Duct Static Control Air Systems

Pick up exactly where the last page left you: it's six o'clock, thirty zone loops have satisfied within the same half hour, and thirty dampers are gliding toward minimum together. The fan is still pushing — into a duct that is closing like a fist. Something has to give, and this page is about the thing that gives. One question, start to finish: how does the supply fan know how much air the building wants — and why does it hold duct static pressure rather than flow?

The Duct Answers for the Boxes

Start with what the fan cannot do. It cannot ask the zones what they need — thirty flow rings belong to thirty box controllers, each minding its own loop, and nobody at the box level is minding the trunk. And it shouldn't chase total flow even if it could sum it, because total flow isn't the fan's promise to keep. The zones decide how much air moves — that was the whole point of the last page. What the boxes need from the fan is simpler and humbler: enough pressure at the trunk that any box can take what it wants. The fan makes pressure available; the boxes decide flow.

And pressure turns out to be a wonderfully honest messenger, because the duct does the arithmetic for free. Thirty boxes all pull from one trunk. When they throttle back, the fan is suddenly pushing the same effort into less opening, and trunk pressure rises. When they open up, the trunk bleeds down. Boxes close, static rises, the fan slows; boxes open, static sags, the fan speeds up. One pressure sensor summarizes thirty demands — no polling, no totalizing, no arithmetic that can go stale. The fan answers to the building, not to a clock.

If that four-beat sentence rings familiar, it's because pump control taught it first with water in it: valves close, DP rises, the pump slows. Same sign, same logic, same cube-law payday. The hydronic arc drew this mirror across two pages — loads throttle, the mover responds — and the forced-air chapter has now drawn both halves of its own: the last page was the throttling, this page is the response. Six o'clock stops being a cliff-hanger and becomes a sequence: the dampers glide closed, the trunk pressure climbs, the loop sees it climb, and the fan backs down the same way it came up. What gives is the fan — gracefully, if this page's loop is doing its job.

One Sensor, Two-Thirds Down the Duct

The loop itself is three parts, and none of them is exotic. A pressure transducer out in the supply trunk — a small box with a plastic sensing tube poked through the duct wall, reading duct pressure against the space around it and reporting back as an analog signal. A setpoint, somewhere around 1.5 in. w.c. on a typical mid-size system — the same "whole inches" magnitude that building pressure measured its envelope against and found fifty times smaller. And a PID loop whose output is the speed reference to the supply fan's VFD. That's the entire machine: one AI, one AO, one fast loop between them.

Fast is worth pausing on. The zone loops you met last page think in minutes — a room warms slowly, a damper strokes, the world catches up. Duct pressure answers the fan in seconds: touch the speed reference and the sensor feels it almost immediately. Loops that quick are tuned gently or they hunt, and this exact loop — supply fan VFD, duct static — is one of the scenes in the PID tuner. Drive it there and feel how little gain it wants; that's the hands-on pair for this page.

Where the sensor lives is the design decision. Mount it at the fan discharge and the loop holds pressure where nobody uses it — on a design afternoon the far end of the trunk has spent most of that pressure on duct friction before the last boxes ever see it, so you're forced to hold a high number all day to protect a worst case the sensor can't even see. That's the same mistake pump control named on the water side: local DP at the pump tells you nothing about what the far end of the loop is feeling. Water's answer is to move the sensor to the most-remote load. Air practice settles for a compromise with the same logic in it: about two-thirds of the way down the trunk — far enough out that the reading resembles what the boxes actually feel, close enough in that one sensor still speaks for several branches. Hold a modest number where it matters instead of a big number where it doesn't, and every hour of the day gets cheaper.

The profile picture is worth keeping. On a design afternoon the trunk is a steep hill — high at the fan, spent along the run. On a mild evening it's nearly flat. The one point that never moves is the sensor's: the loop pivots the whole profile around the number it holds. Everything upstream of that point is the fan's business; everything downstream is the boxes'. Keep that split in mind — it's about to matter.

