How Does a Chilled-Water System Work? The Two Loops Explained

One cold factory instead of a hundred cold boxes

Think about how a small shop stays cool. It has a wall-mounted air conditioner, and inside that unit is everything needed to make cold: a compressor, a fan, and refrigerant. The cold is made right there, in the room, at the point where you feel it.

Now picture a 40-storey office tower. You cannot bolt a separate air conditioner onto every wall and expect it to work well, and you certainly cannot run hundreds of compressors efficiently. So large buildings do something cleverer. They build one central cold factory in the basement or on a plant floor, make cold water there, and pump that cold water out to every corner of the building through pipes.

That central cold factory is the chiller. Understanding how it and its water loops work is the single most useful mental picture in this whole course, because almost everything else in a large building's cooling system hangs off it. So let's build that picture slowly and carefully.

The chiller does not make cold — it moves heat

Here is the idea that trips up almost everyone, so we'll say it plainly: a chiller does not create cold. It moves heat.

There is no such thing as "cold" as a substance you can manufacture and pump around. Cold is simply the absence of heat. When we cool a room, what we are really doing is picking up unwanted heat from inside and carrying it somewhere else — outside the building. A chiller is a heat relay: it collects heat from all over the building and hands it off, stage by stage, until it is dumped into the outdoor world.

Keep that word — relay — in your head. A chilled-water system is a relay race, and the "baton" being passed along is heat. Once you see it that way, the loops stop looking like a tangle of pipes and start looking like a sensible chain of hand-offs.

Three loops: two of water, one of refrigerant

A chilled-water plant is built from three loops. Two of them carry water. One carries refrigerant and lives sealed inside the chiller. Let's meet them in the order the heat travels.

Loop 1 — the chilled-water loop (the building side)

This is the loop that touches the building and does the actual comfort cooling.

The chiller produces cold water — the chilled-water supply — at around 6-7 degC. Pumps push this cold water out through pipes to cooling coils spread all over the building. These coils sit inside air-handling units and fan coil units (the boxes that blow cool air into rooms — we cover them in the next part). Air from the building blows across these cold coils, the air gives up its heat to the water, and cool air continues on into the offices.

Because the water has absorbed that heat, it comes back warmer. This is the chilled-water return, typically several degrees warmer than the supply. The difference between supply and return temperature is called the delta-T (delta just means "difference"), and a healthy design figure is around 5-6 degC — for example, water leaving at 6 degC and coming back at about 12 degC.

That temperature rise is not a defect. It is the whole point. The warming of the water is the physical proof that cooling was delivered. No temperature rise would mean the water travelled all the way around and picked up no heat — it did no useful work. (Delta-T turns out to be one of the most revealing health signals a chilled-water system has, and we devote a whole later part to what happens when it goes wrong.)

The warm return water flows back to the chiller, gets re-chilled back down to 6-7 degC, and heads out again. Round and round, continuously, all day.

So the first hand-off in our relay is complete: heat has moved from the building's air into the chilled water.

Loop 3 — the refrigerant loop (inside the chiller)

Now the chiller has a problem. It has a stream of warm water coming back, and it needs to pull the heat back out of that water to make it cold again. How?

This is where the refrigeration cycle we met earlier in the course does its job. Sealed inside the chiller is a refrigerant loop — the same evaporate-compress-condense-expand cycle that runs in your home air conditioner, just much larger. We won't repeat the full cycle here, but the essential moves are:

- In the chiller's evaporator, cold refrigerant absorbs heat from the returning chilled water, cooling that water back down to 6-7 degC. (Hand-off: heat moves from the water into the refrigerant.)
- The compressor — a large motor — squeezes the refrigerant, which raises its temperature well above the outdoor temperature so the heat can be shed.
- In the chiller's condenser, the hot refrigerant dumps its heat into... something. And what it dumps into is exactly what separates the two families of chiller.

Notice that the refrigerant loop is the bridge. It connects the cold building-side water on one end to the heat-rejection side on the other. It is the engine in the middle of the relay.

Loop 2 — the condenser-water loop (the heat-rejection side)

For a water-cooled chiller, the condenser hands its heat into a second water loop: the condenser-water loop.

This loop carries the collected heat out to a cooling tower, usually sitting on the roof. There, the warm condenser water is exposed to outdoor air and a little of it evaporates, which carries the heat away into the sky. The now-cooler water flows back to the chiller's condenser to collect the next batch of heat. (Cooling towers have their own clever physics, and they get their own part later.)

So the full relay, end to end, looks like this:

Building air → chilled water → refrigerant → condenser water → cooling tower → outdoor sky.

Five hand-offs, and the heat you felt as a stuffy warm office ends up dispersed above the rooftop. That is the entire job of a chilled-water plant.

Air-cooled or water-cooled: two ways to throw heat away

Not every chiller has that condenser-water loop and cooling tower. There are two families, and the only difference between them is how they get rid of the heat at the final stage.

