Recirculating Domestic Hot Water: Buying Back the Ninety Seconds

Three ways to get the water back, and what each one really is

A dedicated return is the proper answer: a separate small-bore pipe from the far end of the hot main back to the heat source, with a pump keeping the whole flow leg warm. Every outlet teed off that main is then a short branch from hot water, and the only wait is the branch itself. It costs a second pipe through the whole building, which is why it is decided at design stage and almost never retrofitted into finished construction.

A crossover arrangement uses the cold main as the return. A thermostatic valve under the furthest fixture opens while the water reaching it is cool, letting the pump push cooled hot water into the cold pipe, and closes once hot arrives. It needs no second pipe, which makes it the retrofit of choice, and it has one consequence nobody mentions at the sale: the cold water at that fixture is lukewarm for a while after each cycle, and in a single-bathroom dwelling that is the tap people drink from.

A thermosiphon return has no pump at all and relies on the density difference between the hot flow and the cooling return to drive circulation. It works in tall buildings with vertical loops and short horizontal legs, and it barely works anywhere else. Where one already exists and performs, leave it. Do not design one into a modern well-insulated horizontal layout and expect it to move water.

There is also the option of not building a loop. Shortening the run, relocating the heat source, dropping a small point-of-use heater at the outlying fixture, or reducing the bore of the final leg all attack the volume side of the equation instead of the flow side, and on a single distant fixture they are frequently cheaper to install and always cheaper to run.

A warm loop is a heat emitter you did not specify

The moment a loop circulates, every metre of it is losing heat to the space around it, continuously, whether or not anybody is drawing water. That loss does not stop at night, it does not stop in August, and in a building with a cooling load it gets paid for twice — once at the heat source and again at the chiller.

The scale surprises people. A loop is typically the longest run of hot pipe in the building, and unlike a heating circuit it is hot for far more hours of the year. On many commercial buildings the annual standing loss from the recirculation loop exceeds the useful heat delivered to the taps. That is not an argument against loops; it is an argument for insulating them properly and controlling when they run.

The return leg counts too, and it is the one that gets forgotten in both the insulation order and the heat loss calculation. It runs only a few degrees below the flow, over the same distance, usually in the same ceiling void, and it is often left bare because it is small. A bare 15 mm return alongside a well-lagged 28 mm flow can contribute a startling share of the loop's total loss.

Everything that touches the pipe drains it as well. Uninsulated valve bodies, unlagged sections through fire-rated walls, bare pipe where it crosses a support, and the hundred short gaps where the sleeve stops for a bracket all add up. Loop loss is not distributed evenly along the run — it concentrates at the places the insulation stops.

Balancing, or why the nearest riser takes everything

A loop with several risers is several parallel paths, and water takes the easy one. Without balancing, the riser nearest the plant receives most of the circulation and the far riser receives almost none, so the building has a hot water problem in exactly one wing and a plant room that appears to be working perfectly. The complaint reads as a capacity failure and it is a distribution failure.

Static balancing valves fix this by adding measured resistance to the short paths until every path sees a similar drop. They work, they hold their setting, and they require a commissioning visit with a differential pressure meter and a set of valve charts. They also assume the loop always runs at the design flow, which stops being true the moment a variable-speed pump modulates.

Thermostatic balancing valves take a different approach: each one throttles itself to hold its own branch return at a set temperature, so the flow distributes itself according to where the heat is actually being lost. On multi-riser buildings they are usually the better answer, and several types include a periodic full-open thermal disinfection position, which matters where the water safety plan calls for one.

Whichever type is fitted, they need to be found and reached later. Balancing valves buried above plasterboard ceilings with no access panel are, functionally, fixed orifices. Schedule them, label them, and record the commissioned setting on the label as well as in the file.

1. Bring the loop up to normal operating temperature with all branches open and the pump at design speed.
2. Work outward from the plant, recording the return temperature at every branch tie-in.
3. Throttle the branches running hottest — they are the ones taking more than their share — and let the far branches recover.
4. Repeat the circuit; each adjustment changes the ones you already set, so two or three passes is normal.
5. Verify the total flow at the pump still matches the duty, and re-check the furthest branch last.
6. Label every valve with its final setting and photograph the label in place.

Temperature is a safety parameter here, not a comfort setting

In a building with a written water safety plan, the recirculation loop is not primarily a convenience feature. It is the mechanism that keeps stored and distributed hot water above the temperature at which bacterial proliferation is controlled, throughout the distribution, including at the ends. Turning a loop off to save energy in such a building is a change to the risk assessment, not a settings change.

That places a floor under the temperature drop you are allowed to design for. A return arriving back at the plant well below the flow temperature means part of the loop has been sitting in the range the safety plan exists to avoid. Where a plan applies, the acceptable return temperature is stated in it, and the flow rate has to be sized to hold that figure, not the drop that would have been comfortable.

It also places a ceiling on what reaches the outlet. Storage and distribution at a temperature that manages bacterial risk will scald, so the protection happens at or near the fixture through a mixing valve of the right type, and the blended leg downstream is kept as short as the layout allows. A loop that circulates blended water at a mild temperature throughout the building has removed the scald risk by creating the other one.

Sources for the actual numbers are the adopted plumbing code, the building's own water safety plan, and the recognised guidance the plan cites. They differ by jurisdiction and by building type, and a residential dwelling, a care home and a hotel are three different answers. Confirm which document governs this building instead of carrying a figure across from the last job.

Controls, and the honest arithmetic of the trade-off

Every control strategy trades wait time against standing loss, and the trade is close to linear: the more hours the loop is warm, the fewer seconds anyone waits and the more heat the building loses. There is no setting that removes both.

A time clock is the crudest useful control and often the most effective in a building with a predictable rhythm. It gives instant hot water during the morning and evening windows and full wait outside them, and it cuts loop hours by more than half in most dwellings. Its weakness is that it knows nothing about the building, so a household that changes shift patterns gets the worst of both.

An aquastat on the return runs the pump only when the loop has cooled below a setpoint, which stops the pump churning water that is already hot. It is a genuine improvement over continuous running and is almost always worth pairing with a clock. Demand control goes furthest: a button, a motion sensor or a flow switch starts the pump on request, and the occupant accepts a short priming wait in exchange for a loop that is cold most of the day.

Whatever is fitted, someone has to be told what it does. A demand-control button that nobody explained gets reported as a fault within a fortnight, and the usual remedy applied by whoever attends is to strap the pump to a permanent live. Leave a one-page description of the control strategy in the plant room and another with the occupant.

What to measure, and what to leave behind

Commissioning a loop is a temperature survey, not a pump test. Bring the system to normal operating condition, then record the flow temperature leaving the plant, the return temperature arriving back, and the return temperature at every branch tie-in. Those readings are the whole diagnosis: a branch return sitting well below the others is a balancing fault, and a system return sitting well below the flow is a flow rate or an insulation fault.

Then repeat the measurement the occupant actually cares about. Time the wait at the furthest outlet with the loop running as it will run, and again with the loop off. The difference between those two numbers is what the system was installed to deliver, and it is the only performance figure the building owner will ever quote back at you.

Write down the pump setting. Circulators get nudged, replaced and reset by people with no knowledge of the balance, and a loop that took a morning to commission can be undone by somebody turning a dial to the maximum in the belief that faster is better. A label on the pump giving the commissioned speed and the design flow prevents most of it.

Hand over the standing loss as well as the wait time. The building owner is buying instant hot water and paying for a heated pipe circuit, and stating both sides of that in the handover turns a future energy query into a decision they already made instead of a defect they discovered.