Brief
The client — a regional decorative-arts foundation, named privately on request — purchased a vacant 1898 textile mill in 2022, on a 4.4-acre lot south of the Monongahela, with the intention of converting it into a study center for a 9,400-piece collection of regional decorative-arts objects: ceramics, glass, metalwork, furniture, textiles, and wallpaper. The mill had operated continuously from 1898 to 1979, was vacant from 1979 to 2017, and was used for film-set storage from 2017 until the foundation purchased it. The building is on the National Register of Historic Places as a contributing property in a textile-industry historic district. The foundation’s rehabilitation is being financed in part through Federal Historic Preservation Tax Incentives, which means a Part 2 application has been filed and approved with NPS Technical Preservation Services.
The collection’s class requirement: Class B for the bulk of the building (±5 °C, ±10% RH, with seasonal RH adjustment of ±10%), with two small Class A reading rooms inside the larger Class B envelope. The Part 2 application disallows changes to the historic windows and disallows visible mechanical equipment on the main north and south facades. We were retained in October 2024 in coordination with the foundation’s historic-preservation architect and a structural engineer who specializes in heavy-timber industrial fabric.
The Part 2 constraint
The 119 mill windows are the building’s defining historic feature. They are 8-light wood-sash assemblies with original 1898 wavy crown glass, set in segmental-arch openings of red brick. The Part 2 application calls them out specifically, and disallows: (a) replacement of any sash or muntin, (b) replacement of any glazing, (c) installation of secondary glazing on the interior face if the secondary glazing is visible from the exterior, (d) any blocking of the daylight delivery into the bays, (e) any window-mounted mechanical equipment of any kind. The mechanical, in other words, has to live on the inside, has to leave the windows alone, has to operate quietly enough that the foundation’s docents will not be making excuses for it during tours, and has to deliver Class B regardless.
This is not a unique constraint — we have run NPS-Part-2-bound projects before — but the 119-window count makes it unusually demanding. The window perimeter is approximately 1,790 linear feet of un-tightened thermal bridge, and the building load calculation is dominated by it.
Reading the wall
Gate M1 was a six-week instrumented walk (longer than our usual two weeks because the building’s seasonal behavior was the question, and a fortnight in early winter wouldn’t have answered it). We deployed 38 sensors across both floors and ran a hygrothermal model in WUFI of the wall section, with and without proposed envelope improvements. The model showed that the existing wall section, given existing window stack-effect leakage, would not support a Class A envelope on the upper floor without sweat at the inner brick face during three-day cold snaps. It also showed that a Class B envelope, with seasonal RH adjustment of ±10%, would be supportable across the full annual cycle with the addition of (1) a tight inner air barrier in the bays, and (2) a careful mechanical sizing for the worst week.
The risk register named Class B as the building-wide regime. The two Class A reading rooms, on the south side of the second floor, were sized as “rooms within rooms”: separate vapor-tight envelopes inside the Class-B shell, each with its own dedicated mechanical zone, each able to be commissioned independently of the larger building.
Mechanical schematic (Gate M3)
Two primary chillers (200 tons each, magnetic-bearing, R-1234ze, N+1), located in a new mechanical room threaded into the existing first-floor power-shaft alley (an existing inset in the building footprint, invisible from the historic facades). Three air-handling units, two on the second floor (one for the main bay, one for the two Class-A reading rooms) and one on the first (for the bulk-storage area). DOAS/ERV at 4,800 cfm of outside air, located in a roof penthouse hidden behind a parapet on the back side of the building. A sensor mesh of 84 LoRaWAN nodes. A Niagara BMS with a federated control structure (each AHU runs its own loop and they negotiate setpoints through a shared BMS).
Two unusual details:
- The window perimeters are receiving an interior air-barrier treatment using Dörken SD-membrane, terminated in a custom millwork-integrated reveal that the historic-preservation architect specified. The reveal makes the air barrier invisible from the gallery floor while keeping it accessible for service.
- The two Class-A reading rooms have their own dedicated dehumidification skids, located in adjacent service alcoves. The alcoves were existing power-shaft chases. Nothing was added to the building’s footprint.
Where the project is now (May 2026)
Gate M3 closed in February 2025 with a stamped drawing set. Gate M4 began in March 2025. As of May 2026, the chillers are installed and pressure-tested, the AHUs are in place, the ductwork is roughly 80% complete, and the window-perimeter air-barrier treatment is roughly 60% complete. The two Class-A reading rooms are framed but not yet mechanically tied in. We are projecting Gate M4 closure by November 2026, with Gate M5 (equilibration) running through 2027 and final commissioning in early 2028.
The project is complicated, and so far it is going well. Three things have surprised us:
- The building’s heavy-timber columns are dimensionally stable in ways the WUFI model did not predict. The columns are taking up moisture more slowly than a generic heavy-timber section would, and releasing it more slowly. This is partly because the original mill ran at relatively stable 60% RH for 80 years, and the columns have memory.
- The 1898 wavy-glass window assemblies are not as leaky as we feared. The instrumented walk found a window-perimeter leak rate of about 0.74 ACH at 50 Pa across the bay, which is a competent number for an 1898 sash. The original glazier was apparently good at his job.
- The Part 2 application has been a help, not a hindrance. The constraint to leave the windows alone forced us to design a more thoughtful mechanical, and the resulting building will use less energy than a Class-B mechanical that had been allowed to compromise the windows.
The casebook entry will be updated at final commissioning, projected 2028. We will write up the energy report and the equilibration log at that point.