Hydronic system balancing field guide for HVAC

Water balancing, and what the report delivers

Hydronic balancing is the work of setting the water flow through every coil and circuit in a chilled or hot water system to the flow the design called for, measured in gallons per minute. Set the flow right and each zone gets the heating or cooling it was sized for. The deliverable is not a system that feels balanced. It is a report, the water side of the test, adjust, and balance package, that lists design GPM against measured GPM for every coil and circuit, the pump data, the valve settings, and the deficiencies the tech could not fix.

The number that matters is the flow at the coil, not the pressure on the gauge at the pump. A system can show plenty of head and still starve the coil at the end of the run, because the water took the path of least resistance through the near circuits and never reached the far one. Balancing is how you make the water go where the design put it.

What separates a real water balance from a clipboard exercise is that the numbers reconcile. The sum of the circuit flows should track the pump total, and the coil delta-T should match the flow it is passing. When those do not agree, you have a bypass, a fouled coil, trapped air, or a measurement error, and the tech who signs the report owns finding out which.

Why is flow, not pressure, the balancing target?

Flow is the target because heat transfer rides on the mass of water moving through the coil, not on the pressure pushing it. The governing relationship in the trade is simple: BTU per hour equals about 500 times the flow in GPM times the temperature drop across the coil in degrees Fahrenheit. The 500 is the rounded product of water's density of about 8.33 pounds per gallon, the 60 minutes in an hour, and its specific heat of 1.0 BTU per pound per degree, all at around 60°F. Move the design GPM through a clean coil and it makes its capacity. Move too little and the zone runs short.

Pressure is a means, not the goal. The pump makes head, the head pushes flow against the system resistance, and the flow does the work. Two coils on the same loop can see the same differential pressure and pass wildly different flows, because their valve settings and circuit resistances differ. That is exactly the problem balancing fixes. You set each circuit's resistance with its balancing valve so the flow lands at design, then the pump carries the total.

Chase pressure instead of flow and you can convince yourself a system is fine while the far coil starves. The gauge at the pump does not know where the water went.

How do you measure water flow at a coil?

You read the pressure drop across a device of known resistance and convert it to flow. The everyday method on a balanced system is the manual balancing valve with two pressure and temperature ports, called PT ports, one upstream of the valve seat and one down. You connect a differential pressure gauge or manometer across the ports, read the drop, and convert it to GPM using the valve's flow chart for the position the handle is set to. The math behind the chart is the valve coefficient: flow equals the Cv times the square root of the pressure drop in psi, with the Cv fixed by the valve position. A fixed-orifice balancing valve keeps the metering orifice constant as you read, so the pressure drop maps cleanly to flow.

Where a balancing valve is not in the line, you read flow off a permanent flow station, a venturi, or an orifice plate, each a calibrated restriction with its own pressure-drop-to-flow curve. An ultrasonic flow meter clamps onto the outside of the pipe and reads flow from the transit time of sound through the moving water, which is the tool of choice when there is no installed device or you need to check one. A calibrated DP gauge with the right valve curve is the workhorse. Get the curve wrong, or read the wrong valve position, and the flow you write down is fiction.

Take the temperature off the same PT ports while you are there. The pressure gives you flow and the temperature pair gives you delta-T, and the two together tell you whether the coil is actually doing its job.

PT ports

The pressure and temperature ports on a balancing valve, read with a DP gauge for flow and a probe for water temperature

Cv

The valve flow coefficient; flow equals Cv times the square root of the pressure drop in psi at a given valve position

Flow station

A permanent venturi or orifice with a calibrated pressure-drop-to-flow curve, used where a balancing valve is not in the line

The balancing valve and the circuit setter

The manual balancing valve, known in the field by trade names like the circuit setter, is a calibrated valve you throttle to set a circuit's flow and then lock. You adjust the handle to the position that gives design GPM, confirm it by reading the pressure drop across the PT ports against the valve chart, then set the memory stop. The memory stop is a mechanical limit that lets the valve be closed for service and reopened to the same position without rebalancing. It is the feature that makes the setting survive the life of the building.

Two families show up. The fixed-orifice valve meters across a constant orifice, so the pressure-drop-to-flow relationship holds at any handle position. The variable-orifice valve changes its internal geometry as you throttle, so its flow chart is keyed to both the position and the pressure drop. Both are valid. Read the chart that matches the valve in your hand, not the one from the last job.

A balancing valve does three jobs at once: it measures flow through its ports, it sets flow by its position, and it isolates the circuit for service. Skip the memory stop and the next person who closes it for a coil pull undoes your balance and never knows.

What is a PICV and how does it change the job?

