Register schedule

The per-register schedule for the proposed scheme. Sizes come from Eldr’s equal-friction math at 0.08 in.wc/100 ft, hand-tuned as the layout firmed up; the reasoning behind the airflow is in the scheme. Re-derived against the current model on 2026-09-18 — the loads moved 2% overall and one duct changed size.

Airflow is biased on purpose — second floor ×1.35, basement ×0.70, main floor unchanged — and computed on a 30°F supply-air rise, the heat-pump figure. (A 50°F rise is a gas furnace and undersizes every heating-driven duct by 1.67×.)

Every size below is per individual duct, not per room. A room with two supplies gets two ducts of the size shown.

Schedule

Rectangular sizes are the shallowest option holding an aspect ratio at or under 4:1. Flatter than that costs friction and makes fittings awkward, so a 3x20 is not a substitute for a 4x14 even though the areas are similar.

Two return decisions are baked into the table above.

The Utility Room return joins the Kids Room branch rather than a basement trunk. That branch passes through the basement ceiling on its way down, so tapping it is the short path — but it means the branch carries 147 CFM, not 64. It is drawn at 8″, which runs 421 fpm. A 6″ sized for the room alone would run 747. A shared branch is sized for the sum, and the room name on it stops being the whole story.

The basement runs two ceiling returns rather than one midline. One shared midline grille would be the cheaper build — three new ducts instead of five — but the two rooms are separated by the stair and utility walls, so it would pull almost entirely from whichever room it sat in and leave the other giving its air up through a doorway. Each room returns where it is supplied, at 87 and 82 CFM.

The Play Room gets two supplies, and the new one is sized as though it were the only one. The existing east-wall duct is small enough to be suspect, so the new south-wall register is sized for the room’s full 117 CFM rather than for half of it. If the east duct turns out to carry its share, the room is comfortably over-served and the damper takes it back; if it carries nothing, the room is still right. Sizing the new duct for half would have made the room hostage to a duct nobody has measured. Treat the east register as recovered capacity, not as a design assumption — it is the one supply in this house whose contribution is unknown.

The second-floor returns are sized for the cross-flow split, not for their own rooms. The Play Room’s is sized at 168 rather than its own 117, so the option to drag air across from the Office exists without re-ducting. Their combined capacity is 302 against the 251 the floor needs — the dampers set the real split, and sizing for the larger case costs a nominal size, not a redesign.

Room-name mapping, since the model and these documents differ: the 2nd-floor “kids room” is Play Room; the main-floor “small kids room” is Kids Room; “bathroom” with supply only is Main Bath. The two basement supplies are named by position in the model and by room here — “west basement” is Future Media Room, “basement east” is Utility Room.

Register and grille faces

Sized on face velocity only. Supplies assume 75% free area — the generous end for a louvred face, so a grille with heavier blades runs faster than shown, never slower. Returns are computed on gross area.

Targets: supplies 500–750 fpm is the normal band and 600 is where this document starts objecting; returns 300–500 fpm, with the sizes below aimed at the quiet end. That aim is a choice, and an expensive one — a return sized at 300 fpm is two-thirds larger than the same airflow at 500. Treat 500 as the requirement and the rest as comfort.

This is not Manual T. Throw, spread, drop and NC need the room’s dimensions against a specific manufacturer’s catalogue, and the choice of where on the wall or floor matters as much as the size. Treat the faces below as the starting schedule a contractor prices, not as a selection.

Faces below are as drawn in the model, sized against available space, with the resulting velocity computed. Every face is now inside its band.

The two kitchen faces

Both are constrained by cabinetry, and both are drawn at the size the airflow asks for:

The kitchen SE register needs 4″ of height, and getting it is carpentry. At the 2″ the cabinetry currently leaves, the same airflow runs 806 fpm — audible, in a spot where somebody stands to cook. At 4″ it runs 407. One cabinet moves over, or the cabinet is modified; the drawn face assumes that happens. Note that 407 fpm assumes a generous 75% free area, so a heavier louvred face runs faster.

The kitchen return was about area, not length. 251 CFM needs roughly 120 in² at 300 fpm, and the 3x20 gave 60. What it did not need was to stay 20″ long: preserving the drawn dimension and growing only the other gives 6x20, a 3.3:1 grille nobody stocks.

8x14 is the pick — a stock size at 323 fpm, and what the model now carries. 6x12 is defensible if space is tight, at the cost of a return you can hear.

If the 3″ height is fixed by cabinetry, no single grille solves it — 3x40 would be needed, and the answer is two openings rather than one enormous one.

Already corrected in the model: Main Bed W supply 4x8 → 6x11, matching its NE twin at the same 113 CFM, and Office SW return 4x8 → 6x8. Neither had a space constraint; both were simply drawn small.

The Main Bed W register may also want to point east rather than the way it is drawn — a throw question, and throw is Manual T’s business rather than something the plan view can settle.

