GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter X Instagram YouTube Icon Navigation Search Icon Main Search Icon Video Play Icon Plus Icon Minus Icon Picture icon Hamburger Icon Close Icon Sorted

Community and Q&A

Measuring COP of Variable Refrigerant Flow (VRF) Cooling System

jameshowison | Posted in General Questions on

Is it possible to measure the COP of a VRF installation when it’s cooling?

I understand the inputs to be energy (watts) and the BTUs needed to maintain a set point on the thermostat.

Measuring watts is straightforward, but what options are there to measure BTUs supplied by the unit?

For heating one could use a portable resistance heater in similar weather conditions (or the oil use calculations if you have an oil furnace).

For cooling, though, any ideas? Would it make sense to use a non-VRF window unit for a period, measuring duty cycle multiplied by name-plate output?

Portable cooling unit is an option but duct leakage and radiation from the hot pipe would matter.

One could model the building but that’s pretty reliant on inputs and the models are far from perfect.

Anything more exotic, like how long ice takes to melt? Some sort of other phase change materials speed of change, while monitoring inside and outside temperature?

Can one estimate load by monitoring inside temps and outside temps in an unconditioned building and running some regression analyses? Feels like the same assumption and modeling issues from the models.

Do HVAC units have some log or way you measure what the outdoor compressor is putting out (like some log of the ‘clicks’ open that VRF systems were, combined with duty cycle)?

Why? Mostly I’m just curious, but also interested in accuracy of Model J and understand energy cost of ducts in the attic etc.

GBA Prime

Join the leading community of building science experts

Become a GBA Prime member and get instant access to the latest developments in green building, research, and reports from the field.


  1. Expert Member
    BILL WICHERS | | #1

    Cooling is just heating in reverse. It’s difficult to know how much energy is stored up in the mass of the materials in the room though. What I would suggest doing is using an electric heater to heat the room some fixed number of degrees, maybe 5 degrees. From the wattage of the heater and the time it takes to raise the temperature of the room, you can calculate how many BTUs were needed to achieve that 5 degree temperature rise.

    Now for cooling, run the air conditioner for a fixed period of time, maybe a few hours. Keep track of how much the temperature falls during that period of time. The air conditioner is removing roughly the same number of BTUs per hour as the heater was adding to achieve similar changes in temperature as measured in degrees over time. In this way, you are using the electric heater as a sort of reference standard to calibrate your measurement process, then using that resulting data to allow you to calculate the performance of your cooling system.

    While this won’t be perfect, it will get you pretty close.


    1. jameshowison | | #2

      Thanks Bill, great thinking! So we artificially raise the temperature a known number of degrees using a calibrated source (resistance heater), then remove that heat using the A/C (measuring the energy taken to do so).

      So if we saw (for example) the heater running at 1,000 W continually for two hours to raise the temp by 5 degrees F.

      And then if turn the A/C on and find that A/C runs at an average of 333W over a period of one hour to bring the temp back down to the original set point (and nothing much changed outside), then we would have an estimated COP of 6?

      It would be interesting to do this at a few different temperatures, but perhaps ideally late at night using the outside temp as the center point to take solar gain out of the equation.

      I think this will work well in smallish spaces (ideally one room), since spreading the heat around won't matter so much (taking duct losses out of the equation).

    2. charlie_sullivan | | #3

      Good thinking. I propose a modification that I think makes it easier and more accurate. Run the heater and the A/C simultaneously, with the A/C thermostat modulating to keep the temperature constant. If you do this on a day when the outdoor temperature is near the setpoint temperature, you have very little heat flow other than out from the heater, into the A/C. I'd also monitor humidity and make sure you aren't also gradually dehumidifying.

      The COP is just the ratio of the heater power and the average A/C power.

      1. Expert Member
        BILL WICHERS | | #4

        You'll never reach a steady state condition this way, so your attempt at improving accuracy will actually act to introduce additional uncontrolled variables. If you do what I suggested, and heat the room, then let it come back to equilibrium again, then cool the room, you have essentially calibrated thermal mass issues out of the system since you're heating up and cooling down the room with everything in it. All you need to do this is a thermometer and a clock, so nothing fancy.

