Why Balancing Is Not Optional

Why this matters

Balancing gets treated as a finishing touch, the thing you do if the customer complains. It is not a finishing touch, it is the step that turns a correctly sized machine and correctly sized branches into a system that actually delivers what was designed. Skip it and you get the specific failure that is hardest to sell a fix for later: the equipment is right, the total flow is right, every measurement at the appliance passes, and half the rooms are wrong. The shop then spends years of warranty visits on comfort complaints that no component replacement can touch.

Before you change a damper or a blower tap

A blower wheel that starts while your hands are in the cabinet is the hazard, and the action is to open and lock the unit disconnect before your hands cross the plane of the wheel, never to rely on the door interlock. Defeating an interlock so the unit runs with the panel off is the specific practice that causes these injuries.

Changing a speed tap is electrical work, so lock and tag the disconnect and prove dead at the terminals under 29 CFR 1910.333(b)(2), with the live-dead-live sequence from NFPA 70E-2021, 120.5. A run capacitor holds a charge after the disconnect opens: discharge it with a suitable resistive tool and confirm zero volts across its terminals before you touch them, because proving the supply dead does not prove the capacitor dead.

Drilling static-pressure test holes puts a bit into a space you cannot see. Check behind the panel for wiring, refrigerant lines and the coil face before you drill, deburr the hole, and plug it afterwards with a proper plug rather than tape.

Cutting or drilling lined plenum and duct board is an inhalation exposure, not a contact one, so wear at least a fitted filtering facepiece with eye protection; where the employer requires respiratory protection, the program at 29 CFR 1910.134 applies, including fit testing before first use and at least annually under 1910.134(f)(2).

On a fuel-burning appliance, an airflow change is a combustion change. After any balance work on a furnace, confirm the appliance is not cycling on its high limit and that the vent connections are intact before you leave; combustion analysis and carbon monoxide testing belong to the combustion cards and should be run by someone equipped to do them.

What balancing actually is

Balancing is the deliberate addition of resistance to the easy paths so that all of them become equally hard, at which point one machine setting delivers the right flow to every terminal at once.

That is worth stating plainly because it sounds wrong. You are intentionally throwing away pressure in the branches that were performing best. There is no way around it in a system where one machine feeds parallel paths of unequal resistance, because every path shares the same pressure difference and takes flow according to its own resistance. The only lever you have that affects one branch without affecting the machine is that branch's own resistance.

The alternative would be a system where every branch has identical resistance by construction. That is what a well designed layout tries to approximate, and it never lands exactly, because runs have different lengths, different numbers of bends, and different terminals.

The two intuitive alternatives, and what each costs

Turn the machine up. This lifts every branch, including the ones already over-flowing, and it lifts them roughly in proportion to what they already had. The imbalance ratio does not change. What does change is velocity everywhere, which raises noise at the grilles that were already loudest, raises the pressure the machine works against, and raises power draw. On a centrifugal machine, power climbs steeply with speed: at a fixed system, flow tracks speed while power tracks roughly the cube of it, so a 10 percent speed increase costs on the order of a third more power for a 10 percent flow gain that lands mostly in the wrong rooms.

Work the complaining room. Larger register, longer flex, a booster. This can help the one room and it changes the shared pressure difference for every other room, so the fix propagates as a new complaint somewhere else. It is also the move that hides the original defect, because the record now shows a non-standard branch and nobody knows why.

Neither of these is stupid. Both are what you reach for if you think of branches as independent. They fail for the same reason: branches are not independent.

The order of operations, and why it is that order

Proportional balancing runs in a fixed sequence, and each step exists because doing it later invalidates the earlier work.

1. Get the system to a stable, representative condition first. Clean or new filter, all terminals open, dampers full open, the machine at its intended setting. Balancing against a half-loaded filter produces a balance that is wrong the day the filter is changed.
2. Measure every terminal and compute each one's ratio of measured to design flow. The ratio, not the raw number, is what you balance on.
3. Find the index run: the terminal with the lowest ratio. That branch is the hardest path in the system and it sets the ceiling for everything else. Its damper stays full open for the whole procedure.
4. Bring the other terminals down to the index ratio, highest ratio first. Working highest first minimizes the number of passes, because each closure raises the ratio of everything else and the biggest offender moves the others the most.
5. Re-measure everything after each adjustment. Coupled branches mean a single-branch check after a single-branch change proves nothing.
6. Only when every ratio matches, trim the machine to lift the whole set to 1.0 together.