Air-cooled chillers skip the second water loop entirely. Their condenser is cooled directly by fans blowing outdoor air across it — like a giant version of the finned unit behind a home air conditioner. The heat goes straight from refrigerant to outdoor air. Simpler, no tower, no condenser-water pumps, and no water treatment to worry about. You often see them on the roofs of mid-sized buildings.

Water-cooled chillers use the condenser-water loop and cooling tower we just described. There are more moving parts — a tower, extra pumps, water treatment — but there is a good reason large buildings put up with the complexity: water-cooled plants are usually more efficient. A cooling tower, using evaporation, can cool the condenser water closer to the outdoor conditions than fans blowing on hot metal can. A cooler condenser means the compressor does not have to squeeze the refrigerant as hard, which means it draws less electricity for the same cooling.

That efficiency gap is why most large commercial buildings — malls, hospitals, big offices — run water-cooled plants, while smaller buildings often choose the simpler air-cooled route. We'll put real efficiency numbers (kW per refrigeration ton, and COP) on this comparison in a later part; for now, hold the intuition: cooler heat-rejection means a more efficient chiller.

Where all the electricity goes

Here is a fact worth memorising: in most commercial buildings, the chiller's compressor is the single biggest electrical load in the entire building. Not the lifts, not the lighting — the chiller.

That is because the compressor is a large three-phase motor, and squeezing refrigerant against a building's worth of heat is genuinely hard work. If you're fuzzy on why big motors use three-phase supply, our Electricity Fundamentals course covers it in three-phase power explained and how motors turn current into torque in how electric motors work.

But the compressor is not the only motor in the plant. A chilled-water system is really a family of motors working together:

- The compressor — by far the largest single load.
- The chilled-water pumps — pushing cold water around the building loop.
- The condenser-water pumps — circulating water out to the cooling tower (water-cooled plants only).
- The cooling-tower fans — pulling air through the tower to drive evaporation.

The pumps and fans are significant secondary loads in their own right. Add them up and the whole plant is a serious slice of the building's electricity bill — which, if your building is on a demand-metered (MD) tariff, also shapes the peak kW you're charged for when several of these motors start together. (More on that in a later part.)

One more subtlety worth knowing: a large motor running lightly loaded has a poorer power factor than one running near its design load, which quietly affects how efficiently that current is used. Our Electricity Fundamentals course explains why in reactive power is not wasted energy. It matters here because chillers spend most of their lives at part load, not full load.

Why the pumping gets clever: primary and secondary

There's one more piece of the picture, and it explains a lot of pipework you'll see on real plants.

The chiller itself likes a steady, constant flow of water through its evaporator — too little flow and it can't transfer heat properly. But the building's demand for cold water is anything but constant. On a hot afternoon with a full office, coils all over the building are wide open and gulping chilled water. At night with the building near-empty, most coils are nearly shut. The flow the building wants swings wildly through the day.

So how do you keep steady flow through the chiller while the building demands wildly varying flow? You decouple the two. A couple of common arrangements do exactly this:

- Primary/secondary pumping uses two separate sets of pumps. A primary set keeps constant flow through the chillers. A secondary set pushes whatever variable flow the building actually needs. A short connecting pipe between them, called a decoupler, absorbs the difference so neither side disturbs the other.
- Variable-primary flow is a more modern single-loop design where variable-speed pumps throttle the building flow directly, within limits the chiller can safely tolerate.

You don't need the plumbing details. The idea to carry forward is simple: the flow through the chiller and the flow through the building are deliberately allowed to differ, because the machine and the building want different things. This decoupling is exactly why real chilled-water performance can only be judged by measuring what's actually happening, not by reading a design spec off a nameplate.

The takeaway

A chilled-water system is not a cold factory — it is a heat relay. Two water loops and one refrigerant loop pass the building's unwanted heat along a chain: room air gives its heat to chilled water (supplied at ~6-7 degC, returning warmer with a healthy ~5-6 degC delta-T), the refrigerant loop inside the chiller pulls that heat back out, and — for water-cooled plants — a condenser-water loop carries it to a cooling tower and out to the sky. Air-cooled chillers skip the tower and dump straight to outdoor air; water-cooled plants add complexity but usually run more efficiently. Throughout, the compressor is the biggest motor and the biggest electricity user in most buildings, with pumps and fans close behind.

Because the flow through the chiller is deliberately decoupled from the flow the building demands, a chiller's real efficiency — its true kW per refrigeration ton — never shows up on a nameplate. It only emerges when you trend chilled-water flow, supply and return temperatures, and electrical power together over time, which is precisely what chilled-water monitoring is built to do. (To be clear, that's loop performance — flows and temperatures — not compressor bearings or refrigerant charge, which are a mechanical technician's domain.)

Next, we follow the cold water to where it actually meets the air you breathe: the air-handling units and fan coil units scattered throughout the building.