        If you try to run the heater and the A/C together, you'll never get things perfectly balanced, and the thermal mass of everything will act to damp the response and make it harder to know what you're really measuring. Is that heat you're putting in with the heater heating up the mass of the drywall, or going right out the A/C at any particular instant? Remember also that you can't continously modulate the A/C or the heater, so they are essentially on/off devices. That also complicates reaching a steady state.

        By simply warming or cooling the room and timing how long it takes to raise or lower the temperature some fixed number of degrees, you have a reasonably controlled experiment. Trying to achieve the steady state condition where the heater and A/C are exactly matched sounds simple, but in reality its much harder to do outside of a completely controlled laboratory enviornment.

        As an example of a somewhat similar experiment, my designed but not completely built R value test rig for insulation samples does what you (Charlie) are describing, but it can control the heating AND cooling and do both fully variably. The way it works is to have a hot plate on one side of the insulation sample and a cold plate on the other. The system maintains fixed temperature on the cold side by modulating a set of peltier junctions that are water cooled. It modulates these with a variable voltage derived from a PWM waveform, since PWM'ing peltiers directly is inefficient. The system then controls the hot side by directly PWM'ing a set of power resistors used as heating elements. The R value is calculated by monitoring how much energy is required by the resistors on the hot plate to keep the hot plate's temperature setpoint stable. The cold plate doesn't measure it's energy, it just maintains a fixed temperature. The reason for this is that the peltiers are not very efficient, so they waste a lot of the applied power, but the hot plate's resistors can be considered to be 100% efficient. Thus, all of the input energy to the resistors either goes through the insulation to the cold side (which is how we can calculate the R value of the sample), or is lost to the enviornment (which is minimized by using a decent amount of insulation on the non-sample side of the hot plate).

        In my test rig, I can continously control the hot AND cold sides of the system, and there is no unknown mass to damp the response (the hot and cold "plates" are pieces of 1/4" thick aluminum plate). In a non-lab setup like a room in a house, there are a lot more variables involved, and anything you can't control has to be calibrated out.


        1. charlie_sullivan | | #5

          The problems you describe would be problems if the A/C cycled rather than modulating. But I think James it talking about one that modulates. You'd turn turn the heater on at constant power, and have the A/C modulate to maintain constant temperature.

          It's still true that there would be some heat going into local heating of the building materials and furniture near the heater, but the longer time ramp method has more problems like that. You are characterizing the ramp up and down as being described by one temperature, but the building material temperatures won't track the air temperature perfectly, and there's diffusion of the heat through the insulation a time scale measured in hours.

          Remember that the time constant of a closed-loop system is shorter than an open-loop system.

          Even if the A/C was needing to cycle, you can log those cycles and make sure they have settle to a constant rate and duty cycle. Then you look at the average power over an integer number of cycles, not over an arbitrary 1-hour period.

          1. Expert Member
            BILL WICHERS | | #7

            The loss through the insulation will be roughly the same in both directions, so it is effectively calibrated out in the method I described. If you have something that effects both "sides" of the experiment relatively equally, then they don't really effect the variable you're trying to measure, since they cancel each other out. That helps you here, since it removes one variable that you can't otherwise control.

            It's true that you could log what the system is doing, but you'd need to watch it over a much longer period of time to get an accurate result. If you force it to go 5 degrees over or under setpoint, you can get a quicker result. If it doesn't actually cycle during that time (basically have it run all out until the desired result is achieved), that's OK too, but you will have some overshoot that way.

            Ultimately you end up "close enough" here and should be fine, which I agree could also be done with your idea of logging the operation over a period of time, you'd just likely need a few days of logging to be equivalent to a few hours of forced operation. Over multi-day time periods, weather effects start to come into play too, which isn't as much of an issue over a period of just a few hours.


          2. charlie_sullivan | | #8

            The bit about logging cycles was just an aside. In this situation, the A/C can modulate, so it will run at constant power after an initial transient.

            No, the effect of the more complex thermal network is not nulled out by ramping up and down. The transient response of a system going from steady temperature to a ramp is different from the transient response of a system going from a ramp up to a ramp down. You could come up with modified procedures to account for that, but they will take longer to run.

            You also have a the potential of the heat driving moisture out of building materials which might then have to be condensed out of the air as the air is chilled back down.

  2. yesimon | | #6

    Best is to measure A/C vs space heater at steady state. See:

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |