VAV Systems Air Systems

The last page ended on a promise: follow the supply duct away from the unit and you find boxes with actuators — and what those boxes are actually doing is a page of its own. This is that page. One question, start to finish: how does one air handler serve thirty zones that all want different things — what is a VAV box actually doing when its zone calls?

One Trunk, Thirty Claims

Start with the problem. One unit, one supply temperature — and thirty zones that have never once agreed on anything: a corner office roasting in afternoon sun, a packed conference room two doors down, a north wall that hasn't felt sunlight since the building topped out. A constant-volume system settles that argument by ignoring it: airflow to each zone is fixed at balancing, and the thermostat war is won by whoever sits closest to the sensor. Variable air volume settles it differently — it gives every zone its own throttle.

The trick is what stays constant. The air handler keeps its supply air cold and steady — around 55 °F, all day, for everyone — and each zone's box decides how much of that air the zone receives. More load, more cold air; less load, less. The box never changes what the air is, only how much of it there is: a VAV box throttles volume, not temperature. Thirty different loads, one supply temperature, thirty different airflows.

If that sentence rings familiar, it should — it's the load-piping story with the fluid swapped. There, every coil got a two-way valve and throttled water; the plant's job was to respond. Here, every zone gets a box and throttles air. Loads throttle, and something upstream has to answer for it — the answering half of the mirror is the last page of this chapter, the same way pump control answered for the valves.

Watch what the trunk feels while the zones do this. Total supply flow is just the sum of what thirty boxes are taking, and that sum swings all day — morning pull-down runs everything near wide open, a mild afternoon leaves most boxes idling near their minimums. The supply fan rides a VFD and slows with the building, which is where VAV earns its keep: cube law pays the fan's part-load bill. How the fan knows how much to slow down is the chapter's last page. This page stays at the boxes — because everything the fan will have to answer for starts with what the boxes do.

Inside the Box

For all the load it carries in a mechanical plan, a VAV box is almost disappointingly simple sheet metal: a round inlet collar, a rectangular body, a discharge to the zone's diffusers. Three parts do all the work. A damper — one blade on a shaft, swung by an actuator. A flow sensor across the inlet — a ring or cross of averaging ports, one set facing into the airstream, one away, with two small tubes running to a pressure transducer. And a controller — a small BAS device, usually the most numerous controller family in the whole building, which is why every networking example you've ever seen names a device VAV-101.

The flow sensor is the part worth slowing down for, because it means the box measures its own airflow. The pressure difference between the upstream and downstream ports — the velocity pressure — rises with the square of the air's speed, so flow follows a square root: CFM = K × √VP, where K is the manufacturer's coefficient for that pickup — literally the CFM it reads at exactly 1.0 in. w.c. A box with K = 1000 reading 0.36 in. w.c. is passing 600 CFM. The airflow tool runs that math in both directions — flow from K and VP, or K back-solved from a balancer's hood reading — and carries the fine print (K conventions vary by vendor, and the whole frame is US-native; the K printed on a metric box is a different number on a different label).

Two more parts appear on boxes that can heat: a reheat coil at the discharge — hot water on a small valve, or electric elements in stages — and a discharge-air sensor (DAT) just downstream of it, so the controller can see what it's actually sending to the zone. What they're for is a few sections down; first, the question that makes the flow ring earn its place: why does the box measure flow at all, instead of just holding a damper position?

Why the Box Chases Flow, Not Position

Try the naive version. Command the damper to 50 % open and walk away. How much air is the zone getting? You don't know — and worse, the answer won't hold still. A damper position only fixes a geometry; the flow through that geometry depends on the pressure behind it, and the pressure behind it depends on what every other box on the trunk is doing and how hard the fan is pushing. Let the neighbors throttle back and the pressure at your inlet rises, and your 50-percent-open damper quietly delivers more air than it did an hour ago, to a zone whose load never changed. Damper position is nobody's setpoint.

So the box runs a cascade of two loops instead. The first watches the zone: temperature against setpoint, and its output is not a damper command but a flow setpoint — some CFM between the box's minimum and maximum, more when the zone is hot, less as it cools. The second loop watches the flow ring: measured CFM against that setpoint, driving the damper to whatever position closes the gap — 30 % today, 55 % tonight, nobody cares. When the trunk pressure wiggles, the flow reading wiggles, and the second loop absorbs the disturbance before the zone ever feels it. That's what pressure independent means: the zone gets the airflow it asked for regardless of what the rest of the building is doing to the duct. The hydronic side pulls the same trick with a PICV — a valve that meters flow, not position — and for the same reason.

One boundary worth drawing before moving on: the box's damper serves the zone's flow, and only that. It doesn't hold duct pressure, doesn't help its neighbors, doesn't coordinate. Thirty boxes each minding their own loop is the whole choreography — and everything the system does well or badly emerges from that.

