Ducted HVAC Cost $745K More on a 57-Unit Worcester Project

We modeled both ducted and ductless heat pump paths for a 57-unit all-electric multifamily project in Worcester, Massachusetts under Stretch Code. The ductless mini-split path passed with a comfortable margin on double pane windows, standard assemblies, and no exotic add-ons. The ducted central heat pump path failed. To make it pass, we had to layer on triple pane windows, tighter air sealing targets, heat recovery ventilation, and induction ranges. Even after all of that, the ducted path still performed worse than the ductless approach on the energy model. The price tag for making ducted work was $745,000 in additional construction cost across the 57 units.

The Real Project

The building was a 57-unit all-electric multifamily development in Worcester targeting Stretch Code compliance. Two HVAC paths were on the table.

The first was a ductless mini-split system with per-unit indoor heads and outdoor condensers sized to the load. Familiar territory for an energy model in Massachusetts. High HSPF2 ratings, no duct losses, no central air handler, and equipment efficiencies that the index responds to favorably.

The second was a ducted central heat pump configuration with a single air handler per unit and a duct network running through conditioned space. This is the path most developers default to because tenants expect ceiling registers and design teams have built it a hundred times.

We ran both against the same building, same envelope, same DHW configuration. The ductless path cleared compliance with margin. The ducted path failed.

The Ducted Compliance Penalty

The ducted central heat pump system performs worse on the energy model for three connected reasons.

The first is equipment efficiency. Ducted central heat pumps at the residential scale typically rate lower on HSPF2 and SEER2 than equivalent ductless mini-splits. The compressor and refrigerant strategy that lets a mini-split modulate down to low part-load operation is harder to deliver on a ducted central system. The model sees that gap and the index responds.

The second is duct losses. Even with ducts entirely inside conditioned space, the index assigns a thermal penalty for the air handling system that ductless equipment does not carry. The model treats a duct network as a real building load, not a paper assumption.

The third is the auxiliary fan energy. A ducted system runs a blower to push air through the duct network every time the equipment calls. That fan energy is captured in the energy model and it accumulates over a heating-dominant climate year.

Stacked together, those three penalties pushed the ducted path under the Stretch Code threshold the ductless path cleared without effort. To recover the headroom, the design had to spend it elsewhere.

What the Recovery Cost

The ducted path could not pass on the original envelope. Bringing it back into compliance required four upgrades layered onto the building.

The first was triple pane windows in place of the double pane package the ductless path used. On a 57-unit building, that is a six-figure window cost swing on its own.

The second was a tighter air sealing target. Hitting the lower CFM50 number that the ducted path needed required additional sealing labor, additional testing iterations, and a tighter coordination loop between the GC and the rater during construction.

The third was heat recovery ventilation. The ductless path passed on a standard ventilation strategy. The ducted path required HRV at the unit level to recover the heating load lost through the ventilation air. New equipment, new ductwork, new commissioning.

The fourth was induction ranges across every unit. A cooking equipment substitution that closes a small but non-trivial gap in the index.

Even with all four upgrades layered on, the ducted path still modeled worse than the ductless path on the unmodified envelope. The total cost of forcing ducted to comply was $745,000 in additional construction across the 57 units, and the resulting building was less energy efficient than the cheaper ductless alternative.

When Ducted Still Works

The 57-unit Worcester numbers do not say ducted central heat pumps are always the wrong call. There are buildings where the math runs the other way.

Large central plant configurations. On certain larger multifamily projects with a central VRF system serving zoned ducted air handlers, the equipment efficiency profile is different and the duct penalty can be absorbed. The path is real, it just is not the residential-scale central heat pump that gets specified by default.

Buildings with significant cooling load. In a cooling-dominant climate, the equipment efficiency gap between ducted and ductless narrows, and the duct distribution can serve cooling effectively. Massachusetts is not that climate, but the same building in a different region runs differently.

Projects where the envelope was always going to be high-performance. If the architect was already drawing a triple pane, HRV-equipped, tight envelope for other reasons, the ducted path does not carry a true cost premium because the upgrades are not incremental.

Tenant comfort or acoustic constraints that ductless cannot meet. On certain high-end residential products, the design team may have non-negotiable reasons to deliver ceiling-register conditioning. The cost premium is then a known program decision, not a compliance surprise.

The Decision Framework

The pattern on this project is not a referendum on ducted equipment. It is a code compliance cost problem. Under MA Stretch Code with all-electric construction, the equipment efficiency gap between ducted and ductless heat pumps is large enough that the ductless path clears compliance on standard assemblies while the ducted path requires expensive envelope upgrades to make the math work. Familiarity with ducted design is not free. On this 57-unit building it cost $745,000.

The honest answer is that the right HVAC path depends on the specific building, the specific code, and the specific envelope the architect is drawing. The way to know which one is cheaper on a particular project is to model both before the design is locked. On this project, the answer was worth $745K. On a different project, the answer could swing the other way.