The main-floor return path is the real bottleneck

It was a single 4x8 running 30 ft with three returns hanging off it — Main Bed 227, Kids Room 64, and the Utility Room’s 83 via the Kids Room tap. That is 374 CFM through 32 in², about 1,680 fpm.

The run is now split at the point where it stops being buried, which is the remedy this section used to argue for:

The two rooms behind the pinch are out from behind it. The Kids Room and Utility Room now return through 60 in² at a quiet 353 fpm, where before they shared a duct running four times over.

The main bedroom segment still reads 1,022 fpm, and that number is notional. It is the existing duct through the inaccessible crawlspace, drawn at the size it actually is. At the ~91 CFM it is believed to pass it runs 410 fpm and is quiet — the duct is not undersized for what goes through it, it is undersized for what the room wants. That gap is a capacity problem to solve at the grille or with a transfer path, not a duct to re-size in the model.

The split still rests on drawn adjacency. Verify the topology on site before pricing anything — specifically that the open basement run really does extend as far as the model puts the split.

The Main Bed return is not in that list, though its number is the worst. 726 fpm assumes 227 CFM actually arrives, and it cannot: behind that grille is the existing 7.5x4 through the inaccessible crawlspace, good for roughly 91 CFM. At 91 the 5x9 face runs 291 fpm and is perfectly quiet. The grille is correctly sized for the duct; the duct is the problem, and the plan of record remains a transfer grille over the bedroom door first.

The Kids Room supply is the opposite case — 6x12 for 64 CFM is generous enough that throw suffers. Harmless in a small room, and worth leaving if the opening is already cut.

How trunks are sized

The duct system is a hierarchy, and each level is sized by the air passing through that particular segment — not by anything upstream or downstream:

Anything serving more than one register is a trunk. So yes — the large duct going up to the second floor is a trunk, and so is the return coming back down from it. Both carry 276 CFM and serve four registers before splitting.

A trunk shrinks as branches leave it. Two standard ways to handle that:

• Reducing trunk — step it down after each takeoff. Keeps velocity roughly constant, uses less metal, needs more fittings.
• Extended plenum — hold one size for most of the run and reduce once near the end. Simpler, slightly more material, and much the more common choice in residential work.

The main trunk may not need to exist

That 16″ figure assumes one duct leaves the plenum and splits later. It does not have to be built that way, and in this basement it probably should not be — a 16″ round plus insulation hanging under joists at 84″ is a real intrusion.

If the plenum carries takeoffs directly, the 16″ duct never exists. Splitting the main floor into two sub-trunks makes the largest single duct in the house a 10″:

Four takeoffs, nothing above 10″, and the south/north split happens to follow the routing already planned — south through the stairs, north through the basement bathroom and west along the girder. The cost is a larger plenum with room for four connections, and enough spacing between them that they do not rob one another.

Returns are not a mirror of the supply side

Total return must equal total supply. But the return side is a smaller number of larger openings:

• Fewer. Rooms without their own return give their air up through door undercuts or transfer grilles. Here that is Main Bath, Main Closet, Upper Bath and Upstairs Hallway.
• Larger per CFM. Returns are designed at lower velocity — roughly 400–600 fpm against a supply branch’s 600–900 — so the same airflow wants a bigger duct.
• Much larger at the grille. A supply register is limited by throw and noise to roughly 100–150 CFM. A return grille does not throw, so one can comfortably take 300–500 CFM. That asymmetry is why one return serves what needed two supplies.

Balance per level

With the Upper Bath and Upstairs Hallway carrying no supply:

Using return placement to balance the second floor

Oversizing the Play Room’s return to drag air across from the Office is a standard technique, and the numbers work. The only constraint is that the floor’s two returns sum to 251.

That drives 53 CFM through the bath and hallway on its way to the Play Room. Shrink the Office return further to pull harder; the pair just has to keep summing to 251.

The Office is the one room whose duct grew in the 2026-09-18 re-derivation: 113 → 134 CFM, which takes its supplies from 5″ to 6″. Its return duct is sized for the room’s own 134 and reads 503 fpm on that basis, but the cross-flow above throttles it to 81 — where a 7″ runs a quiet 305. Sized for the room, damped to the split.

The mechanism only works if the air has a path. Door undercuts of ¾″ or transfer grilles between office, bathroom and play room are what make this real rather than theoretical — without them the rooms pressurise and the flow stops.

Transit air sweeps a room; it does not heat one

This is the trap in leaving the Upper Bath unsupplied. Air arriving from the Office is at room temperature and has no heat to give up. Steady-state estimate for the bath, on a UA of about 16.5 BTU/hr·°F:

Conduction from adjacent rooms through interior walls is ignored, so reality is warmer — but no plausible transit rate reaches 70°F, because the air has nothing to deliver.