A pressure independent control valve, or PICV, holds a set flow regardless of the pressure swinging across it, and it changes the balancing job from setting every circuit by hand to commissioning valves that balance themselves. Inside one body it combines the control valve that modulates to the load with a pressure-regulating cartridge that caps the flow at a preset maximum. As other valves on the loop open and close and the differential pressure wanders, the cartridge absorbs the change and the flow through the coil stays put.

On a PICV system you do not proportionally balance the terminals the old way. Each valve arrives preset, or you set its maximum flow in the field, and because it is pressure independent the setting holds without chasing the neighbors. The balancing work moves up a level: verifying the valves were set right, confirming the differential pressure across each one stays inside the cartridge's control range, and setting the pump to deliver the total at a head that keeps the worst valve in control.

The trap is assuming a PICV means no balancing at all. It does not. Starve the valve below its minimum control differential and it can no longer hold its flow, so you still verify the pump head and the pressure available at the index valve. The valve is only independent inside its working range.

Does proportional balancing apply to water?

Yes. Proportional balancing on the water side runs on the same logic as the air side: you set every circuit as a ratio to each other first, then bring the whole system to design with the pump. You do not chase each coil to its exact GPM one at a time, because throttling one circuit shifts flow to all the others and you would chase your tail until the pump bearings give out.

The reference is the index circuit, the one reading the lowest percentage of its design flow, almost always the circuit hydraulically farthest from the pump. You leave the index balancing valve wide open and never throttle it. Then you work back toward the pump, throttling each over-supplied circuit until it reads the same percentage of design as the index. As you close the near valves, water shifts outward and the index reading climbs while you work. When every circuit on a branch reads the same fraction of design, the branch is proportional.

Then you set the pump to bring the index circuit to 100 percent, and because everything is proportional to it, every circuit lands at design together. Change the pump speed later and they all move by the same fraction, so the balance holds. This is the same index-and-proportion method the air-balancing guide walks through for diffusers, applied to coils and risers instead of branches and outlets.

The pump, the system curve, and the affinity laws

With the circuits proportional, you set the pump to deliver the total flow at the head the system actually needs, and the pump affinity laws tell you how flow, head, and power move when you change speed or trim the impeller. Flow changes in direct proportion to speed or impeller diameter. Head changes with the square of that change. Power, and the motor amps that ride with it, change with the cube. The cube is the one that bites. A 10 percent speed cut to shed excess flow drops the power draw by nearly 30 percent, which is where the savings of a properly balanced variable-speed pump come from.

The pump operates where its curve crosses the system curve. Balancing reshapes the system curve, so after the circuits are set you trim the pump to land on the design flow. On a constant-speed pump you trim the impeller to a smaller diameter or throttle the discharge valve to add head loss. On a pump with a variable frequency drive you set the speed, which is the cleaner fix because it cuts power on the cube instead of burning it across a throttle valve.

Check the motor amps every time. Clamp the motor leg and compare it to the nameplate full-load amps. A pump running past its rating overheats and fails early, and an overamped pump is a finding for the report, not a number you accept because the flow looked right. The affinity laws are accurate for speed changes; for impeller trims they get you close, because cutting the impeller also nudges the efficiency, so verify the flow after the trim rather than trusting the calculation alone.

System curve

The plot of head the piping demands against flow; balancing reshapes it, and the pump runs where its curve crosses it

Affinity laws

Flow scales with speed or impeller diameter, head with the square, and power with the cube of the change

What is low delta-T syndrome?

Low delta-T syndrome is the chronic chilled water plant problem where the temperature difference between return and supply water comes back smaller than design, so the plant has to pump far more water than it should to move the same cooling. Design a chilled water system for a 12°F to 16°F rise and watch the return come back only 6°F or 7°F warmer than supply, and you are in it. The plant is moving the gallons but not picking up the heat.

The cost lands on the central plant, not always the space. Low delta-T forces the secondary pumps to push more flow for the same load, drives the bypass open on a primary-secondary system, and can start an extra chiller that the running machines could have covered if the water came back warm. The plant burns pump and compressor energy to do work the design said it would not have to.

The causes are a short list. Coils overpumped because nobody balanced the flow down to design. Three-way valves and bypasses that let supply water short-circuit to the return without crossing a coil. Fouled coils and air-bound coils that cannot transfer their heat. Control valves that cannot throttle, set wide open as a field fix for a comfort complaint. Balancing the coils to design flow is the first and cheapest line of defense, because an overpumped coil is the most common cause and the one a balance directly corrects.

