Heat pumps · Chillers · Sludge drying systems
Industrial cooling · Process

Chiller Selection: Full-Load COP Is Not the Decision

Full-load COP is easy to compare and easy to market. It is also the least representative number for most buildings. A chiller that runs at 100% for a few hundred hours a year spends the rest of its life at 25–75% load, where the part-load curve, the compressor type and the staging sequence decide the electricity bill.

Selection should start with the annual load profile, then use the part-load performance to compare machines. Full-load COP is a boundary condition, not the decision.

1. One point versus a load profile

COP at 100% load tells you the efficiency at design day. It says nothing about the 50% condition that dominates a mild spring, the 25% condition at night or the 75% condition on a typical summer day. Two chillers with the same full-load COP can have very different part-load curves: one may hold efficiency down to 40%, the other may fall away below 60%.

The load profile is the share of annual cooling energy at each load band. An office in a cooling-dominated climate may spend 60–70% of its hours below 50% load; a data center may spend most of its hours above 75%; a hotel has a morning and evening peak with a deep night valley. The profile, not the building type label, drives the weighting.

2. IPLV is a weighted part-load number

IPLV combines the efficiency at four load points with standard weights. A common weighting is 1% at 100%, 19% at 75%, 30% at 50% and 50% at 25%. The exact weights depend on the standard and the application; check the standard named in the tender.

The weighted average input power per unit of cooling is:

P_weighted = 0.01 × P100 + 0.19 × P75 + 0.30 × P50 + 0.50 × P25

IPLV is the reciprocal of that weighted power fraction, so the result is a COP. The 25% point carries half the weight, which is why a machine that is excellent at full load but poor at quarter load can still have a mediocre IPLV.

3. A worked comparison

Take two 1,000 kW chillers. Machine A has COP 5.8 at 100%, 6.2 at 75%, 6.0 at 50% and 4.8 at 25%. Machine B has COP 6.0 at 100%, 6.0 at 75%, 5.5 at 50% and 4.0 at 25%.

Load Weight Machine A COP Machine B COP
100% 1% 5.8 6.0
75% 19% 6.2 6.0
50% 30% 6.0 5.5
25% 50% 4.8 4.0

Machine B wins at full load by 3.4%. Machine A wins at 50% by 9% and at 25% by 20%. With the weights above, Machine A’s weighted COP is about 5.4 and Machine B’s about 4.7; the full-load winner is the annual loser. The gap is large enough to change the operating cost by double digits.

4. The part-load curve is not only the compressor

Compressor type. A screw compressor with a fixed slide valve may unload in steps; a scroll bank may stage in discrete increments; a centrifugal with variable inlet guide vanes has a different curve; a magnetic-bearing centrifugal with variable speed can hold a high part-load efficiency. The curve, not the type name, is the evidence.

Variable speed. A VSD compressor can reduce power at part load, but the saving depends on the load profile and the condenser-water temperature. A VSD machine that runs at full speed most of the year may not repay its premium.

Condenser-water temperature. A chiller at 30 °C condenser water is more efficient than the same machine at 35 °C. A cooling tower that produces lower water temperature in the shoulder season improves the part-load performance without changing the chiller.

Staging. Two half-capacity machines staged to keep each above 50% can beat one large machine that runs at 25%. The control sequence is part of the selection.

5. The cost of oversizing

An oversized chiller costs more, occupies more mechanical room and runs at a lower load. If the design load is 800 kW and the selected machine is 1,200 kW, the typical load may sit at 40–50%, near the bottom of the efficient band or in the cycling range. The part-load penalty can exceed the price difference.

Selection margin is legitimate, but it should be applied to the load calculation, not stacked on top of a conservative load and a conservative machine. A 10–20% margin on a well-built load is different from a 40% margin on a rough estimate.

6. A selection checklist

Ask for the part-load table at the project’s condenser-water temperature, not only the nominal COP. Ask for the load profile and the annual energy estimate. Ask how the machines will be staged, and at what load the next machine starts. Ask what happens at 25% load on a mild day. If the answer is “it cycles”, the selection is incomplete.

Common questions

Is a higher full-load COP always better?

No. It is better at full load. If the building rarely reaches full load, the part-load curve can reverse the ranking.

What is the difference between IPLV and NPLV?

IPLV uses standard rating conditions; NPLV uses the project’s non-standard conditions. NPLV is closer to the actual plant, but it requires the project conditions to be defined.

How many load points should I compare?

At least 100%, 75%, 50% and 25%, plus the lowest expected load. A single COP cannot show the shape of the curve.

Does a VSD always save energy?

No. It saves when the load profile has long part-load hours and the compressor unloads efficiently. At full load, the VSD may be slightly less efficient than a fixed-speed machine.

Can staging fix a poor part-load machine?

Partly. Staging keeps machines in a better band, but it cannot make a machine efficient at a load it was not designed for. Selection and staging work together.

What is the most common selection mistake?

Buying on full-load COP and ignoring the 25–50% band where the plant actually runs. Part-load performance and staging decide the annual bill.

Why not full-load COP?

The chiller spends most hours at part load. A good full-load COP can hide a poor part-load curve and a bad staging sequence.

What metric should I use?

Load-weighted efficiency at the site’s hours, plus the staging sequence. Ask for the part-load table at the actual condenser conditions.

Further technical reading

LONGCOM can build the load profile, compare part-load curves at the project’s condenser condition and define the staging sequence. For a part-load selection review, contact 400-878-0280 or 13002001280.

Related reading

COP and IPLV guarantee · Chiller plant group control · Selection margin mistakes