Improving heat pump performance in cold / freezing conditions

After all the mild weather, this morning has brought horrible conditions for our heat pump - temperature around 0C, wet atmosphere, misty and frosty - and the performance is much lower.

It took almost 3 hours for the flow temperature to reach target. I’ve also seen long defrost cycles of 13 minutes, which is longer than I remember seeing.

As a result, our house isn’t as warm as I’d like this morning. I need to adjust settings to allow heat pump to come on earlier in cold weather.

Does anyone have any tips for improving performance in these conditions?

I’ve tried lowering the target flow temperature by 3C as I reason that stable operation at 36C might be better than working harder to get to 39C and defrosting often. (Obviously this might not be enough to meet our heat loss, I’ll nudge it back up later.)

Things I’ve considered:

  • lowering the delta from 5C to 4C.
    – this would increase pump speed and I figured this might reduce losses in my long outside pipe run (a bit like higher voltage / lower current reduces I^2R losses), however it would just increase the return temperature and so more heat loss on the return pipe run.

  • limiting the heat pump power
    – would lower fan speed and compressor power reduce frosting? Or would it just frost up more slowly while producing heat more slowly and therefore have no net benefit?

  • increasing target flow temperature: opposite to first post. Get heat into the house more quickly. Perhaps as well it would shorten defrost time if it has hotter water for the defrost?

  • re-enable the back-up heater?!

Hi Jonathan,

I am going to refer back to my experience with the 9kW Daikin when I first encountered defrosting with my original radiators.

At this time I was running at a fixed flow temperature of 42c and on the worst day it was heating for 20 minutes and then defrosting for 10 minutes.

It struggled to get to the requested flow temperature and was never going to get to the designed 50c at -2.3c outside.

At this time I had slightly bigger radiators than you have.

After I changed my radiators to about three times what you have I was able to run at 33c at -4c outside and the defrosting was much reduced but ice formed almost immediately the heat pump started again after a defrost.

Lower flow temperatures are the key here as well as the fact that my house only needed 4kW of heat at that outside temperature. I wasn’t running the heat pump hard, in fact I was able to run it at the minimum possible.

This won’t help you but I was amazed to wake up this morning and find that my new 8kW Daikin hasn’t needed to defrost at all even though the weather was prime icing conditions.

I mention it because it is the answer, big radiators that enable very low flow temperatures to provide the heat needed minimise or even eliminate defrosting.

My wife now understands why I went so big on the radiators, they are good when it is warmer but really pay off when it is cold enough that icing and defrosts are an isuue.

I can compare my new heat pump to other similar heat pumps and they are running at 35c to 40c flow temperatures and suffer defrosting.

Mine is running at 31/32c with 0c outside and no defrosting and not even the merest sign of ice on the evaporator.

I was truly shocked after the 9kW which iced up really easily and had a penchant for defrosting even without ice.

1 Like

Hello @squarepeg77 I noticed the heatpump was off overnight, is there an opportunity to increase night time setback temperature, or find a way to keep the heatpump ticking over over night so that it needs to work less hard in the morning?

While it probably won’t do miracles for performance it would help with comfort?


Hi Matt,
Yes, I looked up your 8kW system on heatpumpmonitor today and was impressed that you had not had any defrost cycles! Most of the 9kW units were performing pretty similar to mine, given variations in local temperature.

I do appreciate our radiators are small, but there’s not much more I’m willing to do. We did swap a K1 for P+ in one bedroom as that room was often a bit too cool and took too long to warm up.

I’ve been studying the ESPAltherma data (I know you aren’t using it) and I think one of the critical values is the “Heat exchanger mid-temp (R5T)” - if that is below zero, the pump is struggling. Get it above 3C or so and it starts to be pretty efficient. There is obviously some relationship between the requested flow temperature, the outside temperature and the heat exchanger temperature. Ask for too much when it’s too cold, and it ices up particularly quickly. I reckon if the target flow temperature is much much more than ambient temperature + 33-35C (not sure exact number) then defrosting is a frequent occurrence.

I dropped the target flow temperature by 5C this morning at 10am, to 33.5C, and managed to double the time between defrosts which seems pretty significant. But that temperature isn’t enough for our radiators to enable our house to gain heat at 1.5C outside.

It looks like another pretty cold night, so I’ve dropped the flow temperature and will run the pump longer and see what tomorrow brings.


Have you considered adding fans under your radiators for forced convection? There’s off-the-shelf solutions or you can just build them yourself from PC fans. You can get around 50-100% more power out of a radiator at the same flow temperature. I’m running 30°C flow at 0°C outside temperature which wouldn’t be possible without those fans.

Hi @TrystanLea,
I’m running a software thermostat / schedule in Home Assistant and it wasn’t allowing the heat pump to come on before 4am. That worked fine in the warmer weather but I got caught out today. I’ve changed that now to 2am and will see how it does tonight. (I’m not using the Daikin Madoka thermostat as it is “quirky”.)

My system has used 20.5kWh of electricity today over 15 hours of active heating. With a minimum power usage of around 900W on the 9kW Daikin, I could have ran it continuously since midnight and in theory used about the same amount of electricity! And as you say, that would have probably resulted in more comfortable temperatures this morning.

