Hydronic Loops Hydronics

"Hydronic" systems use water to transfer thermal energy throughout a building. We'll focus on the heating side — I understand boilers more than chillers — but most of the ideas transfer to chilled water just the same. The equipment is the easy part: a boiler or chiller adds or removes heat from the water, pumps push it, coils take what they need. The part that trips people up is the piping topology — how the loops are arranged, and why. Get the topology and a wall of pumps and tees stops looking like spaghetti and starts looking like three or four moves repeated. Here are the three you'll meet most.

2-Pipe Direct Return

The simplest layout. The source — a boiler (or a chiller; the picture is the same for chilled water) — feeds a supply main that runs past every load. Each load taps off the supply, does its job, and dumps into a return main that heads straight back to the source. "Direct return" is exactly that: each load's return is "go back the way you came."

The catch: the nearest load has a short total trip — a little out, a little back. The farthest load's water has to cross the whole building and the whole length back. Water, like current, takes the path of least resistance, so the near load gets too much flow and the far load gets too little — left alone, this layout self-unbalances. Every load needs a balancing valve, and somebody has to set them. Cheap to pipe, fussy to commission.

2-pipe direct return A boiler with a pump feeds a supply main running left to right past three loads in parallel. Each load returns into a separate return main that runs right to left, straight back to the boiler. The nearest load has a short loop; the farthest load's water travels the full length of the building and back, so flow naturally favours the near loads unless balanced. BOILER PUMP LOAD 1 (nearest) LOAD 2 LOAD 3 (farthest) supply main → ← return main (straight back to the source)
supply water return water

2-Pipe Reverse Return

Same supply main, same loads. The trick is in the return: instead of running back toward the source, the return main keeps going the same direction as the supply — past every load — and only then loops the long way home. So the first load supplied is the last to return, and the last supplied is the first to return. Add up the pipe each load's water travels and it comes out about the same for everyone.

Roughly equal path length means roughly equal resistance, which means the loads naturally split flow close to evenly — "self-balancing." (You'll still want balancing valves, but they're trimming, not doing the heavy lifting.) The price: that loop-back is a whole extra run of pipe — you're essentially doubling the return — so it costs more to install. Trade commissioning effort for pipe and labor; pick your poison.

2-pipe reverse return A boiler with a pump feeds a supply main running left to right past three loads in parallel. Each load returns into a return main that also runs left to right, past all the loads, and only then doubles back the full length of the building to the boiler. The first load supplied is the last to return, so every load's total pipe path is roughly equal — naturally self-balancing — at the cost of the extra doubled-back run of pipe. BOILER PUMP LOAD 1 (nearest) LOAD 2 LOAD 3 (farthest) supply main → Return runs WITH the supply then the long way back. The extra pipe is the price of self-balancing
supply water return water

Primary-Secondary — the "Twin-T" Boiler Injection

Here's the one that takes people forever. The natural assumption — and it's wrong — is that the whole building's water flows through the boiler: the boiler is just the spot in the loop where the water gets heated, like a kettle on the way around. It isn't. If it were, a 100-GPM boiler couldn't serve a 200-GPM building, and a boiler that hates low flow and thermal shock would get both every time the building turned down. So that's not how it's piped.

The boiler sits on its own short loop — the primary — with its own dedicated boiler pump keeping flow through the boiler constant, always, no matter what the building is doing. That constant flow is what protects the boiler. A separate injection pump taps hot water off the primary and pushes it into the system loop — the building loop — which has its own system pump moving water through the loads. At any moment, only the slice of flow the injection pump is pulling ever touches the boiler; the rest of the building's water is just circulating through the loads, mixing with whatever hot water the injection pump adds. The two tees where the loops meet sit right next to each other — the "closely-spaced tees," the "common pipe," the "twin-T" — so close that there's essentially no pressure drop between them. That's the whole mechanism: with no pressure difference across that little stub, the three pumps — boiler, injection, system — each do their own thing instead of fighting each other's discharge pressure.

Worked example · follow the water

A boiler sized for 100 GPM through itself — held there by the boiler pump. The building's system pump moves 200 GPM through the loads. Right now the injection pump is moving 40 GPM from the primary into the system. So:

  • 100 GPM circulates through the boiler — always, no matter what the building wants.
  • 40 GPM of that gets pulled into the system by the injection pump, where it mixes with cooler return water before it reaches a single load.
  • The other 60 GPM stays in the primary and goes straight back to the boiler to be reheated.
  • 200 GPM moves through the loads — and most of it never touches the boiler. The loads see the mixed temperature, downstream of the injection point.
  • That blend is the payoff: 40 GPM of 180 °F primary blends with 160 GPM of 140 °F system return → a 148 °F supply to the loads, the figure the widget shows.

Colder outside, building wants more heat? The injection pump speeds up — pulls more hot water from the primary into the system — and the supply temperature to the loads climbs. The boiler's flow rate never changes. The injection pump is the control point.

Why it matters on the floor: those closely-spaced tees look like a mistake the first time you see them — like somebody piped a short-circuit straight across the boiler. They aren't. That little stub of common pipe is the entire reason a building whose demand swings all day can be served by a boiler that insists on a steady diet. Recognize the twin-T and the rest of the plant snaps into focus; miss it and you'll spend an afternoon convinced the pipefitter botched it.

Primary-secondary "twin-T" boiler injection Two separate loops. On the left, a small primary loop: a boiler and a dedicated boiler pump with water always circulating around it. On the right, a larger system loop: a system pump moving water through two building loads. The only connection between them is a short bridge with a pair of closely-spaced tees and an injection pump on it; the injection pump pulls a portion of hot water from the primary loop into the system loop, where it mixes with cooler return water. Most of the building's water never passes through the boiler — only the slice the injection pump diverts. The injection pump's speed is the control point for supply temperature. BOILER BOILER PUMP constant flow INJECTION PUMP the control point SYSTEM PUMP LOAD LOAD PRIMARY LOOP the boiler's own loop SYSTEM LOOP closely-spaced tees ≈ no Δp
hot supply water cooler return / mixed water

Do you see a difference with this system I didn't mention? Find out what it is →

Injection flow
40.0 GPM
Pump speed
45 Hz
% of design speed
100 %
Primary loop: 180.0 °F
System supply: 148.0 °F
Normal operation
← Back to Education