Hot Water Recirculation Pump Sizing Calculator — Heat Loss Flow and the 2 ft/s Ceiling

Recirculation Flow From Heat Loss

The flow in a recirculation loop is not a demand figure. It is the flow that carries away enough heat to keep the water at temperature while nobody is drawing anything.

The relationship is the same one used for hydronic heating, applied backwards. In heating you compute the flow needed to deliver a quantity of heat into a space. Here you compute the flow needed to replace a quantity of heat lost from a pipe. The arithmetic does not know the difference.

Flow in gallons per minute equals the system heat loss in BTU per hour divided by 500 times the design temperature drop in degrees Fahrenheit. A published ASPE presentation gives it in exactly that form, and it is described elsewhere as the ASPE verified formula. At a 10 degree drop, which is the common design assumption, it reduces to the heat loss divided by 5000.

The 500 is the product of three physical quantities: 8.33 pounds per gallon, 60 minutes per hour, and the specific heat of water at 1 BTU per pound per degree Fahrenheit. Multiplied out it is 499.8, rounded to 500 by convention.

Which heat loss belongs in that calculation depends on what the design has to achieve, which is why this calculator asks before it computes.

Where the criterion is the temperature arriving at the farthest fixture, published guidance uses the supply piping heat loss to that fixture. The return leg is downstream of the fixture, so its losses do not affect what arrives there.

Where the criterion is a minimum temperature held everywhere in the loop, or a minimum return temperature, the return losses are included and the temperature drop is reconsidered against that objective. A designer working to a return temperature target with a supply only heat loss will undersize the flow.

What does not enter the calculation at any point is the building's water demand. Not the fixture count, not the fixture units, not the peak flow, not the occupancy, not the number of storeys. A reader who has just sized the distribution system from fixture units has to set that work aside entirely: the two calculations share a building and nothing else.

This calculator does not compute the heat loss itself. A real figure comes from walking each pipe size and length with published heat loss per foot, accounting for insulation and the temperature difference to the surrounding space. The published ASPE example does exactly that, summing 8.90 BTU per hour per foot on 1,500 feet of 2 inch supply with 2 inch insulation, 7.59 on 1,000 feet of 1-1/2 inch, 7.25 on 500 feet of 1 inch and 6.33 on 100 feet of 3/4 inch return, for 25,198 BTU per hour in total. Because everything on the page derives from that one number, an invented figure would propagate into the flow, the velocity verdict and the pump selection at once, which is why the estimate route on this page returns a state rather than a number.

Why There Is No Static Head

This is the single most consequential thing to understand about recirculation pump sizing, and it is the opposite of what every other pump calculation on a project teaches.

A recirculation loop is a closed circuit. Hot water leaves the heater, travels up through the building in the supply, and comes back down through the return to the heater inlet. The pump does not lift water anywhere: the water is already in the pipe, and the column coming down the return pushes with exactly the force the column going up the supply resists.

The two cancel. What the pump has to overcome is friction, and nothing else.

A published ASPE presentation on domestic hot water recirculation states it in those words: the pump head is determined from the piping system's friction and pressure loss, no static head. The same document repeats it from the other direction, that static head should not be included because the calculation is for a closed system.

The numbers this produces look wrong to anyone used to booster or well pumps. A published residential worked example arrives at 1.9 gallons per minute at 5.2 feet of head. The commercial example in the ASPE presentation, a 3,100 foot loop moving 10.3 gallons per minute, comes out at 31.5 feet. Neither figure is small because the buildings are small; they are what friction alone produces over those lengths.

The error this prevents is expensive. Fifteen storeys is roughly 150 feet of static head that does not exist. A pump chosen for 150 feet instead of 6 is not slightly oversized; it is a different class of machine, and it will drive the loop at a velocity the copper cannot survive. The head term that was never real produces a pipe failure that is.

This calculator therefore has no building height input and no static lift input. That is a deliberate omission rather than an oversight: offering the field, even labelled as unused, invites the mistake the page exists to prevent.

One qualification keeps the statement exact. This describes the loop in operation. Filling a system, purging air and commissioning are separate operations with their own requirements, and the head figure here does not describe them.

The 2 ft/s Velocity Ceiling for Recirculation Lines

The velocity limit that applies to a recirculation line is far lower than the one that applies to the same copper elsewhere in the same building, and the reason is that the flow never stops.

The Copper Development Association gives the recommended maximum velocity for water in a copper tube system as 5 to 8 feet per second for cold water systems, 4 to 5 for hot water systems below 140 F, and 2 to 3 for hot water systems above 140 F.