Skipping step 3 is the common error. If you balance to design numbers one branch at a time instead of to a common ratio, you chase your own tail, because every branch you set moves every branch you already set.

The record, filled in

This is the artifact the job produces. Design flows come from the load calculation, not from the register size.

Reading the as-found column. Total flow was 640 against a design 600, 7 percent high, and every measurement at the appliance would have passed. Bedroom 1 was 21 percent below its design flow and Living was 30 percent above. The total was the only number in that column that looked acceptable, and it was the least informative one.

How the balance ran. Bedroom 1 at 0.79 is the index run and its damper stayed open throughout. Living, the highest ratio, was closed down first; that pushed flow into the other three and raised their ratios, which is expected and is why the next measurement is a full set rather than one terminal. Kitchen followed. After two passes all four ratios sat close together and total flow had fallen, because the system now carried more resistance than it did wide open. The blower then went up one tap to lift the whole set together, landing at 611 total, 1.02 of design, with every terminal within about 2 percent of its own design flow.

What it cost. Total external static rose from a measured 0.62 to 0.78 in w.c. on the same manometer and the same test-hole locations, and blower current rose from 4.1 A to 4.6 A on the same clamp meter, about 12 percent. That current increase is real and it is the price of the correct answer: the alternative is a system that draws less while delivering 21 percent short to one bedroom and 30 percent long to a living room.

The failure mode if you stop early. A tech who sets Living and Kitchen down to their design numbers and leaves, without lifting the blower, ends up with a system whose total has dropped below where it started. That matters because the appliance has its own minimum airflow requirement, and dropping under it is not a comfort issue, it is a coil-icing issue on cooling and a limit-cycling issue on gas heat.

What the record proves that a total never can

A total flow reading confirms the machine is inside its own operating window. It says nothing about distribution, and the two failure modes it hides run in opposite directions.

• Correct total, wrong split. Every appliance-side check passes. The complaint is comfort, it is chronic, and it survives every component replacement anyone tries.
• Correct split, wrong total. Every room is proportionally right and all of them are short, which reads as "the system just cannot keep up on design days" and gets sold as an undersized appliance when it is a distribution restriction.

Only a per-terminal record separates them, and only a dated per-terminal record lets the next tech tell drift from a bad original balance.

When the balance will not hold

• A system with modulating flow. Where terminals throttle themselves, a fixed-damper balance is set at one operating point and the machine's control has to hold pressure across the range. Balance at the design condition and then confirm at a low-flow condition.
• Occupants closing registers. Every closed register raises the pressure difference and pushes flow to the rest. A balance report is a snapshot of a configuration, so note on it that the terminals were all open.
• Filter loading. A filter is a rising resistance in series with the whole system, so it moves the total without changing the split much. That behaviour is its own subject and the filter card in this group carries it.
• Anything that changes a branch. A remodel, a rerouted run, a removed restrictor. A branch whose resistance changed voids the balance for every branch, not just itself.

How to verify you got this right

Re-measure every terminal after the machine trim, not just the ones you adjusted, and confirm the ratios are still together. Ratios that drifted apart during the trim mean a damper is not holding its position.

Confirm total external static against the appliance manufacturer's published maximum, using the same test holes and the same instrument reference you used before. A balance that lands inside design flow but past the machine's static limit is not finished.

Record design flow, as-found flow, final flow and damper position for every terminal, plus the filter condition and the machine setting at the time. A balance record without the filter condition and the machine setting cannot be reproduced, and a balance nobody can reproduce is an opinion.

References

• 29 CFR 1910.333(b)(2) for electrical work at a blower tap or disconnect, with NFPA 70E-2021, 120.5 for the live-dead-live proving sequence
• 29 CFR 1910.134, respiratory protection, including fit testing under 1910.134(f)(2), when cutting lined duct or plenum
• ACCA and ASHRAE guidance on load-based design flows and proportional balancing procedure
• Manufacturer documentation for minimum airflow, maximum external static and blower speed tap data on the specific appliance
• See related: How a Branch Steals From the Branch Next to It; What Static Pressure Tells You About a Duct System; How a Filter Changes a System as It Loads