@cathoderay To answer your question regarding this

Posted by: @cathoderay

↑

Max norm and min, in their output tables, but so far as I know no one knows how or when to apply the different levels

The inverter pcb contains a VFD (variable frequency drive) circuit that drives the 3 phase compressor motor with a synthesised three phase that varies in frequency from 25 Hertz to  a max frequency of about 75 hertz

The frequency will affect output, hence at 25 Hertz the output equates to the minimum value, norm is at 50 Hertz and max is at max frequency which i assume is limited to around 75 hertz in this application (although often goes upt o 120 hertz in many industrial application and right up to 500 hertz if application can take it).

The speed of the compressor is predominantly driven by the error signal difference of Set point flow temperature and current flow temp.This control signal gets modified by a PID algorithm that then is fed to the VFD.

To complicate matters futher the VFD signal is also constrained by evaporator and condenser temperatures.

From what I can ascertain the evaporator fan speed is derived from the compressor speed, and its the fan speed that massively determines the output of any heat pump; if it ain't running fast enough then defrosts occur sooner and more frequent. For this reason I would use a non contact tachometer to measure the evaporator fan speed at full load conditions i.e during a cold start of the heating system and compare with design spec.

The performance graph you have managed to plot would suggest to me that the cold weather 'underperformance" is due to you heat pump, and not primary or secondary pipework.

It would also be very interesting if you were able to plot s similar graph with the weather comp disabled to see what effect it has specifically on low temperature performance.

We know weather comp is a good thing to employ in warmer weather but what if...?

@bobtskutter, @iaack - thank you both for you comprehensive replies.

Posted by: @bobtskutter

↑

TL;DR I think you have the weather compensation set too high at low OAT, try dropping the upper setting to 50C. 

I don't in fact think this is the case. Recall the main problem is defective performance of the system at low OATs, ie at low OATs, the IAT falls further than it should. Look at the all important IAT on the chart (I think you have analysed the one I posted in the original post). From around 1800 on 2nd Jan, it starts to fall. Lowering the set LWT (by lowering the WCC) will only make that worse. If anything, the WCC and so the set LWt needs to be higher. I am using the current cold weather to test that - I currently have the left hand end of the WCC as high as I dare have it, and the early signs are that it has reduced the deficit, ie the IAT is more stable. here is the last 48 hours worth of data: 

I increased the left hand end of the WCC to 57 at -2 degrees last Wednesday (8th Jan), 57 degrees being as high as I dare go because I also have my auto-adapt script running, which has the potential to add another 3 degrees when the IAT falls too low, which would take the set LWT to its operational limit of 60 degrees. Look at the all important IAT line. Apart from the dip at 1300 on 11th caused by switching to DHW heating for an hour (a similar dip is just visible at the extreme left hand of the line, the IAT has remained remarkably constant at 17.6-17.77 degrees. Although this is both an MCS fail, and is lower that my preferred IAT of 19 degrees, at least it hasn't fallen, despite the extreme low OATs in the early hours of the 11th, when the OAT fell to -4 degrees, which is 2 degrees below design OAT in this part of the world (south UK). Maintaining this IAT needed very high set and actual LWTs, as can also be seen in the chart. If I were to lower that set LWT to say 50 degrees, then I have no doubt the IAT would fall, be several degrees.

Posted by: @iaack

↑

From what I can ascertain the evaporator fan speed is derived from the compressor speed, and its the fan speed that massively determines the output of any heat pump; if it ain't running fast enough then defrosts occur sooner and more frequent. For this reason I would use a non contact tachometer to measure the evaporator fan speed at full load conditions i.e during a cold start of the heating system and compare with design spec.

The performance graph you have managed to plot would suggest to me that the cold weather 'underperformance" is due to you heat pump, and not primary or secondary pipework.

Thanks for the detailed explanation.

I agree the heat pump itself is the most likely place to look for a cause. The problem is there is a mismatch between what is supposed to happen, and what actually happens, eg my system only varies the flow rate in limited ways. Freedom the Midea manual and I think you all suggest it should vary more, but it doesn't. Others who have posted Midea/Midea clone data seem to experience much the same behaviour. 

I'm still not clear how the min/norm/max settings work out in practice. Probably me just being thick. I get the compressor frequency has something to do with it. Is it that for each given OAT/LWT combination (in the Midea data tables) the compressor runs different frequencies? But that is not what we see on the charts, the compressor speed (dull red dotted line) is effectively all over the place (yes I know the chart is an unholy mess, even more so because I have added fan speed (divided by 10, ie 60 on chart = 600RPM) as I also collect that data, but it does give an overview of sorts):

It seems that in the week shown the fan is either mostly at 500RPM (left hand end, mild OAT) or mostly at 600RPM (rest of the time, cold OAT), ie any control is pretty crude, and doesn't seem to vary with compressor speed.

