How a BMS Optimises the Chiller Plant: Staging and Temperature Reset

The biggest motor in the building, run by software

Walk into the plant room of any shopping mall, hospital or office tower in Kuala Lumpur and you'll find the building's single largest electricity consumer: the chiller plant. This is the machinery that makes cold water, pumps it through the building, and pulls heat out of the air so people stay comfortable in the tropical heat. In many commercial buildings the chiller plant alone accounts for half or more of the entire electricity bill.

Here is the part that surprises people. Two identical plant rooms — same chillers, same pumps, same cooling towers — can post electricity bills that differ by a third or more. The difference isn't the hardware. It's the control: the sequences the Building Management System (BMS) runs to decide how many machines to switch on and how cold to make the water. In the last few parts we met the BMS as the brain of the building. Now that brain meets the building's biggest load.

This part is about three control moves that separate an efficient plant from a wasteful one: staging, chilled-water reset, and condenser-water reset. None of them requires new equipment. All of them are just software deciding well.

A quick refresher on the plant

Before the control, the cast. A basic water-cooled chiller plant has three loops of moving water and one refrigerant loop inside the chiller itself.

- The chiller makes cold water. Inside it, a compressor squeezes refrigerant so it can absorb heat from the building's water and dump it outside.
- Chilled-water pumps send that cold water — typically leaving the chiller around 6–7 °C — out to the air-handling units around the building, where it soaks up heat and returns warmer.
- Condenser-water pumps carry the heat the chiller rejected out to the cooling tower on the roof.
- Cooling-tower fans blow air across that water so it evaporates and cools, then it flows back to the chiller to absorb more heat.

Two numbers let us talk about all of this cleanly. Cooling is measured in refrigeration tons (RT): 1 RT = 12,000 BTU/h = 3.517 kW of cooling. And efficiency is measured in kW/RT — the electrical kilowatts the plant draws to deliver one ton of cooling. Lower is better. An efficient plant runs around 0.5–0.7 kW/RT; a poorly run one drifts past 1.0 kW/RT for the exact same cooling. Everything in this part is, in the end, a fight to push that kW/RT number down.

One more thing worth flagging early: the compressor, the pumps and the tower fans are all electric motors — three-phase motors, the same machines we covered in the Electricity Fundamentals course (how electric motors work and three-phase power explained). We won't re-derive how they work here. The point is that when the BMS "runs the plant," what it is physically doing is deciding which large three-phase motors spin, and how fast. On many big centrifugal chillers those motors don't even sit on the 400 V low-voltage supply the rest of the building uses — a large chiller is frequently fed at medium voltage (3.3, 6.6 or 11 kV), because moving that much power at 400 V would need impractically fat cables. Big or small, LV or MV, it's still a motor the BMS commands, usually through a variable-speed drive (power electronics: rectifiers and inverters).

Move one: staging — run the fewest machines you can

Most plant rooms have more than one chiller. A building might have three, each sized for, say, 500 RT, giving 1,500 RT of installed capacity. But the building almost never needs all 1,500 RT. So which chillers do you run, and how many?

The tempting mistake is to spread the load: switch on all three chillers and let each idle along at a third of its capacity. It feels gentle. It is actually wasteful. Every running chiller drags its own condenser-water pump and its share of cooling-tower fans along with it, and every machine has fixed losses just from being energised. Run three chillers at 33% each and you're paying three sets of overheads to do the work one-and-a-half machines could handle.

Staging (also called sequencing) is the control logic that fixes this. The BMS watches the actual cooling load and runs the fewest chillers that can meet it, each loaded near its efficient point rather than crawling along near-empty. As the building warms up through the day and load climbs, the BMS stages up — brings the next chiller online — before the running machines get pushed past their comfortable range. As the evening cools and the building empties, it stages down, dropping a chiller so the survivors stay well loaded instead of all sagging toward idle.

A well-tuned staging sequence carries a deadband and time delays around each stage-up and stage-down point (we defined deadband back in the sequences part — it's the buffer that stops the plant from flip-flopping a chiller on and off every few minutes). The goal is simple to state and worth a lot: keep each running chiller near the loading where its kW/RT is lowest, and don't pay overheads for machines you don't need.

Why part load is where the savings live

You might assume a chiller is sized to run flat out. In reality it almost never does. Buildings are sized for the worst hour of the hottest day — full sun, full occupancy — and that hour is rare. Every other hour, the plant is coasting at partial capacity.

There's a standard way the industry weights this. AHRI (the US cooling-equipment standards body) publishes a part-load rating that blends a chiller's efficiency at 100%, 75%, 50% and 25% load into a single figure. It's important to read this correctly: AHRI's part-load weighting gives full load only about 1% weight, with most of the weight on the 50–75% band. That is a standardised weighting assumption for comparing machines on paper — not a measured record of how many hours your building actually spends at each load. But it points at something real: a typical building genuinely spends most of its operating hours somewhere around 50–75% load, and hardly any time flat out.

