@bobtskutter - I think I get at least some of the theory but I am still not sure of the analysis:

(1) the PHE data recording was very crude. I can't vouch for the reliability of readings let alone validity (I use the medical terms, other disciplines use others). I think at the time I recorded it we took the view it was better than nothing, itself a dangerous assumption! I'm not even sure what were were looking for, possibly just some evidence of some heat transfer. You will recall the PHE is literally black box, no one knows what is inside it. The plumber who fitted it installed it the wrong way round, ie non-contraflow. That is the sort of level of understanding we were (and perhaps still are) dealing with.

Another problem is that although I am pretty sure the data were collected on 14th December 2023, which was a relatively mild day for the time of year (no defrosts), I don't know the times of the readings, meaning we don't know where in the cycle (the heat pump would have been doing 'normal' cycling at the time, ie using compressor on/off as part of its modulation regime) the readings were taken. It is possible, given the pattern in the the primary in, they were taken during the trough of a cycle, when the compressor was off. Here's some cycles from the early evening of the 14th Dec 2023 (from my modbus monitoring, which is reasonably reliable and so far as I know valid): 

You can see that for part of each cycle, when the compressor is off, there is no LWT (PHE primary in)/RWT (PHE primary out) delta t. Without knowing when in the cycle the PHE readinqs were taken, we are pretty much stuffed.

2. Running mode:

Posted by: @bobtskutter

↑

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).

It was in fact running in weather compensation mode, not fixed LWT mode. The set LWT doesn't vary much, but it does vary a bit, even during this stable OAT period. The LWT/RWT pattern you see here is bog standard normal for my heat pump at moderate OATs. Actual LWT cycles above and below set LWT, giving a mean LWT close to the set LWT. This is what happens in 'normal' cycling, defrost cycling is very different.

As I don't have a fixed LWT, that won't be the cause of the limited (8kW, the max I think it has ever put out (at -4 OAT, LWT 58 degrees) is ~8.6kW) output! We still don't have a clear explanation for the limited output. It may just be that is what a Midea 14kW comedy badge rating heat pump puts out at low OATs and high LWTs:

It is unlikely to be the rads are too small, if anything they are slightly over-size. The primary pipework (including the PHE) may be restricting the primary flow, but again unlikely. Or it may be the Midea control of the primary circulating pump. There is still an unresolved question as to why a variable flow rate pump with a PWM lead connected by and large (there are dips at the peak of each cycle) does not vary its flow. 

3. Other comments:

Posted by: @bobtskutter

↑

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

They will (40 degrees, say mean rad temp 38 degrees > IAT) but a mean rad temp of 38 degrees is not enough to keep the house warm in low OATs. In the most recent cold spell, I had to set the WCC to take the LWT almost to its limit (it reached 59 briefly, limit is 60). The did keep the IAT stable, but at a bit below the design IAT (actual IAT 17.6, design OAT 21 degrees. 

Posted by: @bobtskutter

↑

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

Unfortunately (and annoyingly!), no. It does have a push button control and LEDs and it is currently 'CC3' (constant curve 3) which in plain English is apparently constant speed, maximum setting.

4. Getting reliable and valid PHE temperature readings: I could do this, but it is not a trivial undertaking, and not cheap, perhaps £100 for standalone data loggers, a bit less for modbus sensors perhaps, bit a lot of wiring in to do and some programming to set it up. Going back to the original thread title, (where 'My Midea Heat Pump' should really be 'My Midea Heat Pump System' ie the whole thing), if we were sufficiently sure that the PHE was the main cause of the problem, the I could take it out of the circuit. But I would need a lot more certainty that we currently have, removing it will not be that easy (very tight space, both circuits have to be drained down, refill now open circuit with glycol treated water etc etc) and once out, getting it back in again if I want to do that - as I said, the available space is very tight.

