GBA Logo horizontal Facebook LinkedIn Email Pinterest Twitter 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

Cycling of heat pump

davidsmartin | Posted in Energy Efficiency and Durability on

I have a Fujitsu heat pump with two indoor units each in relatively small rooms. (For good or bad, this decision was made because I was not satisfied if only one room had a unit that sufficient heat could be transmitted to the second room and I wanted each room to have its own thermostat. I recognize that they are larger than I need but I did not see good alternatives) Currently both units are set on “minimum heat” (50 degrees F) because the rooms are not occupied for days at a time.

When the temperature outside is zero degrees F, my power consumption is as shown in the first attachment. The graph shows whole house power consumption but the pattern is clear. The compressor comes on and power jumps by about 800 watts. After about six minutes, power jumps by another 500 watts for about three minutes. Then power drops to almost nothing for three minutes, then the cycle restarts. The entire cycle lasts about 12 minutes.

When the temperature is warmer, around 35 degrees F cycling is much more rapid. The compressor comes on, (power up by 600w), runs for one minute, Then the compressor turns off and there is a small draw probably for the fan for a minute. Then there is nothing for four minutes. The cycle lasts only six minutes. Note that the scale is different on the second graph, with the top of the graph 1 kw vs 2 kw on the first graph.

It is interesting that the two units seem to operate in tandem, both on the same cycle, sharing the output of the compressor when it is running. They never seem to get out of synch.

I assume the unit would be more efficient if it ran longer cycles. Is there any way to increase the difference between the temperature at which they demand more heat and when they turn off? Is this the way they are supposed to work?

The graph comes from my new “Sense” power monitor that is supposed to tell me how much power various things in my house are using. After a month it has not yet recognized the heat pump but hopefully it will figure things out eventually.

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. user-626934 | | #1

    This is “classic” Short-cycling on a mini-split. It’s doing what FUJITSU has programmed it to do. Are you using the wireless controllers?

    Side note: the Sense is not likely to ever recognize the widely varied power inputs of a mini split.

  2. davidsmartin | | #2

    I am using the standard wireless remote controllers.

    Why has Fujitsu programmed it with short cycles? Wouldn't it be more efficient with longer cycles?

  3. Calum_Wilde | | #3

    If the rooms are unoccupied try opening the doors between them and turning off one head.

    I'm still doing trial and error to see how low I can go, but it looks like my 12k btu Fujitsu is able to heat my whole 2400sqft house down to about 2-3°C with no solar gain. The temperatures from room to room are surprisingly even, but that might be a factor of how air tight the house is.

  4. user-1137156 | | #4

    This is classic short cycling. that results from over sizing How is the temperature being sensed? What indoor temperature change accompanies each cycle? Do these particular units have the "follow me" feature that senses temperature at the remote, If they do try using it it with the remote completely hidden from the unit, like under a bed. With the IDT set at 50f it is very probable even with correctly sized units. What is the IDT change per cycle? Cycling is inevitable close to the "set point"
    where even the best unit will rapidly supply too much heat. to make the cycles longer will mean greater excursion of IDT If you can't program them, you are "stuck" You could experiment, you said this is multi split with the IDT @ both heads set to 50f Try seting one to 60 observing it till it shuts down then re select 50, time that cycle.

  5. davidsmartin | | #5

    Jerry -- What is IDT?

    Fujitsu does not have the "follow me" feature -- the temperature is measured only at the unit's air intake.

    Calum -- I will wait until the outside temperature is above single digits F and then try running only one of the interior units with open doors.



  6. Expert Member
    Dana Dorsett | | #6

    Having the two units work in tandem is more efficient than independent cycling, since when only one of the heads is running it would have to run at a higher blower speed to be able to take the full output of the compressor. Even though the steady-state efficiency is quite high, it's spends a lot of each cycle ramping up, and is probably never reaching it's steady-state.

