Why Chiller Efficiency Is Measured at Part Load: Understanding IPLV

The number on the nameplate describes a moment that almost never happens

Every chiller ships with a headline efficiency figure stamped on its nameplate: something like 0.60 kW/ton. It reads like a promise. Buy this machine, the number says, and each ton of cooling will cost you 0.60 kilowatts of electricity.

There is a quiet catch. That figure is measured at full load — the chiller running flat out, delivering 100% of its rated cooling on a design day so hot it might occur a handful of afternoons a year. It is the equivalent of judging a car's fuel economy only at top speed on a racetrack. Impressive, precise, and almost completely disconnected from how the thing actually spends its life.

Because here is the reality that this part is built around: a chiller sized for the hottest afternoon of the year spends almost all its hours coasting. It rarely sweats. And a machine that is efficient when sweating is not necessarily efficient when coasting. To judge a chiller honestly, we need a number that reflects the coasting, not the sprint. That number is IPLV.

If you have not yet read the part on kW/ton — the basic unit of chiller efficiency — it is worth a look first, because everything here builds on it. In this part we take that single number and ask a harder question: at which load should we measure it?

Part load is not the exception. It is the job.

Let's define the term plainly. Part load simply means the chiller is producing less than its full rated cooling — say 60% of capacity on a mild, cloudy morning instead of 100% on a blazing afternoon. Full load means it is delivering everything it was designed for.

Now, how a chiller's hours actually split between these states is where intuition goes wrong. Most people picture a chiller humming along at 100%, occasionally easing off. The truth is the mirror image.

The AHRI 550/590 standard — the North American rulebook the whole industry uses to rate chillers — assumes a chiller's operating hours are distributed roughly like this:

- 100% load: about 1% of the time
- 75% load: about 42% of the time
- 50% load: about 45% of the time
- 25% load: about 12% of the time

Read that top line again. The condition the nameplate advertises accounts for around one hour in a hundred.

An important honesty note here: those percentages are the standard's weighting assumptions, not a measured duty cycle for your building. AHRI picked them so that every manufacturer computes the same blended figure the same way and the numbers can be compared fairly. Your building's real profile might lean heavier or lighter depending on its occupancy, its use and the local climate — and in tropical Malaysia, where the cooling load never truly switches off, the curve looks different again from a temperate office block. But the shape of the lesson survives everywhere: part-load hours dominate, and full load is a rare visitor.

Why does a chiller so rarely run flat out? Because it was deliberately sized for the worst case. You buy enough capacity for the hottest, fullest, most punishing hour so that on that one afternoon the building still stays cool. Every other hour of the year, that generous capacity is more than the building needs — so the chiller throttles back. Oversizing, which is extremely common, only pushes the machine further down the load curve, making part-load behaviour even more decisive.

The consequence is unavoidable: part-load efficiency dominates your real energy bill, and a single full-load figure hides most of the story.

IPLV: one number that respects how chillers really run

If part load is where the hours are, we need an efficiency figure that lives there. That is exactly what IPLV — Integrated Part Load Value — is designed to be.

IPLV, defined in AHRI 550/590, takes the chiller's efficiency measured at four load points — 100%, 75%, 50% and 25% — and blends them into a single weighted number using those hour-fractions above. Think of it as a grade-point average for the chiller: instead of judging it on one exam it aced (full load), you weight every "subject" by how much time it spends there.

The blend is not a simple average, though, and this trips people up. The formula is:

$$\text{IPLV} = \frac{1}{\dfrac{0.01}{A} + \dfrac{0.42}{B} + \dfrac{0.45}{C} + \dfrac{0.12}{D}}$$

where A, B, C and D are the chiller's kW/ton at 100%, 75%, 50% and 25% load respectively. The weights 0.01, 0.42, 0.45 and 0.12 are the hour-fractions from the table above.

Two things must be true for this to make sense:

1. A, B, C and D must all be in the same units — all kW/ton (or, if you prefer, all COP). Mixing a kW/ton in one slot with a COP in another produces a meaningless number. Keep the units consistent and the formula behaves.
2. It is a reciprocal (harmonic) blend, not an arithmetic mean. The reason is subtle but sound: efficiency in kW/ton is a rate (electricity per unit of cooling). To average rates fairly by how much cooling is done in each band, you average their reciprocals. It is the same reason you can't find your average speed on a trip by simply averaging the speedometer readings.

A worked example

Suppose a water-cooled chiller measures:

- A = 0.60 kW/ton at 100% load
- B = 0.55 kW/ton at 75% load
- C = 0.52 kW/ton at 50% load
- D = 0.58 kW/ton at 25% load

Notice the machine is actually better in the middle of its range than at full load — more on why in a moment. Plugging in:

$$\text{IPLV} = \frac{1}{\dfrac{0.01}{0.60} + \dfrac{0.42}{0.55} + \dfrac{0.45}{0.52} + \dfrac{0.12}{0.58}} \approx 0.54 \ \text{kW/ton}$$

So the full-load nameplate says 0.60 kW/ton (a COP of about 5.9), but the IPLV of about 0.54 kW/ton (a COP of roughly 6.5) is the figure that better reflects a typical year. If you had specified this chiller on its nameplate number alone, you would have underestimated how good it really is over its working life — and two chillers with identical nameplates can have very different IPLVs.