Those figures describe hot and cold water service generally, where flow occurs in bursts as fixtures open and close. Code commentary on limitation of velocity goes further specifically for this case: it states that this velocity is too great for systems where the flow is continuous, as in the case of recirculated hot water piping, and that the continuous flow rate for hot water with modest chemical content should be limited to not more than 2 feet per second for such continuous systems.

An ASPE Pipeline article on corrosion in hot water recirculating systems puts it as an instruction: design all hot water circulating systems to keep velocities below 2 feet per second. The same article cites the Copper Tubing Institute at 2 to 3 feet per second for hot water recirculation pipe, and advises considering a lower velocity on half inch lines because of burrs or imperfections from faulty workmanship.

This calculator therefore uses 2 feet per second as its default, which is 0.61 metres per second. A ceiling of 3 feet per second is available as a deliberate override. That figure has its own sources rather than none: it is the top of the Copper Development Association range for hot water above 140 F, and a published forensic account of a recirculation failure recommends designing these systems so that temperatures generally do not exceed 140 F and flow velocity is no greater than 3 feet per second. What it is not is a continuous circulation figure, and the result says so when it is selected.

The 5 to 8 feet per second figure that most people associate with copper is never applied on this page. It belongs to cold water distribution, where the flow is intermittent and the tube spends most of its life at rest.

Exceeding the ceiling fails the result here rather than raising an advisory. A flow computed exactly right from a heat loss computed exactly right, pushed through a segment one size too small, is a design that will leak, and calling that a pass with a note attached would be the wrong message on the one page where oversizing is the danger.

Maximum Continuous Flow by Copper Pipe Size

The velocity ceiling has a practical form that is easier to work with than a velocity: the maximum flow each pipe size can carry continuously.

At 2 feet per second in Type L copper, using actual inside diameters rather than nominal sizes: a half inch line carries 1.45 gallons per minute, three quarter inch 3.02, one inch 5.14, one and a quarter inch 7.83, one and a half inch 11.09 and two inch 19.29. In litres per minute those are 5.5, 11.4, 19.5, 29.7, 42.0 and 73.0.

At the 3 feet per second override the same sizes carry 2.18, 4.53, 7.72, 11.75, 16.63 and 28.94 gallons per minute. Every figure is exactly half as much again, because the flow at a ceiling is directly proportional to the ceiling.

Read the other way, that table sizes the pipe. A loop needing 4 gallons per minute cannot use three quarter inch pipe at the 2 foot ceiling, because three quarter inch tops out at 3.02. One inch is the smallest size that holds it, and it runs at 1.56 feet per second with room to spare.

The bores behind those figures are 0.545, 0.785, 1.025, 1.265, 1.505 and 1.985 inches, which are 13.8, 19.9, 26.0, 32.1, 38.2 and 50.4 millimetres. They are not arbitrary: copper tube outside diameter is the nominal size plus one eighth of an inch, and the Type L wall of 0.040 to 0.070 inches comes off that twice. Velocity depends on the bore squared, so using a nominal 0.75 for three quarter inch instead of 0.785 reads about 10 percent low, and friction depends on the bore to a higher power still.

One point belongs to the next section and is worth flagging here. The size to check against this table is not automatically the return line. It is the narrowest pipe anywhere in the continuous recirculation path.

Controlling Segment Versus Return Pipe

Most guidance on recirculation velocity speaks in terms of the return line, and most of the time that is right. It is not always right, and when it is wrong the error is large.

The recirculation flow is the same through every segment of the continuous path. Water leaving the pump passes through the supply main, the branches that carry recirculation flow, and the return, and none of it disappears along the way. Velocity therefore depends only on the bore of each segment, and the segment that governs is the narrowest pipe carrying that flow.

In a purpose designed loop the return is usually the smallest pipe, which is why the shorthand works. But a supply main that necks down before the last fixture, with the recirculation flow continuing through it, governs instead. The ASPE presentation makes the same point from the design side, warning that smaller supply and return piping is assumed until the flow is determined and that it is not always three quarter inch.

The scale of the difference is not marginal. Take a loop carrying 4 gallons per minute. Through a three quarter inch return, bore 0.785 inches, it runs at 2.65 feet per second, which is over the 2 foot ceiling but inside a 3 foot one and looks tolerable. Through a half inch segment on the same loop, bore 0.545 inches, it runs at 5.50 feet per second, which is more than double the reported figure and outside every ceiling in circulation, including the 4 to 5 that applies to hot water below 140 F in ordinary intermittent service.

A calculator that asked for the return line would report 2.65, pass it against the looser ceiling, and say nothing about a segment eroding at 5.50.