Setpoint Reset: Cube Law One More Time

A fixed setpoint has the same quiet flaw on air that it had on water: it's sized for the worst case, and the worst case almost never shows up. That 1.5 in. w.c. was chosen so the farthest box can still make design flow on the hottest afternoon of the year. On the three hundred other days, the boxes are half closed, the trunk needs far less — and the loop dutifully holds the design number anyway, with every box pinching off the difference across its own damper. Pressure the fan paid cube-law money to make, throttled away at thirty dampers, on purpose, all day.

The fix is the same one pump control reached for: reset the setpoint down when nobody needs it. The air-side recipe is usually called trim & respond — the BMS polls the box dampers, and as long as no box is near wide open, it trims the static setpoint down a small step at a time; the moment some box drives toward its stops asking for more than the trunk offers, it responds back up. The steady state is elegant: the setpoint settles wherever the most-open damper in the building sits nearly wide open — meaning that box is doing essentially no throttling, and the fan is making only the pressure someone is actually using. There's a floor under the walk-down (hold too little static and the far boxes can't make even their ventilation minimums), but between floor and design the setpoint follows the building.

Why bother? Because the fan is the biggest motor on the unit, and power follows the cube of speed — half speed costs an eighth (≈13%) of the power. Every tenth of an inch the setpoint walks down lets the fan run that much slower for the same delivered air, and the savings compound exactly the way the affinity-laws tool computes. A VAV system with a fixed setpoint captures some of the cube-law money; a VAV system with reset goes back for the rest of it. In the widget below, the reset toggle is the difference between the fan idling through a mild evening at nearly a fifth of design power and idling at a twentieth.

Static Is Not Flow

Now the hard lesson, and it's the reason this page exists. Everything above says the fan infers what the building wants from one pressure reading. An inference is only as honest as the assumptions under it — and the load-bearing assumption is that the path between the fan and the boxes is the path the designer drew. Break that assumption and the loop keeps working perfectly while the number it serves stops meaning anything. Duct static is a pressure promise, not a flow promise. Three ways that bites, in escalating order:

First — by design. 1.5 in. w.c. at 30,000 CFM and 1.5 in. w.c. at 7,600 CFM are indistinguishable at the sensor. That's not a failure — it's the whole design. The loop holds pressure precisely so that flow is free to collapse when the zones stop asking. But it means a healthy static reading tells you nothing about airflow, which is why the last page's DX interlock had to be proven on airflow, never on static. The building from that page starved its coil all afternoon with the static trend sitting flat on setpoint, green as grass.

Second — under restriction. Now ice the coil, the way that building iced it. The coil sits upstream of the fan, and ice is a restriction: dragging the same air through a choking coil costs more fan than it used to. But the loop doesn't know about coils — it knows the sensor is reading low-of- setpoint by a hair, and it does its job: speeds the fan up until static is back on the line. Same flow. Same static. More speed. The loop actively hides the failure, and the mask is deepest at low flow, where friction is cheap and the extra effort rounds to nothing — exactly the hours when the coil is most at risk. You will not find this failure on the static trend, because the static trend is the thing the loop is defending. You find it where that building's own trend finally showed it: on suction pressure, or on fan speed — Hz at a given flow is the honest number, and Hz creeping week over week at the same load is a restriction growing somewhere the sensor can't see.

Third — the old fix. Same building, earlier chapter of its story. The coil is starving for air, so someone reaches for the obvious lever: make more air — crank the fan. Except the boxes are pressure-independent, and you know what that means now: every box is holding a flow setpoint, and when trunk pressure rises they pinch. The flow barely moves — the zones already set it — and all that extra fan becomes pressure with nowhere to go. On the trend, duct static climbs, then flatlines at exactly 2.5 in. w.c. and sits there, hour after hour, suspiciously smooth. A flatline at a round number near the top of a sensor's range isn't a measurement — it's the sensor railed at full scale, a 0–2.5 in. w.c. transducer wired 0–10 V, pegged at 10 V, reporting only that the truth is somewhere above it. (That volts-to-inches arithmetic is exactly what the signal-scaling tool does; it's worth being fast at, because a railed signal is invisible unless you know where the range ends.) The real static was well past the reading, past where a high-static cutout should have tripped — and there wasn't one. More fan never was the answer: the fan makes pressure, the boxes make flow, and no amount of pressure changes what the zones asked for.