One more term you'll meet: NPLV — Non-Standard Part Load Value. It is the same idea and the same formula, but computed for your specific site conditions (your actual condenser-water temperatures, for instance) instead of AHRI's standard reference conditions. IPLV is the catalogue figure for comparing machines; NPLV is the tailored figure for your plant.

Why part load can actually be more efficient — "condenser relief"

Here is the genuinely counter-intuitive bit. You might assume a chiller working less hard is like an engine idling — running, but wastefully. Often the opposite is true. Many chillers deliver a lower (better) kW/ton at 50-75% load than at 100%. Why?

Recall from the earlier chiller part that a chiller is a heat pump: the compressor's job is to push heat "uphill" from the cool chilled water to the warmer condenser side, where it is rejected. The steeper that hill — the bigger the temperature gap between the two sides — the harder the compressor works and the more electricity it draws per ton.

Now think about when a chiller runs at part load. Usually it's because the day is milder — cooler outdoor air, a cooling tower rejecting heat into less punishing conditions, so the condenser water returns cooler. Cooler condenser water means a lower pressure for the compressor to pump against. The hill gets shorter. The compressor does the same cooling for less electricity.

This effect is called condenser relief, and it is real money. It is precisely why a chiller can post a better efficiency number at half load on a pleasant morning than at full load on a brutal afternoon. Part load isn't just where the hours are — it's often where the good hours are.

How a chiller turns itself down — and why the drive matters

For a chiller to run efficiently at part load, it has to actually modulate its capacity smoothly rather than crudely switching on and off. There are two main ways it does this:

- Inlet guide vanes on centrifugal machines — adjustable blades that throttle how much refrigerant vapour reaches the compressor.
- Variable compressor speed — slowing the compressor down when less cooling is needed.

That second method is the same idea we met back in Electricity Fundamentals: a variable-speed drive (VFD, or inverter) electronically varies the frequency fed to the motor so it can run at any speed instead of just full tilt. If you'd like the underlying physics of how an inverter builds a variable-frequency supply, the power electronics: rectifiers and inverters part lays it out, and the how electric motors work part explains why motor speed follows frequency.

The efficiency payoff is large because of a physical law we've leaned on throughout this course: for the pumps, fans and compressors that move fluids, power tends to fall with roughly the cube of speed. Slow a compressor to 75% and, all else equal, its power can drop far more steeply than 75%. A variable-speed chiller can therefore ride the part-load curve gracefully; a fixed-speed one that can only throttle or cycle wastes much of the opportunity. This is a big reason variable-speed chillers post such strong IPLV figures.

Where the savings leak away: too many machines, too lightly loaded

A single efficient chiller is only half the plant. Most buildings run several chillers, and here a whole plant can quietly squander its part-load advantage through poor staging — the logic that decides how many machines run and at what load.

Picture a plant with three chillers meeting a load that two could handle comfortably. If the controls fire up all three and spread the work thinly, each machine limps along at, say, 30% load — and at very low load, efficiency often falls off again (notice how our example's kW/ton crept up at 25%). Worse, every running chiller drags its own condenser-water pump and cooling-tower fan into service, whether it's pulling its weight or not.

Run two chillers at a healthy 65% each instead, and both sit in that sweet spot where condenser relief and smooth modulation combine — while the third machine rests and its auxiliaries switch off entirely. Same cooling delivered, noticeably less electricity consumed.

Getting that decision right — how many machines, loaded how heavily, in what order — is the art of staging and sequencing, and it deserves its own treatment. We come back to it in detail in a later part on running a multi-chiller plant well. For now, hold onto the principle: a plant full of efficient chillers can still be an inefficient plant if it stages them badly.

The takeaway

A chiller's full-load nameplate number describes a condition it may see one hour in a hundred. Real efficiency lives at part load, and IPLV is the single weighted figure — built from performance at 100/75/50/25% load — that captures it honestly, with NPLV doing the same for your specific site. Better still, thanks to condenser relief and variable-speed control, a well-run chiller is frequently more efficient at part load than at full — provided the plant is staged to let it.

There is one unavoidable consequence of all this. Because true efficiency drifts constantly with load, weather and the slow fouling of heat-exchanger surfaces, the only way to know a plant's real part-load kW/ton is to measure it continuously — trending the chilled-water flow and supply/return temperatures against the electricity actually drawn, hour after hour. A nameplate can't tell you; only continuous chilled-water and energy monitoring can.

Next, we turn to a subtle failure that silently wrecks part-load efficiency and hides from every gauge on the plant: low delta-T syndrome.