Air in the system ruins flow and heat transfer

Air trapped in a hydronic system wrecks two things at once: the heat transfer at the coil and the flow reading at the valve. Air is an insulator, so an air-bound coil transfers less heat for the same water flow, which reads as a low delta-T you will chase for hours if you do not know the air is there. Air pockets also block flow and put bubbles across the PT ports, so the differential pressure you read is noise and the GPM you calculate is wrong.

Get the air out before you balance, not after. A closed hydronic system carries an air separator, usually placed at the hottest, lowest-pressure point in the loop where air comes out of solution most readily, and it coalesces the small bubbles into large ones that rise and vent. Air vents at the high points and at each coil let trapped air out. The fill and purge sequence runs water through the system at a velocity high enough to sweep entrained air to the separator, commonly above about 2 feet per second, then bleeds it off.

A system that keeps making air after it is purged has a real problem: a vent sucking air on the suction side of the pump, a make-up valve cycling, or a leak that draws air when the loop goes to vacuum. Find it. You cannot balance a loop that is swallowing air, because every reading moves under you.

Strainers and the startup flush

Balance a dirty system and you balance it twice, because the debris that fouls a strainer or a coil moves every reading you just set. New piping carries cutting oil, pipe dope, weld slag, and construction trash, and all of it heads for the first restriction, which is a control valve, a balancing valve orifice, or a coil. The system gets flushed before it gets balanced, with the startup strainers or temporary bags in place to catch the trash, and the strainers cleaned until they stop loading.

The sequence matters. Flush the loop, pull and clean the startup strainers, confirm they come out clean twice in a row, then balance. Balance before the flush and the first strainer cleaning afterward changes the system resistance and throws your settings off. A clogged strainer ahead of a coil reads exactly like an undersized circuit: low flow that no amount of valve opening will fix, because the restriction is the screen, not the valve.

This is the same discipline the chilled water side shares with the flush and fill on any new closed loop. Clean water, clean strainers, then set the flow. The trade learned this the expensive way, on coils that plugged a month after a balance that read perfect.

Two-way, three-way, and variable versus constant flow

The control valve type at the coil decides whether the system flow is variable or constant, and that changes how you balance and what the plant sees. A two-way valve throttles flow to the coil as the load drops, so the system flow falls with the load. That is a variable-flow system, and it is what a variable-speed pump and a modern chilled water plant want, because less load means less flow means less pumping energy.

A three-way valve keeps the coil flow roughly constant by diverting water around the coil through a bypass when the load drops. The coil sees variable flow but the circuit sees constant flow, so the plant pumps full flow regardless of load. Three-way valves are a classic cause of low delta-T, because the bypassed water returns to the plant at supply temperature and dilutes the return. On a balancing job, the bypass leg gets balanced too, so the diverted flow matches the coil flow it replaces.

The variable-primary-flow plant pushes variable flow all the way to the chillers, with a minimum-flow bypass to protect the machines. Balancing that plant means setting the coil flows, confirming the minimum-flow bypass holds the chiller's minimum, and making sure the control valves can close down without starving the worst circuit. Know which plant you are on before you set a single valve, because the right balance on one is the wrong balance on the other.

Confirming the coil makes its capacity

A coil at design flow is not proof the coil is doing its job. You confirm capacity with three numbers together: the flow through the coil, the temperature drop across it, and the entering and leaving water temperatures against design. Run BTU per hour equals 500 times GPM times delta-T and compare it to the coil's rated capacity. A chilled water coil at design GPM with a design delta-T is making its tons. At design GPM with a low delta-T, it is overpumped or fouled and not transferring what it should.

The flow per ton is the quick gut check. At a 10°F chilled water delta-T, a ton of cooling needs about 2.4 GPM, because a ton is 12,000 BTU per hour and 12,000 divided by 500 times 10 is 2.4. Design for a wider delta-T and the flow per ton drops: at 12°F it is about 2.0 GPM per ton, at 16°F about 1.5. That relationship is why the high-delta-T design saves pumping energy, and why accepting a low delta-T quietly throws it away.

Read the water temperatures off the PT ports at the coil with the coil under load, not at rest. A coil that will not make its delta-T at design flow is telling you something: it is fouled, it is air-bound, the air side is not moving its design CFM, or the entering water is warmer than design because the plant itself is struggling. The coil is the place the whole system's health shows up in two thermometer readings.

Reverse-return, direct-return, and the self-balancing myth

Reverse-return piping is often sold as self-balancing, and it is not, quite. In a direct-return loop the first coil off the supply is also the first back to the return, so it sees the shortest total path and the least resistance, and it hogs flow while the far coil starves. Reverse-return flips the return so the first coil supplied is the last returned, which makes every coil's total path length roughly equal, so the circuits come out closer to balanced on their own.