    If the room would stay at or above 50F with the doors open by convecting warmer air from an adjacent conditioned space, simply turning those heads would probably use less power (despite potentially higher average room temperatures, due to longer more efficient cycles & modulation on the head(s) serving the adjacent fully conditioned space(s).

    IIRC if you went with wall thermostats rather than sensing room temperature at the head, you would lose the MINIMUM HEAT function, but it would likely deliver longer more efficient cycles.

    David: I'm interpreting "IDT" as "InDoor Temperature", but perhaps Jerry will clarify that.

  7. user-1137156 | | #7

    Thanks Dana
    I apologize to all for using undefined shorthand!.
    IDT=In Door Temperature. ODT= Out Door Temperature COP= Coefficient Of Performance=Thermal power out/ (divided by) electrical power in. Note thermal power in BTU converts to watts by dividing by 3.412. units must match

  8. GBA Editor
    Martin Holladay | | #8

    Dana and Jerry,
    Remember, we are all striving, as much as humanly possible, to keep this website acronym-free. That includes neologisms like IIRC, Dana.

  9. Expert Member
    Dana Dorsett | | #9

    The convention in academic and journalistic writing is that when that sort of short-hand abbreviation is being used, at the first instance of usage it is defined in parentheses. eg, either:

    "With the IDT (indoor temperature) set at 50f it is very probable even with correctly sized units..."


    "With the indoor temperature (IDT) set at 50f it is very probable even with correctly sized units..."

    After which the abbreviated form can be used without confusion.

    Many abbreviations such as "BTW", "PS", "IIRC", and "ASAP" precede the wide world web (WWW), many even precede the internet in business & personal correspondence, and are not really neologisms (they are abbreviations). "IIRC" (if I remember/recall correctly) has been widely used in internet forums for more than three decades, well before the www, and while it wouldn't be used in academic papers or journalism, it is broadly understood in web forums, even web-forums in languages other than English. "IIRC" is not new, not freshly coined, and has been in common use on electronic forums even before many now-adult users of that abbreviation were even born.

    Shall we spell out BTU (British thermal unit) in discussions too? How about kwh (kilowatt-hour)?

    There are many newer commonly abbreviations that have arrived with text and instant-messaging, eg. IDK (I don't know) or U (you), that may be less appropriate on web forums but are used very seldom in a forum context, unlike BTW (by the way), AFAIK (as far as I know), or IIRC.

  10. GBA Editor
    Martin Holladay | | #10

    I'm an educated reader who studied English literature at Yale. As a student, I learned viz., ibid., op. cit., etc., et al.

    I also learned BTU and BTUH.

    As a builder, I learned OSB and HRV.

    Many of our readers don't know what an HRV or a WRB is, though, so I try to remember to write out “heat-recovery ventilatitor” or “water-resistive barrier” at first usage. Sometimes I fail, but I try to do that.

    I'm not so good at reminding readers what BTU and OSB stand for -- but I think that is not a problem. Readers can let me know if it is a problem.

    I got through my academic and professional decades without every learning what IIRC and LOL stand for. These are ugly abbreviations and I will never use them.

    Younger readers may think I'm odd, but I was raised on written dictionaries published in the 1960s and 1970s. Newer dictionaries include LOL, I'm sure, and IIRC, but not everyone knows what these mean.

    IIRC is particularly confusing to me, because it looks like IRC (International Residential Code).

    All of us who post comments here at Green Building Advisor hope that our sentences are easy for readers to understand. If readers scratch their heads at "IDT" or "IDU," -- or even "IIRC" -- there is no communication. Our attempt at communication has failed.

  11. user-4053553 | | #11

    @ Martin
    Just be glad you don't have young children today, or you would have to learn all sorts of new acronyms

    I tried to post two 30 sec youtube clips but the website says i triggered the spam filter so search these (separately) on youtube for fun



    [Editor's note: Here are the links]

  12. _Stephen_ | | #12

    @David: Do you know the relative heat load of the rooms and the size of the head units?