This is the problem again, what we observe doesn't fit with what we are supposed to see...

I can zoom in on any section of the above chart, if that would help. 

This is a very complex problem.

I know with a gas boiler increasing the flow temperature makes the device burn more gas, but how does a heat pump make it's outlet temperature hotter?  It would increase the compressor speed to increase the compressor discharge pressure, which would then increase it's temperature.  The high temperature (and pressure) discharge gas is cooled and condensed by transferring energy into the water circuit. The cool high pressure liquid is then depressured over the letdown valve where it vapourises back to gas but this time it's cold.  The cold gas passes through the air heat exchanger and warms up a bit by absorbing energy from the air.  If the cold, low pressure gas is too cold it will condense water vapour out of the air and that will freeze on the tubes.  Eventually the system will have to go into defrost mode which will reverse the refrigerant flow to take energy from the heating flow and put it into the air heat exchanger.  This is what's happening when you have a high leaving water temperature, your unit is freezing up and spending a lot of time in defrost mode.  My theory is if you lower the LWT the unit will spend less time in defrost mode and your average LWT will be higher.

COP is inversely proportional to LWT - OAT.  You get the best COP at the lowest DT you can see it in your data.  I really think you need to find the lowest temperature your radiators will work at to help reduce the number of defrost cycles your unit is going through.

Ultimately, I plan to install a heat pump to heat my garage and workshop space which is likely to be oversized.  So I'm very keen to understand the system performance and your data is very helpful.

Regards

Bob

Posted by: @bobtskutter

↑

This is a very complex problem.

I know! I have been working on it for ages!

I am reasonably confident my rads are big enough for the design LWT of 55 degrees. I checked and double checked. The LWT was deliberately set high (despite knowing about the unhelpful aspects of high LWTs) so I didn't need to have huge radiators (a problem because of both available space and aesthetic constraints). It was a trade off. We didn't under-size them for a LWT of 55, but at lower LWTs (and so lower room mean rad/room delta t's) they do become inadequate. Lowering the LWT will mean inadequate heating, which is what happened when I had the lower LWT before I increased it recently.

I'd be interested to hear what others have to say, but I suspect here in the UK we are pretty much stuck with defrosts in low OAT conditions, a combination of current heat pump design and our moist climate.    

Posted by: @cathoderay

↑

I'd be interested to hear what others have to say, but I suspect here in the UK we are pretty much stuck with defrosts in low OAT conditions, a combination of current heat pump design and our moist climate.   

Sadly I think you are right, until someone develops a cost effective no-frost coating which can be applied and prevents (or reduces) condensation build up by stopping the water adhering.  

Right, this carries on from this post

https://renewableheatinghub.co.uk/forums/renewable-heating-air-source-heap-pumps-ashps/ashp-for-small-garage-and-workshop/paged/3#post-40358

(Hopefully I got that right).

How can I sanity check the amount of ice accumulating in an 8kW ASHP?  I think the number is somewhere between 4 and 8kg.

Not really sure to be honest.  There is a comment on openenergymonitor.org that would suggest 4kg for a 10kW machine.  I tried to estimate the volume of the air heat exchanger based on the machine dimensions and get 0.044m3, so 4kg of ice would be about 10% of the volume.  8kg of ice would be about 20% of the volume.  The ice is likely to be on the outer surface so 4kg seems more likely to me than 8kg, having seen frosted up air exchangers on the offices at work.

I wonder, do you know the flow and return temperatures in your secondary circuit? see attached picture.

I need to know 5 of the 6 variables in that drawing to specify the plateHex.

I know LWT / RWT / primary flow.  If I know "to rads" and "from rads" temperature I can build a model of the plateHex and use that in the data analysis to simulate how the system behaves.

Regards

Bob

@bobtskutter In order to calculate the energy required to defrost the evaporator you are going to need some additional info including evaporator mass, composition and average temperature. Interesting  to know what estimates for these are you considering?

Posted by: @bobtskutter

↑

I wonder, do you know the flow and return temperatures in your secondary circuit? see attached picture.

Yes and no. There is no routine data collection as there is with the primary circuit, but I have done brief manual IR thermometer measurements (on black masking tape on the copper to get emissivity nearer where it should be). The trouble is there are four pipes to measure (because the primary temps here will be different to those measured at the heat pump by the Midea sensors) and only one observer! It doesn't help that the temps are constantly changing, when I measure pipe 4, pipe 2 that I measured some seconds ago will already now be at a different temperature. What I do is cycle round the four pipes recording temps for ten minutes or so. I'll look out the last set of measurements I did. From what I remember, the temperature drop across the PHE wasn't in fact that large.

I haven't got the dinky strap on temperature sensors from the likes of Amazon because they only display the temperature, they don't record it. I suppose I could set up camp outside the airing cupboard and take a photo of the sensors once a minute and then transcribe the results, but I think that might amount to letting my HDHD (Heatpump Deficit Hyperactivity Disorder) get the better of me!   