This calculator therefore asks for the smallest pipe in the continuous recirculation path and says so in the field label rather than only in the helper text. A field called return pipe size gets filled with the return pipe size whatever the helper says.

Finding that segment is a drawing exercise rather than a calculation. Trace the path the recirculation flow takes from the pump discharge back to the pump suction, note every change of size along it, and use the smallest.

Oversized Recirculation Pump and Pinhole Leaks

On most pump pages an oversized pump wastes money. Here it destroys the piping, and it does so slowly enough that the connection is usually lost by the time the leaks appear.

The failure is erosion corrosion: a mechanically induced failure where fast moving water strips material from the inside of the tube. The Copper Development Association describes it as readily identifiable by the horseshoe shaped pitting throughout the inside of the tubes, which is its distinctive signature. Trade coverage describes the same damage as velocity channels, copper eroded away by high speed water.

Published guidance names high velocity from an oversized pump as the number one cause of pinhole leaks in recirculation pipes. The Copper Development Association names the same two causes from the design side: an undersized piping system or an oversized recirculating pump may cause high water velocity.

What makes it worse than a proportional penalty is that turbulence rises with the square of the flow. A pump delivering twice the required flow produces four times the turbulence. Three times the flow produces nine times. There is no conservative direction in oversizing here, only an accelerating one.

A published case shows what that costs. At a recently opened fifteen storey hotel near Lake Tahoe, a guest reported a small leak in a ceiling. Investigation found a pinhole in a copper pipe on the recirculating hot water system, then another, then another. By the time consulting engineers were involved, nearly every room showed water damage on walls or ceilings, and the recirculating system had to be repiped at a six figure cost, before counting lost revenue.

The remedy is named at source rather than inferred. The Copper Development Association states that installation of a smaller capacity pump, or a throttling bypass on the existing pump, should help in lowering the velocity of the water in the system. A larger pipe achieves the same thing from the other direction, and this calculator names both.

Water chemistry shifts the margin rather than the arithmetic. Chlorine in the supply is aggressive to piping as well as to bacteria, and chloramine, which combines chlorine with ammonia so the disinfectant lasts longer, stays corrosive for longer too. A system in aggressive water is designed below the ceiling rather than at it.

Pump Head and Duty Point

The head a recirculation pump must develop is the friction and pressure loss around the supply and return circuit. What goes into that figure, and how precise it is, are worth being clear about.

The starting point is the developed length of the circuit, measured from the pump to the farthest point and back. During design the actual route is often unknown, and a common allowance adds 50 to 100 percent to an estimated length to account for fittings. That allowance is scale dependent and it is worth saying so: in the published ASPE example the fittings add 29 feet to a 1,500 foot run and 22 feet to a 1,000 foot run, which is about 2 percent, not 50. The larger figures belong to a short circuit full of changes of direction, which is what a residential recirculation loop usually is.

The friction is then taken from an allowable uniform friction loss. Published design guidance gives 2 feet of head per 100 feet as a working value, with the author noting it can be any reasonable value that does not produce excessive velocities. A 200 foot circuit with a 50 percent fittings allowance is 300 equivalent feet, which at that value gives 6.0 feet of friction head, or 1.83 metres.

That number is a design allowance rather than a computed gradient, and the distinction matters. The designer chooses a uniform loss value and then selects pipe sizes that meet it. It is not a prediction of what a particular pipe will do at a particular flow: 4 gallons per minute through three quarter inch copper actually produces a Hazen-Williams gradient with C of 140 nearer 4.7 feet per 100 feet, more than twice the allowance. So the allowance is a target used to choose pipe size, not a post selection calculation for any arbitrary pipe.

Component losses are added to the friction head: balancing valves, thermostatic recirculation valves, check valves, isolation valves, strainers and the heater or storage vessel. Published guidance puts them on the return connections at the water heater, at the pump and at any thermostatic mixing valve cold water inlet. A duty point without them is preliminary, so this calculator lists what belongs there when none are entered rather than treating the friction head as the total.

The result of all that is a required duty point, expressed as a flow at a head. For the worked example above, 4.0 gallons per minute at 6.0 feet, or 9.0 feet with 3 feet of component losses.

What this calculator does not do is confirm a pump. Selecting a model means reading a manufacturer curve at that flow and head and checking that the pump sits on it comfortably rather than at the end of its range. The duty point is the requirement; the curve is the confirmation.

Insulation and the Heat Loss the Flow Comes From

Insulating a recirculation loop is usually presented as an energy measure. On this page it is a sizing input, because the flow is the heat loss divided by a constant and a temperature drop, and insulation is the largest lever anyone has on the heat loss.