When the Loop Can't Save You

Everything above trusts one small pressure signal absolutely, which is why the failures worth knowing are all versions of that signal going wrong — and why the protections worth having are the ones that don't depend on it.

The high-static cutout. The six-o'clock scene with no working loop — a fan at speed, a duct closing like a fist — ends with pressure the ductwork wasn't built for: seams boom, flex pops off its collars, access doors bow. The answer is a separate mechanical pressure switch, set well above setpoint — up at the 4 in. w.c. class of number on a system holding 1.5 in. w.c. — hard-wired to stop the fan, usually with a manual reset so a human has to come find out why. Separate is the load-bearing word. It cannot be logic on the transducer, because the moments it exists for are exactly the moments the transducer's number is a lie: a railed sensor, a plugged tube, a crashed controller. The building in the story above ran with no cutout at all — the same uninsured posture as its missing low-pressure cutout, one story earlier. Hardware answers when software's number is gone.

The sensing tube. The transducer is only as good as the plastic tube feeding it. Tubes get pinched by a panel, plugged by dust, kinked at a fitting, and — on tubes run through cold spots — blocked by frozen condensation. A tube stuck reading low is the dangerous direction: the loop sees a starving duct, drives the fan toward 100%, and the real static runs away high while the sensor swears it's low — that's the day the cutout earns its keep. Stuck high is the quiet twin: the fan idles, the far zones starve, and the complaint calls start at the end of the longest run. Pump control's drifted-transducer warning is this same paragraph with pipe in it — one small sensor, absolute trust, calibrate it and protect it.

The duct itself. A blown flex connection or an access door left open upstream of the sensor is a hole the loop cannot tell from demand: static can't build, so the fan pegs at 100% and stays there, conditioning a ceiling plenum. The tell is the same honest number as before — fan speed that no longer matches the load — plus zones downstream of the breach going quiet all at once. The loop has no opinion; it's holding a setpoint it can no longer reach, forever.

Drive the Fan, Watch the Duct

The whole page under one slider. Drag how many of the thirty boxes are calling and watch the loop answer: static held, fan speed following the building down. Flip the setpoint to reset and watch the same evening get cheaper. Then break it the two ways the story broke: ice the coil and watch the loop hide it — same flow, same static, more speed, with the honest number sitting in the second strip — and finally try the last preset, which replays an earlier fix attempt on a pair of air handlers you've already met.

What the widget holds still: loop dynamics (speeds land instantly — the PID tuner has the hunting and the tuning), trim & respond's slow walk (reset shows where the setpoint lands, not the minutes of trimming), and the box side (thirty boxes compressed into one demand slider — their own story is the last page). The high-static cutout isn't modeled either; the last preset is what its absence looks like. Statics follow the units toggle — the building-pressure posture — while the transducer's 0–2.5 in. w.c. range and the fan's Hz stay in their IP field frames.

The Whole Path, One More Time

That closes the six-page chapter. Walk it once more, end to end, the way the air does. Page one built the path — return grille, mixing box, filter, coils, fan — and called itself the map. Page two gave the mixing box its one good trick: when the outside air is better than the return, open the dampers and cool for free. Page three followed the air you threw away and made the envelope balance its ledger. Page four taught you to name the box you're standing in front of, whatever the nameplate mumbles. Page five went to the far end of the supply duct and found thirty throttles, each answering its own room and nobody minding the trunk. And this page put the answer at the head of the duct: a fan that holds one pressure, two-thirds of the way out, and by holding it lets thirty boxes take whatever they need. Loads throttle; the mover responds — drawn on water once, and now drawn on air.

With the chapter in hand, a handful of tools stop being black boxes. The airflow calculator is a flow ring with the sheet metal off. The affinity-laws tool is this page's savings argument with real numbers in it. The PID tuner's supply-fan scene is the loop you just drove, hunting and all. So here's the field move this chapter has been building toward: next time you stand in front of an air handler, read it the way these six pages did — walk the stations, name the unit, find the dampers' minimum, ask where the relief goes, follow the trunk to a box, and put your cursor on two numbers before you believe anything anyone tells you: fan speed, and what the flow actually is. Static will be sitting on setpoint. It usually is.