    I'm concerned because the heat pump I'm putting in has a minimum output of 18k BTU, and the design heat load of my 2050 sq ft home is 20 k BTU. I'm somewhat worried about short cycling. How oversized are you in this situation?

  13. Expert Member
    Dana Dorsett | | #13

    Stephen: If the MINIMUM output of the compressor at +47F is nearly the same as your design heat load, it's WAY oversized! If it's that high because you installed 5-6 compressors on seperate mini-splits rather than a multi-split, you'll be better off consolidating them into a single multi-split, or no more than two multi-splits even at a lower rated HSPF.

    Is this a ducted 1 or 2 stage old school heat pump?

  14. _Stephen_ | | #14

    Hi Dana,

    This is a Mitsubishi hyper-heat system. It's the smallest central ducted system they make at 30k BTU. It's just got a pretty poor turn down ratio. Min output is 18k BTU at any temperature.

    I'm hoping there's enough hysteresis in the system that the run times are reasonable.


    I nerd sniped myself. There should be approximately ~1700 lbs of air in my house, and the aggregate specific heat capacity for that will be around .557 MJ.
    There will be about 4400 lbs of drywall in the house, with a specific heat capacity of around 2.165 MJ
    There will be about 25000 lbs of concrete in the basement slab with a specific heat capacity of about 8.499 MJ

    This means that to raise those items of my house that I would consider to be pretty decently thermally coupled to the air, and the air itself by 1 degree kelvin, I'd need to produce about 11.241 MJ.

    At 18k BTU, this would take 35 minutes. Now, doubtless, i'm not going to get the whole thermal slab up to temperature, but also doubtless, i'm also going to be trying to raise the temperature of thousands of lbs of wood structure that I haven't taken into account. I think so long as the thermostat has 1 degree kelvin worth of hysteresis, I should be fine... I can probably also keep the run time in the units longer by keeping the basement set to a higher temperature to take advantage of all the thermal mass of the basement slab.

    Unless someone sees a flaw in my analysis?

  15. Calum_Wilde | | #15


    Re: abreviations

    I think it would be fair to say that your experience doesn't reflect that of the average user's. For what I assume the demographic is here, the technical terms are much more unfamiliar to read, and vastly more difficult to look up. Typing IIRC or LOL into a search engine gives the correct definition for those terms as the top hit. But something like WRB is much harder to land on the correct definition for.

    Maybe you're old, maybe you're not, that doesn't matter. But what does matter is if your language skills are stuck a generation behind. Language is an ever evolving and adapting, almost living, animal. You do yourself no favors by refusing to learn and adapt with it. One thing that is obvious about you is your passion and willingness to learn, even if it means tossing old ideas for new ones as research teaches us more. So why reject advances in language? None of us can afford to be so blind as to think that their generation had it right and everything else after is some how less than when it comes to every other aspect of what this site is about. And after all, you've been a major player in developing this site for many years now.

    So, again, why reject advances in language?

    I would postulate that common terms such as IIRC, LOL, etc. would be fine if they creep into posts more often. I doubt that would discourage the vast majority of users and they are quickly leaned by anyone that doesn't know them. I would continue to discourage "tech speak" though.

    P.S. Tone is not always easy to convey, so I'll say it plainly. I highly respected you and the efforts you put into educating us on all things green building. I in no way meant to seem snarky in my above post. Challenging, perhaps, but not from a place of disrespect.

  16. GBA Editor
    Martin Holladay | | #16

    Thanks for your comments. Here's where we agree: When the words of a writer are confusing to readers, the writer has failed to achieve the desired goal, which is communication. I think your point about technical abbreviations (like WRB) is a good one.

    I'm glad you reminded me that "technical terms are much more unfamiliar to read, and vastly more difficult to look up." To the extent that Green Building Advisor has been guilty of a lack of clarity, I'll try to do better.