The relationship is exact and needs no rule of thumb. The flow is directly proportional to the heat loss, so whatever insulation does to the heat loss it does to the required flow, one for one, and to the velocity with it. Halve the heat loss and you halve the flow, the velocity and the erosion exposure at the same time.

A range of about a quarter to a third of the uninsulated flow circulates as field guidance for what insulation achieves. It is not printed as a figure here, because it could not be confirmed in any of the sources this page cites, and a quantitative multiplier with no source behind it is exactly the kind of number that propagates into a pump selection and is never questioned again. What can be said with a source is the direction and the mechanism.

The figures that are published are per foot heat losses for insulated pipe. The ASPE example uses 8.90 BTU per hour per foot for 2 inch copper with 2 inch insulation, 7.59 for 1-1/2 inch with 2 inch insulation, 7.25 for 1 inch with 1-1/2 inch insulation and 6.33 for 3/4 inch return with 1-1/2 inch insulation. ASPE also states plainly that insulation reduces heat loss but does not eliminate it, which is why a well insulated loop still needs a pump.

The Copper Development Association arrives at insulation from a different direction and reaches the same place. It states that heating water above 140 F can accelerate the process of erosion corrosion, and that to avoid cold hot water concerns, insulation can be added to the hot water supply lines. In other words, insulation is the alternative to raising the temperature, and raising the temperature is what damages the pipe.

So insulation does two things on a recirculation system: it reduces the flow the pump has to move, and it removes the reason to run the loop hotter. Both reduce erosion. ASHRAE 90.1-2019 Addendum aq sets minimum insulation thicknesses for service water heating piping, so on most projects the question is how much rather than whether.

Treating it as a line item to be value engineered out at the end is therefore a decision to buy a larger pump and larger pipe, and to run them harder for twenty years.

Non-Ferrous Pump Body for Domestic Hot Water

A recirculation pump and a hydronic heating circulator can share a duty point and still not be interchangeable, and the difference is what the water does to the pump body.

A closed hydronic heating loop recirculates the same water indefinitely. That water gives up its dissolved oxygen early in the system's life and then behaves: it is treated, it is not replaced, and a cast iron pump body lasts in it for decades. Cast iron circulators are entirely standard in heating work for that reason.

A domestic hot water recirculation loop is not closed in the same sense. Every time a fixture opens, fresh water enters the system, and that water is oxygenated, chlorinated and continuously renewed. The ASPE presentation on domestic hot water recirculation puts the requirement without qualification: all domestic water pumps must be bronze or stainless steel fitted.

So the correct pump here has a bronze or stainless steel body. Fitting a cast iron circulator because the flow and head match is one of the more common errors when a heating contractor is asked to add a recirculation loop to a building, and the pump fails from the inside rather than from the duty.

The distinction also runs the other way. A bronze or stainless pump will work perfectly well on a closed hydronic loop; it simply costs more than the job requires. The substitution is only a problem in one direction.

This calculator states the requirement on every result rather than only when a pump type is entered, because the assumption that a circulator is a circulator is the default position for anyone coming from heating work.

Balancing Recirculation Branches

A recirculation calculation describes a loop where the flow goes where the design intends. Without balancing, it does not.

Published guidance on domestic hot water recirculation states that balancing valves must be sized for each branch and riser to provide the desired temperature drop. ASHRAE Standard 188 goes further and requires that all water systems be balanced and a balance report provided to the building owner, which makes this a documented deliverable rather than a commissioning courtesy.

The reason is that water takes the path of least resistance. On a system with several risers or branches returning to a common line, the shortest and least restrictive path takes most of the flow, and the longest path takes least. That is exactly backwards from what the design needs, because the far branch is the one the calculation was built for.

The consequence is two failures from one cause. The near branches run fast, which is an erosion problem: they may exceed the velocity ceiling even though the total system flow is correct. The far branch runs slow, which is a temperature problem: the water cools before it reaches the fixture, which is the condition the recirculation system exists to prevent.

Adding pump flow does not fix it. More flow makes the near branches faster without materially helping the far one, so an unbalanced system that is short of temperature at the end of the run tends to get a larger pump, which accelerates the erosion on the branches that were already fast.

Balancing valves, or thermostatic recirculation valves that modulate to a set return temperature, distribute the flow according to the design rather than according to the resistance. Manual valves are pressure dependent and hold a setting; thermostatic types adjust flow as temperatures move.

A single loop with no branches does not have this failure mode. Flow control still matters and the short circuit problem does not arise, which is why this calculator asks for the balancing status rather than issuing the same warning to every system.

