top of page

Chiller Trend COP: Divide Total Cooling Energy by Total Electrical Energy

Oct 26
4 min read
AI-generated illustrative scene: two chillers and insulated pipes in a mechanical plant room. Not product or project evidence.

AI-generated illustrative scene, not a product screenshot or a completed-project photograph. The worked calculation below uses explicit example inputs.


A BMS dashboard may display a COP value every few minutes and then average those values for the daily report. That average answers a statistical question about the displayed numbers. It does not necessarily answer the commercial and engineering question: how much cooling did the chiller deliver for each unit of electricity consumed during the period?


The period result requires matched thermal and electrical energy totals. This article develops an original two-interval example and a practical data-quality workflow. It is not a substitute for an equipment rating procedure, an annual simulation or an agreed measurement-and-verification plan. No actual plant performance or financial saving is claimed.


Define what the ratio contains


COP is useful cooling output divided by input when both are expressed on compatible energy or power bases. Trane's chilled-water systems guidance explains this basic efficiency definition and the distinction between equipment and system considerations. For a logged reporting period, use thermal kWh divided by electrical kWh, with identical time coverage and an explicit equipment boundary.


Here the boundary is the chiller package only. External chilled-water pumps, condenser-water pumps and tower fans are excluded. A plant COP would require the appropriate additional electrical consumption and a thermal measurement boundary that represents the useful cooling delivered. Do not compare those two labels as though they measure the same thing.


Original example: two hours, two operating points


Assume that each example interval lasts exactly one hour and the stated values are representative averages over that interval. During hour A the chiller supplies 600 kW of cooling while consuming 100 kW electrical. During hour B it supplies 100 kW of cooling while consuming 50 kW electrical. Assume no gaps, meter resets or sign errors.


Hour A: COP = 600 ÷ 100 = 6.0; cooling = 600 kWh; input = 100 kWh.


Hour B: COP = 100 ÷ 50 = 2.0; cooling = 100 kWh; input = 50 kWh.


Simple mean COP = (6.0 + 2.0) ÷ 2 = 4.0.


Period COP = (600 + 100) ÷ (100 + 50) = 4.667.


The arithmetic mean gives both displayed COP values equal weight. The energy ratio gives each interval a weight proportional to its electrical energy. In this example, the higher-COP hour accounts for twice the electrical energy of the lower-COP hour. The resulting period COP is therefore closer to six than the simple mean suggests.


The difference is not evidence of a faulty chiller. Both statistical calculations are internally consistent; they answer different questions. For delivered cooling per consumed electricity, retain the energy ratio. If an analyst also wants the distribution of instantaneous COP values, report its median or percentiles separately and explain the valid-data filter.


Two equal-duration intervals have COP 6 and 2. Their simple mean is 4, while 700 kWh cooling divided by 150 kWh electrical gives period COP 4.667.

Original interval example; package boundary excludes external auxiliaries.


Unequal intervals need explicit durations


Suppose hour A's condition lasts two hours rather than one, while B still lasts one hour. Cooling becomes 1300 kWh and electricity becomes 250 kWh, producing a period COP of 5.2. A time-weighted average of the two COP values would be (6 × 2 + 2 × 1)/3 = 4.667. Even time weighting alone does not reproduce the energy ratio when input power differs.


For piecewise-constant interval averages, sum Qᵢ × Δtᵢ and Pᵢ × Δtᵢ separately. For instantaneous samples, choose and document an integration method consistent with the sampling behavior. Do not assume that irregular readings represent equal time intervals. A cumulative energy meter can be preferable when its boundary, resolution and reset behavior are understood.


Handle zero output and missing data honestly


Standby electricity within the chosen reporting boundary should not disappear merely because cooling output is zero. For example, adding 10 kWh of included standby input to the original two-hour totals changes period COP to 700/160 = 4.375. That is a boundary-consistent accounting change, not a new operating-point efficiency.


Missing thermal readings are different from zero cooling. If electrical data exist for an interval but thermal data are absent, flag the gap and report matched coverage. Do not insert a zero or discard only the inconvenient input. Near-zero electrical denominators can also create extreme instantaneous ratios; investigate resolution, timestamp alignment and meter validity before interpreting them.


Make the comparison reproducible


Store the start and end times, timezone, included meters, unit conversions, valid interval mask and excluded periods alongside the result. Check that cooling and input refer to the same timestamps. Review whether package-integrated pumps are already included in the electrical meter so that a later plant calculation does not count them twice.


A useful daily report therefore shows cooling energy, electrical energy, aggregate COP and data coverage together. Compare like operating periods before inferring deterioration, and retain weather, load and setpoint context for a deeper investigation. The Chilled Water System Design Handbook is a related design reference; the calculations here are a standalone worked example, not a claim that the handbook automatically processes BMS logs.



Continue the calculation review



Sources and calculation basis


Primary references checked on 26 October 2026. The worked examples and diagrams are original illustrative calculations, not manufacturer selections or project approvals. Apply the stated limitations and the requirements governing the actual project.



Comments

Rated 0 out of 5 stars.
No ratings yet

Add a rating
bottom of page