  17. Jon_R | | #17

    > pretty decently thermally coupled to the air

    Over short time frames (minutes), drywall and slabs don't transfer very much heat to the air (I calculated it once). Heated air cycle times are mostly controlled by air volume and hysteresis. On the other hand, 1C is a more than typical swing.

    Adding a little insulation around a thermostat sensor will increase hysteresis.

  18. _Stephen_ | | #18

    If I'm just heating the air, and there's no heat transfer to anything else, my run time would be 105 seconds to heat a 2000 sq ft house by 1 degree Kelvin.

    For the more normal 60k BTU furnace that would typically be installed in this house, the minimum run time would be closer to 30 seconds. Given that these houses have neither 30 second runtimes, nor 10 degree overshoots, I must conclude there is significant heat transfer between the air and the environment. There's 1000 sq ft of slab that's decently conductive, and 6000 sq ft of drywall with which the air is in direct contact. That's gotta be soaking up at least a bit of heat on the order of minutes. Its it the full amount? Sure, of course not. But it's gotta be significant compared to the specific heat capacity of air which is genuinely tiny.

  19. davidsmartin | | #19


    Sorry for the slow response. The outside unit is a Fujitsu AOU24RLXFZH connected to two non-ducted ASU12RLS3. My calculation of the design load for a 60 degree temperature differential is that one room would need about 4K BTU while the other would need 5K BTU. The minimum output of the outside unit is about 6K BTU. So in my case, it seems that the unit should be able to run continuously at the minimum output whenever the temperature differential is greater than 40 degrees.

    I have observed rapid cycling when it is warm outside which is not surprising, though as I said, I wish there was a way to increase the differential between the starting and stopping temperatures. I am a bit surprised that even when it is very cold outside it still does not run continuously, cycling about five times per hour. Perhaps this is because of the need to defrost?

  20. Expert Member
    Dana Dorsett | | #20

    There is the defrost issue, but also the solar gain issue. During the day the load isn't as well coupled to outdoor temperature as it is at night, and windows can be supporting a large fraction of the load on cold but sunny days- even more so with higher performance houses.

  21. _Stephen_ | | #21


    I'm surprised it's cycling that much still. I hope my heat pump doesn't cycle as badly.


    You didn't want to wade into my analysis on how long my cycle time would be based on the specific heat capacity of the various bits? Rats.

  22. Expert Member
    Dana Dorsett | | #22

    The furnace model isn't a good one since the thermostat has a built-in (sometimes programmable) hysteresis of (typically) 0.5-1.5 F from the beginning of a call for heat to the satisfaction temperature.

    Ductless heads are trying to modulate the output to achieve to a tight range setpoint temperature, integrating the temperature sensor data over time. Some are using a proportional-integral or proportional-integral-differential approach so as not to ramp it up to the max output just to get to the setpoint quickly at the cost of lower efficiency or even overshooting the setpoint temperature.

    I have no direct knowledge of the specific control algorithms that Fujitsu I(or any other vendor) uses, but they would all cycle extremely rapidly and changing speeds wildly almost all the time if they weren't doing some math of the sensed air temperature.

  23. Jon_R | | #23

    Say a house has 7000 sq ft of interior surface and an average air film R value of .8. With an average delta-T of .75F, I get a heat transfer from the air into the surfaces of 109 btu/min. That's not much compared to a furnace producing 1000 btu/min.

    This is assuming no heat loss to the exterior (ie, 70F outside). But normally there is heat loss to the exterior and this reduces the furnace surplus output. I just noticed that the conductive component has the side effect of causing more convective airflow over exterior walls/floors/ceiling (which lowers the R value by perhaps 4x).

    If a furnace in the above case can modulate down to around 20K btu/hr, it shouldn't short cycle even if it is significantly over-sized.

  24. _Stephen_ | | #24

    Jon, I'll have a heat pump that can modulate down to 18k BTU / hr. Sounds like it won't be too oversized.

Log in or create an account to post an answer.


Recent Questions and Replies

  • |
  • |
  • |
  • |