Workmanship and Local Velocity

A design velocity inside the ceiling is a necessary condition and not a sufficient one. Several installation defects create local velocities far above the average, and no calculation sees them.

The Copper Development Association lists them directly.

Burrs left on the inside of the tube from unreamed ends cause interruption of smooth flow, resulting in localized high water velocity and cavitation. Immediately downstream of an unreamed tube end the local flow pressure is drastically reduced, allowing air bubbles entrained in the water to escape and scour the tube and fitting wall, creating pits that may eventually cause failure.

Protrusions into the flow stream do the same. Excessive lumps of solder or brazing material, or an improperly fabricated tee where the branch protrudes into the run pipe, interrupt smooth flow and produce the same localized high velocity and cavitation.

Numerous abrupt changes in direction have a cumulative effect, and the recommendation is specific: where structural conditions cause numerous directional changes, long radius fittings at 1.5 times the diameter should be used to minimize the interruption of laminar flow.

That last point lands squarely on recirculation systems, which by their nature run the length of a building and back and are full of turns.

Excessive amounts of dissolved gases, vapours or suspended solids in the water conveyed are named as a further factor: at high velocities these impinge on the metal surface and cause erosion corrosion.

None of this changes the number this calculator produces. What it changes is how much margin that number deserves. A loop designed at 1.8 feet per second with unreamed ends and short radius elbows can erode where a loop designed at the same figure with good workmanship does not, and the difference is invisible in the arithmetic.

Temperature, Water Chemistry and Legionella

Three conditions outside the pump calculation change how much velocity the copper will tolerate, and one of them pulls against a requirement from a different discipline entirely.

Temperature is the clearest. The Copper Development Association states that heating the water above 140 F can accelerate the process of erosion corrosion, and its velocity guidance reflects that directly: 4 to 5 feet per second for hot water below 140 F, and 2 to 3 above it. A published forensic account of a recirculation failure recommends designing these systems so that temperatures generally do not exceed 140 F, depending on local code requirements, with velocity no greater than 3 feet per second.

Water chemistry shifts the margin. Chlorine is added to water to kill bacteria and it is aggressive to piping as well; chloramine, which combines chlorine with ammonia to make the disinfectant last longer, stays corrosive for longer too. Dissolved gases and suspended solids impinge on the tube wall at high velocity and do the same. None of these changes the arithmetic; all of them argue for designing below the ceiling rather than at it.

The tension is with Legionella control, and it is not a small one. The favourable growth range for the bacteria is roughly 68 to 122 F, almost all die above 130 F, ASHRAE Guideline 12 recommends the circulating hot water system should not drop below 120 F, and water management practice stores and distributes at or above 140 F with a return temperature of at least 124 F. Every one of those figures pushes the loop temperature up. Copper erosion pushes it down, and the copper guidance changes ceiling precisely at 140 F. The two requirements meet at the same number and pull in opposite directions.

This calculator names that trade off and does not resolve it. Resolving it belongs to a water management plan under ASHRAE Standard 188, which considers the building type, the occupants, the risk criteria and the mitigation available, including thermostatic mixing valves at fixtures to deliver tempered water from a hotter loop.

The controls question sits in the same place. The 2021 IECC requires controls that automatically turn the pump off when the loop is at the desired temperature and there is no demand for hot water, while ASHRAE 188 does not recommend an aquastat to control a circulating pump. That conflict is a conversation with the authority having jurisdiction, and it is worth having before the pump is ordered.

What the calculator can do is make the cost of the decision visible. A loop run hotter for water safety is a loop that needs a lower velocity ceiling, which means either a lower flow or a larger controlling segment. That is a design consequence rather than an opinion.

What Is a Hot Water Recirculation Pump

A hot water recirculation pump keeps a loop of hot water moving so that hot water is available at a fixture without running the tap and waiting. Without one, the water sitting in a long supply run cools between uses and has to be flushed down the drain before hot water arrives.

The loop is a circuit rather than a supply. Hot water leaves the heater, travels the supply main, and a return line brings it back to the heater inlet, so the water in the pipe is continuously replaced with hot water from the source. The pump is what drives that circuit.

What makes sizing it different from every other pump on a plumbing job is that it serves no demand. When a tap opens, the water it delivers comes from the heater through the supply, exactly as it would without a recirculation system. The pump's only job is to replace the heat the pipe loses to the building while nobody is drawing anything.

That has three consequences that run against normal pump instinct. The flow follows from a heat loss rather than a demand. The head is friction only, because a closed loop has no net lift. And the pump must be small, because it runs continuously and a continuous high velocity erodes copper in a way that an occasional peak does not.

Most of the trouble on recirculation systems traces back to a pump that is larger than any of that requires.