The Floors: Ventilation, Then Reheat

Follow a zone as it cools off and the flow setpoint rides down — and notice where it stops. Not zero. Every box carries a minimum, and the first reason is the one this chapter keeps returning to: occupied zones need their share of the code-required fresh-air dose, and that share arrives riding the supply air. Close a box entirely and the temperature might be fine while the ventilation quietly isn't — the ventilation floor is not negotiable while the building is occupied, and at the box scale the minimum CFM is that floor.

But the minimum creates its own problem. Minimum flow is still 55 °F air, arriving whether the zone wants cooling or not — and a small interior room at minimum flow will drift cold by mid-morning. That's what the reheat coil is for: the box holds flow at minimum and warms the air on its way out. The better sequences run it as another cascade — the zone loop asks for a discharge temperature, the DAT sensor closes that loop on the reheat valve, capped somewhere around 90 °F so the warm air still mixes down into the room instead of stratifying at the ceiling. Yes, it's the plant heating air the plant just paid to cool — which is why reheat is metered carefully, and why the minimum it runs at is worth getting right rather than getting generous.

And one floor lives at the unit, not the box. The building-pressure page filed this wrinkle: on a VAV unit the supply fan slows as the boxes throttle, and a fixed outside-air damper position passes less actual outside air at lower fan speed. A minimum-OA setting that was honest at design flow is a fiction at half flow — which is why serious VAV units measure their outside air or actively control to it, rather than trusting a damper angle set one afternoon at full speed.

The Machine Has a Minimum Too

Every floor so far protected the zone — its fresh air, its comfort. The last one protects the machine, and it's the one this page's author has the scars for. The unit's cooling coil was sized for design airflow, and on a chilled-water coil, part flow is graceful: less air means less heat picked up, the valve throttles back, the water side shrugs — that's the hydronic story again. A DX coil doesn't shrug. It's the evaporator of a refrigerant circuit, and a running compressor moves refrigerant whether or not there's warm air to boil it. The field rule of thumb is somewhere near 400 CFM of air per ton of active cooling (an IP-native number, like the K-factor — it's the frame the trade speaks). Starve a running stage much below that and suction pressure dives, the coil surface drops below freezing, and the condensate that should be dripping off the fins freezes onto them instead. Ice chokes airflow further, which starves the coil further — a runaway with a ratchet. Left alone it ends with a coil iced solid and, eventually, compressors slugging liquid and short-cycling toward the scrap bin.

Now connect that to what the boxes do. On a mild afternoon most zones satisfy, thirty flow setpoints ride down to their minimums, and total airflow across that coil collapses — while a handful of zones still calling keep a compressor stage running. Nothing about the box side is malfunctioning; every box minding its own loop is exactly what starves the coil. So a DX-VAV system has to be designed with somewhere for the mismatch to go, and the lineup looks like this: box minimums summed generously enough that the coil's floor is never violated; staging interlocked to proven airflow, so a stage locks out rather than run starved — the interlock chooses the compressor over the complaint; a bypass damper from supply to return, so the coil sees constant volume while the zones see variable (an older answer, with its own problems); or a dump zone — a box into a lobby or corridor that opens when flow collapses. A properly designed unit has at least one of these. When none of them exist, the system runs fine on the design day, and eats itself on the mild ones — the widget's last preset is that building.

Drive a Box, Watch the Coil

One box under your hand, twenty-nine siblings and a two-stage DX unit behind it. The slider is the zone; the box answers with a flow setpoint between its minimum (200 CFM) and maximum (1,000 CFM), the flow ring reads its velocity pressure (K = 1000 — at max flow the pickup reads exactly 1.0 in. w.c.), and the damper lands wherever the flow loop needs it. Change what the siblings are doing and watch the damper move while the flow doesn't — that's pressure independence earning its name. Below the seam, the system strip keeps the coil's books. Then try the last preset, which reproduces a pair of air handlers this page's author is still babysitting.

What the widget holds still: the duct-static loop (the fan is assumed to keep up — that story is the next page), humidity and latent load, and real staging logic (two clean stages stand in for whatever the nameplate actually does). The other twenty-nine boxes come in three coarse moods rather than twenty-nine sliders. Velocity pressure and CFM-per-ton stay in their IP field frames — the same US-native posture as the airflow tool — while flows and temperatures follow the units toggle.

Loads Throttle; the Mover Responds

That's a VAV system from the box's side of it: one cold trunk and thirty throttles; a box that measures its own flow and chases CFM, not damper position; floors under everything — ventilation at the box, reheat to keep the minimum livable, outside air measured at the unit, and airflow for the coil so the machine survives its own part-load days. It's the air-side half of a mirror this site has drawn once before: loads throttle, the mover responds. This page was the throttling half.