existing VFD relocated as part of the retrofit.

The BAS work.

The existing plant controllers were the same generation as the fans. Enclosures were structurally sound and stayed. The internals came out and were replaced. Chiller plant, boiler plant, and air handler all migrated to modern Distech Controls hardware.

The sequences were rewritten from scratch against Taylor’s recommendations: ASHRAE Guideline 36 control sequences on the AHU, chiller lockout tied to a variable supply-air setpoint, dry-bulb economizer control instead of the failed enthalpy control, chilled water valve programmed to actually modulate. The problems Taylor documented in June 2023 don’t exist anymore because the controls were built to prevent them.

Athena did not self-perform the building automation on this project. The Distech install and programming was handled by the design-build team’s specialty controls contractor. We led the mechanical scope, the general contracting, the crane and rigging planning, and the commissioning coordination. Every sequence verified against the design SOO before signoff.

For the record, self-performing our own BAS is our default when project shape allows it. On Collaborative Design-Build teams the CSU model brings specialty firms together on one contract. Everyone brings a specialty and stays in it.

What CSU got.

Not “new fans.” A right-sized fan system, running modern sequences, on modern controllers, doing what the original design intended and never actually achieved. The overcooling complaint that started this whole project has an addressable root cause instead of a workaround.

Two Nortek FanWall arrays. Three VFDs commissioned. Chiller plant, boiler plant, and AHU on Distech controllers with a fresh sequence of operations that reflects how the building actually runs today.

Existing panel enclosures retained. Existing field wiring, where it made sense, retained. That’s Athena’s default posture on retrofits: replace what has aged out or was wrong the first time, keep what still works, save the owner the money we do not need to spend.

Why Collaborative Design-Build was the right call.

The alternative on a project like this is a lump-sum bid off finished drawings. That would have meant engineering the fix blind: no site walks with the design team, no destructive investigation of existing conditions, no way to catch the space constraints until construction pricing was already locked.

Collaborative DB let us find the tight spots at 30% design and re-engineer around them. Lump-sum bids surface those same problems at 60% construction, when they become change orders. It also let CSU stop the work between phases if the numbers had come back wrong. They did not. The GMP came in at 90% CD, after two prior estimating rounds at 30% and 60% had already surfaced and resolved the biggest cost drivers.