Closer is not balanced. Reverse-return equalizes the piping length, but it does nothing about coils of different sizes, valves of different resistances, or fittings that vary circuit to circuit. The flows will be in the same neighborhood, which beats direct-return, but a job that needs its design flows still gets balancing valves and still gets balanced. Treat reverse-return as a head start, not a finished job.

The self-balancing claim costs money when a designer leaves the balancing valves out to save first cost on a reverse-return system, and then the flows still come in uneven with nothing in the line to set them. The piping geometry helps. It does not replace a valve you can read and lock.

How does glycol change the flow setting?

Glycol changes the flow setting because it does not carry heat as well as water, so you have to move more of it to do the same job. A glycol mix has a lower specific heat and a higher density and viscosity than plain water, and the combination means the 500 factor in the heat-transfer equation no longer holds. The trade calls it the double penalty: the fluid holds less heat per pound and it pumps harder.

The practical effect is a correction factor on the flow. To move the same BTU per hour, a glycol system needs more GPM than a water system, and the increase grows with the glycol percentage and varies with temperature. A common working number is that a 30 percent glycol mix uses roughly 450 in place of 500 in the heat-transfer equation, so the design flow climbs accordingly. Get the correction factor from the fluid manufacturer's data for the actual mix and temperature, because the published factor is specific to the product and the concentration.

Two things follow on a balancing job. The design GPM you balance to should already have the glycol correction baked in if the engineer did it right, so verify the design basis before you assume. And the higher viscosity raises the pressure drop across every valve and coil, so the valve charts and the pump head both shift. Balance a glycol loop against water-based numbers and every flow you set is wrong in the same direction.

What goes in the water-side TAB report?

The water-side TAB report lists design GPM against measured GPM for every coil and circuit, plus the pump data, the as-left valve settings, the coil temperatures, and the deficiencies. For each pump it shows the nameplate data, the design and measured flow, the head, the impeller size or VFD speed, and the motor amps and voltage. For each coil and circuit it shows the design GPM, the measured GPM, the percentage of design, the balancing valve setting, and the entering and leaving water temperatures with the delta-T. The deficiency list captures what could not be balanced and why.

Format follows the standard the spec names. NEBB and AABC both publish hydronic procedural standards and report forms, and a project specified to one of them expects that body's forms filled out completely, with the firm's certification and the supervising professional's stamp where required. The report is signed and certified by the balancing firm, which is how the engineer and the building department know a qualified agency stands behind the numbers.

The engineer of record reviews it. A water report that shows every coil dead on design with no deficiencies on a real building earns a hard second look, because a balance with zero findings is usually a balance that was written, not performed. The credible report names the coils that would not make flow, the circuits stuck on low delta-T, and the pump that needed an impeller it does not have. Those are the items the engineer rules on before the system is accepted.

Chilled water in the data center

In a data center the chilled water balance is where plant efficiency lives or dies, because the cooling runs every hour of every day and the plant delta-T is the lever on the power bill. The CRAH units, computer room air handlers, run chilled water coils that pull heat off the server aisles, and on a variable-flow plant their two-way valves modulate flow to the IT load. Set those coil flows right and the return water comes back warm, the chillers run loaded and efficient, and the pumps ride down with the load.

Low delta-T hits a data center harder than an office because the load never stops. An overpumped CRAH coil or a bypass leaking supply into the return drags the plant delta-T down around the clock, forcing extra pumping and an extra chiller to cover a load the warm-return plant could have carried. The balance sets the coil flows to design, confirms the two-way valves can throttle without starving the far units, and verifies the delta-T at the coils and back at the plant.

This is the water-side companion to the airflow and containment work in the cooling pillar. The aisle containment and the CRAH airflow set how well the air picks up heat from the servers; the water balance sets how well the coils hand that heat to the plant. Both have to be right, because a clean airside balance cannot save a plant leaking its delta-T through unbalanced coils.

Field example: setting one chilled water coil

A chilled water cooling coil on an air handler is designed for 80 GPM at a 12°F rise, 44°F entering and 56°F leaving, which is 480,000 BTU per hour, or 40 tons, about 2.0 GPM per ton. The balancing valve has a memory stop and PT ports. The job is to confirm and set the flow, and to prove the coil makes its delta-T.

As found, the balancing valve was wide open and the coil was passing 105 GPM, 131 percent of design. The flow looked generous, which is exactly the trap. Read the water temperatures off the PT ports under load and the leaving water is only 53°F, a 9°F drop instead of 12°F. The coil is carrying close to its load, but the extra 25 GPM is doing nothing except dragging the return temperature down and the plant delta-T with it.

