Ask an operating engineer how their plant is doing and they will answer in kW per ton. $\mathrm{kW/ton} = \dfrac{\dot{W}}{T}$ — kilowatts per ton equals W-dot over T — where $\dot{W}$ is the electrical power the compressor draws, in kilowatts, and $T$ is the cooling it produces, in tons of refrigeration. Read the name and the algebra follows: kilowatts PER ton puts the tonnage underneath.

This is the one figure of merit in the chapter where lower is better, and that single fact catches more errors than any unit check. A modern water-cooled centrifugal runs about 0.5 kW/ton at full load; an old reciprocating machine or an air-cooled package will sit near 1.0 or above. If your answer moves the wrong way when the chiller improves, the ratio is upside down.

It converts to COP through the same constant as before: since one ton is $3.517\ \mathrm{kW}$ of cooling, $\mathrm{COP} = \dfrac{3.517}{\mathrm{kW/ton}}$. So 0.5 kW/ton is a COP of about 7.0, and 0.7 kW/ton is about 5.0. Note which constant did that work — 3.517, the tonnage bridge, not 3.412, the EER bridge. The two are three percent apart and they are not interchangeable.

One caution the datasheet will not print for you. A quoted kW/ton is almost always the FULL-LOAD figure at design conditions, and a chiller spends most of its life at part load with colder condenser water, where it does considerably better. That is why energy codes also quote IPLV — a weighted seasonal average — and why a plant judged on its nameplate alone is being judged on its worst afternoon.