A certain amount of heat must be put into the house over the course of the day to keep temperatures within the acceptable range, and obviously this is mostly dependent on the outside temperature. There are many ways of delivering the heat - shorter time and hotter flow temperatures, or longer and lower flow temperatures. I had previously thought there was no point lowering the target flow temperature too much on the Daikin as they seem most efficient in the mid-to-high 30s C. For the past several weeks I’ve been using a shallow weather curve of 38C at -2C and 36C at 18C, then relying on thermostat control.

But I’m not seeing much short-cycling now, so I figure I should just ask for ~35C and let the “overshoot” allow the flow temperature to go up to 39C.

I might try and work out what the daily heat requirement is from the forecast weather temperatures and then see how many hours of heating are required to achieve that at a typical heating output.

We get some thermal expansion / contraction noise from our en-suite towel rail which can sometimes keep me awake, so I have been motivated to push the heating-on time as late as possible. But if the heating is on consistently then it’s not a problem.

Hi @Andre_K, yes I have thought of this. I actually tried a Nocuta USB PC fan very briefly but returned it as it was not PWM-able and was too noisy for my liking.

I’ve seen the Speedcomfort fans, but understand they don’t come on at low radiator temperatures. Is that right?

I might try again with a PWM-able PC fan.

Are you running DIY PC fans, or an off-the-shelf solution? If DIY, what diameter and number of fans do you find necessary for a given radiator size?


I have a number of different solutions with varying complexity because I just kept adding them to rooms iterating a bit over my approach.

I have one Speedcomfort and yes it only comes on at 30° and would never switch on for my flow temps but you can just unplug the temp sensor and it’s always on. But I still find it too noisy.

Then I have some 120 and 140 mm non-pwm fans combined with adjustable power supplies. If you run a 12 V fan at 7V it’s barely audible.

The most high-tech solution is an ESP32 where I use one pin for PWM to drive pwm fans. There’s also a temp sensor on there that switches the fans based on radiator temperature. I have either DS18B20 sensors or wireless IR sensors for that (MLX90614). Theres Esphome running on the controller so I can vary fan speed from Home assistant, switch the fans on/off etc.

I typically use almost the full width of the radiator. I just tie multiple fans together with cable ties and glue strong neodymium magnets on them with cyanoacrylic glue. This way they just attach to the radiators.

It’s definitely not the prettiest solution but it really works miracles. Depending on the quality of PWM fan you’re using they become inaudible below 30%.

If it’s of interest I’m happy to share pictures/schematics/code.

1 Like

Hi @matt-drummer, I see from your overnight data that your 8kW unit is not immune to defrosts either, although it’s still performing very well.

My experiments have confirmed what I think you’ve mentioned before: when the outside temperature is around freezing the Daikin 9kW defrosts on a pretty regular schedule every 23-24 minutes. I see your 8kW system has a fairly similar interval today.

I’ve varied the target flow temperature in the range 32.5C to 37.5C this morning and the defrost interval hasn’t really changed. So it makes me think the logic is fairly fixed and if outside temperature is below a threshold then defrost every 20-something minutes.

However, as expected a lower flow temperature does deliver slightly better COP.

I think as the temperature gets a little above freezing, lower flow temperatures might lengthen the time between defrosts, but I’m not sure.

Hi Jonathan,

Yes, it suffered today.

But for a slightly different reason this time I think.

I haven’t seen any ice but I wasn’t awake all night so it may have done.

It stopped just before 03.00 and then struggled to get going again.

I have this problem with it at the moment, once it is running in a stable state it keeps running.

I am sure it is not immune from ice and defrosts but I think I can make it better.

My problem is that at such a low flow temperature my radiators cannot deliver the heat produced at the high flow rates in the initial start up. The return temperature rises too much and as the flow rate drops the heating cycle sometimes fails as the flow and return are too close.

I am running with radiators and it has a fixed dt of 10c which I never get at these flow temperatures of course.

My solution with the 9kW was to limit the pump speed and I am going to try this on the new heat pump.

This stops the return temperature rising so much and results in a wider dt between flow and return through the initial phase of the heating cycle.

It might work, it might not.

The 9kW was certainly better like this.

My 8kW is quite new, I have no evidence or science behind this but I feel that there is a bit of a running in period.

I tried to run at less than 30c flow at the beginning but it wouldn’t run continuously like this, it failed every time but yesterday I tried again and managed to get it to start and run at 29c.

I learnt quite a lot just sitting and watching the 9kW during the initial phase of the heating cycle and worked out where the most inefficient parts of this were.

I always found this the worst part of the 9kW, it was painful at the start of a heating cycle and limiting the pump speed made it much better.

Of course, that is in my house with my heating system and aiming to run at low flow rates and as wide as possible dt between flow and return.

My 9kW did not defrost every 20 minutes in freezing conditions so I don’t think that there is any fixed logic in this respect. Once I had changed my radiators, limited the pump speed and got the flow temperature down it was pretty good.

Look at my data for 18 January 2024, the 9kW was icing up all day, it was immediate after defrosts but ran for much longer, I remember the day well as it was the day that proved my heat loss and after that I had conclusive proof that the 9kW was too big, not that I really needed any more proof!

Hi @Andre_K,
I would be interested in more information on your fan projects. Perhaps you could create a new thread for it?

1 Like