Throttle the balancing valve to 80 GPM, confirmed by reading the pressure drop across the PT ports against the valve chart for the new handle position, and set the memory stop. With the flow at design and the same load, the leaving water climbs back to 56°F and the delta-T recovers to 12°F. The coil still makes its 40 tons. The 25 GPM you gave back is pump energy the plant was spending for nothing, and a return temperature the chillers can actually use.

What to document

The balance is the report, so the record is the work. For every coil, circuit, and pump, capture enough that the engineer can reconcile the numbers and the next tech can pick up the baseline years later.

Capture the coil or circuit identifier, the design GPM, the measured GPM, the percentage of design, the balancing valve setting as left, the entering and leaving water temperatures and the delta-T, and the pass or fail against the spec tolerance. For each pump add the head, the impeller size or VFD speed, the motor amps and voltage, and the design versus measured flow. Note the fluid and glycol percentage if any, because every flow target depends on it, along with the instrument calibration dates. The table below is the core circuit record; the pump data rides alongside it.

Common mistakes

• Balancing the coils before the system is flushed and the startup strainers are clean, so the first cleaning afterward moves every setting.
• Reading flow through bubbles because the loop was never properly purged of air.
• Accepting a low delta-T at the coil instead of chasing the overpumped flow, the open bypass, or the fouled coil behind it.
• Setting the pump by the discharge gauge instead of trimming the impeller or the VFD to land the measured total flow.
• Using water-based flow targets and valve charts on a glycol loop, so every set flow comes in short.
• Throttling a circuit without setting the memory stop, so the next service close undoes the balance.
• Reading the wrong valve's flow chart, or the wrong handle position, and writing down a flow that was never there.
• Treating reverse-return piping as self-balancing and leaving out the valves that would have let you set the flows.

Field checklist

Standards and references

The hydronic TAB procedural standards and the certifications behind them come from NEBB, the National Environmental Balancing Bureau, and AABC, the Associated Air Balance Council. Each publishes a procedural standard and report forms covering water balancing, and a project specifies one of them; the named standard governs the procedure, the forms, and the certification on the report. Confirm which one the spec calls out before you start, and use that body's current edition.

ASHRAE Standard 111, the practices for measurement, testing, adjusting, and balancing of building HVAC systems, covers the measurement methods on the water side as well as the air side. ASHRAE Standard 90.1 sets the energy requirements a hydronic system has to live within, and it is the reason variable-flow pumping and proportional balancing are written into so many specs. Above the standards sits the manufacturer's valve data, the Cv and flow charts that set the targets you read against, and the project specification and engineer of record, who set the tolerance and accept the report.

Cite the body that owns the point, and verify the edition and section, because these documents revise on their own cycles. The flow targets themselves come from the design, not from any standard's rule of thumb, and the contract documents control when they are tighter than common practice.

Units, terms, and conversions

Hydronic balancing carries its own vocabulary and a couple of unit systems, so the same quantity reads differently across a balance report, a valve catalog, and a metric drawing.

Flow is GPM, gallons per minute, in the field, and liters per second or cubic meters per hour in metric sources, where 1 GPM is about 0.0631 liters per second. The temperature difference across a coil is delta-T, in degrees Fahrenheit or Celsius. Capacity is BTU per hour or tons of refrigeration, where 1 ton is 12,000 BTU per hour. Head is in feet of water or kPa, and the pressure drop across a balancing valve is read in psi or kPa against the valve's flow chart. Heat transfer ties them together: BTU per hour equals about 500 times GPM times delta-T for plain water, with the 500 corrected down for a glycol mix.

GPM

Gallons per minute, the water flow rate a balance sets at every coil and circuit to match design

Delta-T

The temperature difference between water entering and leaving a coil; the chilled water value the plant's efficiency rides on

Balancing valve

A calibrated, lockable valve that measures flow through its PT ports, sets it by position, and isolates the circuit

PICV

Pressure independent control valve, which holds a preset flow regardless of differential pressure within its control range

Cv

The valve flow coefficient; flow equals Cv times the square root of the pressure drop in psi at a given valve position

Proportional balancing

Setting circuits as a ratio to the index (lowest-percent) circuit, then scaling the whole system with the pump

Low delta-T

A return-to-supply temperature difference smaller than design, forcing excess flow and wasting plant capacity

Related tools

Calculators and readiness checks for this work

Compare your options

FAQ

People also ask

• What is air balancing?
• How do I measure total external static pressure?
• What is low delta-T syndrome?
• What is a hydronic balancing valve?
• What is primary-secondary chilled water pumping?