A bathroom wants its own heat. Three ways out, and the middle one is usually best here:

1. Supply it from the riser and accept the register landing over the stairs.
2. Electric resistance — a toe-kick heater or heated floor. Standard for exactly this case, cheap to run for a room this size, independent of the duct layout, and it makes the bathroom warmer than ducting would. It also avoids forcing the north run through the main-floor bathroom and kitchen void.
3. Accept roughly 60°F on design days.

The Upstairs Hallway needs no such treatment — it is transit space, tolerates being cool, and borrows heat from the rooms it connects.

Zoning — air is not water

The house already runs three heating zones, so three zones feels like the obvious carry-over. It is not, and the reason is worth being blunt about:

Zoning water is easy. Zoning air is hard. A zone valve closes and the boiler simply makes less hot water. A zone damper closes and the air has to go somewhere — the blower is still turning.

The arithmetic here is unfriendly. Against a whole-house design of ~1,180 CFM, and a 3.5–4 ton air handler whose minimum airflow is roughly 40% of nominal (550–640 CFM):

Neither the basement nor the second floor can be its own hard zone on a single unit this size. Either alone starves the blower and drives static pressure up.

What works, roughly in order of preference:

• Manual balancing dampers and no zoning. A modulating compressor running long low-output cycles does much of what zoning promises, because the classic reason to zone is a single-stage unit blasting and shutting off. Simplest, cheapest, and reversible.
• Modulating zone dampers that throttle rather than shut, so every zone keeps some flow. Avoids the minimum-airflow cliff entirely.
• A dump zone — usually the basement, damper never fully closing, so minimum flow is always available.
• A zone panel that reduces blower speed to match the open zones. Correct in principle, but see the caution below.
• A bypass damper. The old answer and the bad one: it recirculates supply air to the return, degrading dehumidification, and can freeze the coil in cooling or overheat it in heating. Avoid.

The caution that links back to controls: a zone panel able to stage the blower generally has to communicate with the unit, and zone panels are frequently where the open-controls option quietly disappears. Zoning and third-party thermostats pull against each other, the same tension as modulation and open control.

Recommendation: manual dampers now, zoning later if the balance disappoints. The stack effect is real and does not care about modulation, so this may not be the final answer — but it is the cheap experiment, and the ductwork for it is identical either way.

Throttling is not zoning, and the distinction matters

Everything above concerns hard zone dampers, where a whole floor’s damper shuts. Smart-home throttling of individual branches is a different and much safer thing, because throttling redistributes air rather than removing it — a partly-closed damper raises static slightly and the air leaves through the other branches. The system only starves when most of it closes.

A single motorised branch damper reacting to a local sensor — the computer rack under the NW basement supply, for instance — is a balancing damper that moves by itself. It is not a system-level risk.

Compressor turndown and blower minimum are separate constraints. The unit modulates capacity to roughly a third; the blower minimum exists to keep enough air over the coil, above freezing in cooling and below the high limit in heating. That minimum scales with output, so at a third capacity the required airflow falls too and demand tracks requirement. The failure mode is not low flow — it is dampers throttling while the compressor stays at full, which is precisely what two independent control systems can do to each other.

The architecture that works: the unit’s own thermostat decides whether the system runs and at what capacity; the smart-home dampers decide where the air goes within that. Clean separation, and it sidesteps the communicating-thermostat trap entirely, because the dampers are just motorised balancing dampers the unit never needs to know about.

The rule that keeps them from fighting: never throttle more than ~30–40% of total design airflow at once, and never fully close a damper. Give the smart-home logic a hard floor it cannot cross regardless of what the sensors say.

And if nothing calls, the system is simply off — the minimum-airflow problem only exists while it is running. For a basement that holds temperature unconditioned, letting it drift when it is the only caller uses thermal mass the house already has rather than conceding anything.

Duct materials

Flat oval deserves consideration here. Spiral-formed then flattened, it keeps most of round’s low friction in a profile that fits where round will not, which suits a house whose recurring constraint is headroom. Available double-wall.

Flex is fine and widely abused. Pulled tight and under about six feet it is a legitimate final connection and kills vibration. Sagging through a joist bay for twenty feet it is the commonest cause of a system that underperforms its design — and it is invisible once the ceiling closes.

Trunks cannot be resized decently. The practical hedge is not oversizing — a too-large trunk runs slow, and slow is what makes near takeoffs steal from far ones — but adding a parallel duct later, which is a further argument for the four-way plenum split: four modest ducts leave room beside them, one 16″ artery does not.

Notes for editing

• Splitting a room’s supply does not change its total — two registers at half the CFM each, and each duct drops about one nominal size.
• Velocities above ~700 fpm on a branch start to be audible. Everything here is well under.
• Under 10 CFM, do not run a duct. Closets, chimney voids, the bar area and the stair void are served by transfer.
• Future work is out of scope here. The garage mudroom conversion is deliberately absent; see the future-options section of ducting-scheme.md.