Ahhhh....

When your system is operating at it's equilibrium point the energy from the heat pump is released into the room by the radiators, but there is accumulated energy in the circulating water.  At the equilibrium point, the energy in the circulating water doesn't change.  The problem for CathodeRay is the temperature of the radiators isn't hot enough to warm the room up.  To get a higher radiator temperature you would need to increase the LWT even more and let the system find a new equilibrium point.  However, the plateheat exchanger complicates things because it creates a temperature loss in the system (that's a bad description, I know).  The plateheat exchanger means the secondary flow temperature is lower than the primary flow temperature (it must be to transfer energy through the heat exchanger), and that secondary flow temperature is too low to heat the rooms.  It's a perfectly viscous circle!

My calculation basis is to start from the equilibrium point, which in this case it's 8.4kW power output into the house and calculate the heat transfer performance of the system (which in my work language means build a dynamic simulation of the system).  Once I have a simulation of the system I can predict how the system would cool down if the heat pump wasn't operating.  In reality it cools down faster because the heat pump is taking energy from the system to melt the ice.  The difference between the predicted cool down rate and actual cooldown rate tells me how much energy is being used to melt the ice.

But I can't do that because the plate exchanger changes the system dynamics and adds a lag to the system. 🙁  I need to think some more on this.

Bob

Posted by: @bobtskutter

↑

It's a perfectly viscous circle!

Yes, this and the paragraph it ends is a good description. I might emphasise the secondary circuit mean radiator temp is too low, the rads can't shed enough heat a bit more. By the way, I see the PHE as a sort of choke that resists transfer (flow) of heat. You therefore need a higher primary LWT than would otherwise be necessary to get an adequate flow temp in the secondary circuit. It is this higher primary LWT that causes the performance hit.

Posted by: @bobtskutter

↑

I need to think some more on this.

Naturally! Having to do this is it seems a rather common experience when trying to understand what actually happens - as opposed to what people think happens - with heat pumps. 

@bobtskutter - here are the PHE in and out temps, taken in Dec 2023. They are NOT precise measurements - taken manually using an IR thermometer pointed at black masking tape on the copper pipes about 2" from the PHE. Unfortunately I don't seem to have any record of the time they were taken, only the date, so we can't marry them accurately to the modbus data. However, I can say (from the modbus data) the OAT (AIT) was likely to have been between 5-8 degrees when the readings were taken.

@cathoderay thanks for sharing that data.  There's not much numerical analysis I can do with it because I need to know 1 flow.

Please let me talk (type?) a little about heat transfer between different fluids.  Apologies if you know this already.

"Energy travels down hill" - thermal energy can only travel between two fluids if there is a temperature difference between them.  A good example would be a saucepan of boiling water sat on the hob would boil at 100C - i.e. the liquid water is turning to steam.  If you put another container in the boiling water and fill that with water, the second container will not boil.  It's temperature will increase and the differential temperature between the two liquids will go down.  When both liquids are the same temperature (100C) there is zero differential temperature, so no energy can flow to actually turn the water to steam.

"You can not heat something hotter than the primary inlet temperature" or "you can not cool anything below the secondary inlet temperature".

Please look at data points 20->35

There are times when Prim_in = Sec_out or is very close to being zero, and Prim_out - Sec_in is quite large.  This means your radiators are dumping a lot of energy into the room, because Sec_out - Sec_in is about 6C.  But your plate heat exchanger has run out of differential temperature to drive more energy into the secondary circuit, because Prim_in = Sec_out.  In heat transfer language it means the heat exchanger has "pinched out".  https://en.wikipedia.org/wiki/Pinch_analysis

It gets worse because your heat pump is set to control to a fixed LWT - this might explain why your heat pump is putting out about 8kW instead of 14kW (previous data analysis).  Increasing the LWT will increase Prim_in - Sec_out and that will get more energy to flow through the plate heat exchanger, but it will hurt your COP.  We need to increase Prim_in - Sec_out without increasing Prim_in.

It looks like your rads will operate with a flow temperature of 40C (data points 26 to 35).

Assuming your radiators are large and can dump a lot of energy if you increase the flow on the secondary circuit you will reduce Sec_out, because there is more mass flowing through the plate exchanger to take the energy, that will increase the overall energy flow but at a slightly reduced "radiator temperature".  You don't need to reduce sec_out by much to get increased heat transfer, 1 or 2 degree's should do it.

Do you have a variable speed pump on your secondary circuit?  Can you increase it's speed?

I know I've gone on a bit, but having the plate heat exchanger "pinched out" will stop energy flow.  That might explain why the LWT goes so low during a defrost cycle, because the plate exchanger has "pinched out" due to not enough flow on the secondary side.

I've reattached your data with some comments.

Regards

Bob