The lesson is strategic. If a chiller lives almost its entire life at part load, then efficiency at part load is the whole game — and part-load efficiency is overwhelmingly a matter of control. That's exactly where the next two moves earn their keep.

Move two: chilled-water temperature reset

Every chiller fights against lift — the temperature gap the compressor has to bridge between the cold water it's making and the warm water it's rejecting outside. The bigger that gap, the harder the compressor works and the more power it draws. Shrink the gap, and the motor eases off.

Chilled-water temperature reset attacks the cold side of that gap. Instead of stubbornly holding, say, 6.5 °C chilled-water supply all day long, the BMS nudges the setpoint upward when the building's cooling demand is light. On a mild, half-empty afternoon the building simply doesn't need the coldest possible water — the air-handling units are barely working. So the BMS lets the chilled-water target drift up a degree or two. That smaller lift means the compressor draws noticeably less power, with nobody feeling any difference in comfort. When demand climbs again — a hot afternoon, a full building — the reset pulls the setpoint back down to keep everyone comfortable.

The BMS decides how far to reset by watching a real signal from the building, usually how hard the busiest air-handling unit's control valve is working. If even the neediest zone is coasting, there's slack to give away, and the setpoint floats up. If a zone starts struggling, the setpoint comes straight back down. It's a continuous, gentle negotiation — and it's the same feedback-control idea we covered in feedback control and PID explained, just applied to a whole plant instead of a single loop.

Move three: condenser-water temperature reset

The other side of the lift is the hot side — the condenser water heading out to the cooling tower. The colder that water comes back from the tower, the smaller the compressor's lift, and again the less power it draws. A useful rule of thumb: every ~1 °F (about 0.6 °C) of warmer condenser water raises chiller power by roughly 1.0–1.5%. Cold condenser water is money in the bank.

Condenser-water reset is the control move that goes and gets that cold water. It lets the cooling-tower fans run harder to push the condenser-water temperature as low as conditions allow, trading a little more fan energy for a larger saving at the compressor — a good trade, because the compressor is far and away the bigger motor.

But there's a hard physical ceiling here, and in Malaysia it bites. A cooling tower cools by evaporation, and evaporation is limited by the wet-bulb temperature — essentially, how much more moisture the air can absorb. Our air is already thick with humidity, so the wet-bulb sits high, and no amount of fan power can make condenser water colder than the wet-bulb allows. A cooling tower in a dry climate can deliver much colder water than the identical tower here. This is why condenser-water reset in the tropics is about capturing every degree the humid air will physically give — not chasing temperatures that our climate simply won't permit. The control still helps; it just works within a tighter envelope than the textbooks (often written for temperate climates) assume.

The whole point: control, not just hardware

Here's the uncomfortable truth these three moves add up to. The same plant room can be efficient or wasteful with no change to the equipment at all — purely from whether these sequences are switched on and tuned, or left off.

Plenty of plants have staging, chilled-water reset and condenser-water reset available in the BMS and never actually running. Someone put the plant into a fixed manual mode during a commissioning hiccup years ago — all chillers on, setpoints nailed to their coldest values — and nobody switched the automation back on. The hardware is capable. The control is asleep. That plant will quietly run past 1.0 kW/RT forever while an identical one down the road holds 0.6.

How much is at stake? Honestly, it depends almost entirely on how bad the starting point was. Studies cite meaningful plant-wide savings from optimised sequencing and reset, though the headline figures — often quoted anywhere from 10% to 40% — depend heavily on how poor the baseline was. A plant that was already reasonably well run has less to gain; a plant stuck in fixed manual mode has a lot. A more modest and dependable figure applies to plants that are already decent and simply add continuous monitoring-and-optimising on top: typically in the 5–10% range. The honest anchor isn't a big percentage — it's the quality of what you started with.

Either way, the savings are only real if the sequences are actually enabled and left running. Control that exists on paper but sits switched off saves nothing.

The takeaway

The chiller plant is the biggest electrical load in most commercial buildings, and it spends nearly all its life at part load — which means efficiency is decided far more by control than by hardware. A BMS earns its keep here through three moves: stage the fewest chillers that meet the load so each runs near its efficient point; reset chilled-water upward when the building is coasting to shrink compressor lift; and reset condenser-water downward as far as the humid tropical air will allow. Done well, the same plant holds a low kW/RT; left in manual, it burns money quietly.

There's a catch that the plant room itself can't show you: a BMS controls the plant now, but the only way to know whether it's still holding its design kW/RT months later — rather than drifting as sequences get disabled or components age — is to keep trending flow, temperatures and power over time, which is exactly the chilled-water and energy monitoring that CobiNeural is built to do.

Next, we turn from the plant that makes the cold water to the air side that delivers it — how a BMS runs the air-handling units and keeps indoor air both comfortable and healthy.