My current compromise solution is to accept I have to run the heat pump at near it's operating limits in cold weather, and tolerate the fact it can't reach design temperatures, because in fact I can live with 19 degrees rather than 21 degrees (though 17.6 is getting a a bit low, but it was -3 to -4 degrees outside, which is below the design OAT of -2 degrees. I accepting this as a solution, I take operating limit to mean what is says, it is a fixed limit without constraints (such as it can operate up to 60 degrees, but only for 10% of the time), ie if it is -4 for days at a time, and my heat pump runs with a LWT of 58 degrees for days at a time, it is still with its operating limits. What operating at that limit does to my heat pump's COP (~1.5) and energy use (~100kWh/day) doesn't bear thinking about, except to say it doesn't happen very often.     

Some further info:

I happened to notice a bit over an hour ago my heat pump was in steady state (no cycling), as it often is at OATs of ~4 degrees, making it a good time to grab some PHE readings. I took five sets of readings in short order at ~1021, then again 15 mins later at ~1036, when it turns out the set LWT went down. Here are the means of the five readings superimposed on the chart of LWT etc for the hour 1000 to 1100. The bar chart at the bottom shows the energy in, energy out and COP (not brilliant!) for the hour to 1100: 

 Here is the raw data, with means and delta t values (and hoping columns will be preserved when posted!):

1020-1022					        	mean	delta t
Primary In	         47	45	48	49	49	47.6	
Primary Out	         44	44	45	45	45	44.6	3
							
Secondary In	        41	42	41	42	42	41.6	
Secondary Out	        46	44	44	48	46	45.6	4
							
1035-1037							
Primary In	        46	48	48	47	48	47.4	
Primary Out	        43	44	44	44	44	43.8	3.6
							
Secondary In	        40	41	41	41	41	40.8	
Secondary Out      	47	47	47	48	48	47.4	6.6

Hello, A quick reply, I need to go out but will be back later.

You said the plate exchanger was installed non-countercurrent.  Please can you post a diagram of how it's configured.  Draw something on paper then take a photo and post it.  I'd like to make sure I know which connection is which.

Please can you post a photo of the secondary pump so I see what vendor / model it is, then I can estimate the flow and check secondary circuit performance.

The ASHP data sheet you posted says it will output 11kW at 55C flow temperature, but it doesn't tell you what differential temperature is needs between LWT and RWT to make that happen.  I'm pretty confident your plate exchanger is acting as a bottleneck for heat transfer from the ASHP into the radiators.  If the RWT is too high (because the plate exchanger is a heat transfer bottleneck) the ASHP must modulate down to stay at the LWT.  This is what I mean by "your ASHP runs at a fixed LWT".  Sure, the LWT is set by weather compensation, but the actual control electronics set the compressor speed to hit that LWT, which limits the power output.  There are systems which run to fixed power and the temperature goes up and up - think standard immersion heater.

The most recent trend you posted shows the outside air temperature lags the change in LWT setpoint by about 1hr, this is most likely due to an averaging function being used in the controller to smooth out sudden changes in outside temperature.

You must have a modbus table to be able to get the data out of the Midea controller.  Look for "corrected air temperature" or "adjusted air temperature", that's what will be used for control.

One thing that is weird is the ASPH shuts down when the RWT matches the LWT setpoint.  Are you confident you're reading the correct LWT setpoint from the modbus table?  Is there an "adjusted LWT setpoint"?

Regards

Bob

@bobtskutter - thanks for your reply. Some answers:

Posted by: @bobtskutter

↑

You said the plate exchanger was installed non-countercurrent.  Please can you post a diagram of how it's configured.  Draw something on paper then take a photo and post it.  I'd like to make sure I know which connection is which.

It soon got corrected (I noticed it was the wrong way round, and mentioned it. It was fixed by rotating the pump through 180 degrees). It is now contraflow (primary in at same end as secondary out). 

Posted by: @bobtskutter

↑

Please can you post a photo of the secondary pump so I see what vendor / model it is, then I can estimate the flow and check secondary circuit performance.

I already know what it is, and have the manual for it. It is a Grundfos UPM3 AUTO 25-70 130 ZZZ. It's currently set to CC3 (Constant Curve 3).

Posted by: @bobtskutter

↑

This is what I mean by "your ASHP runs at a fixed LWT"

OK I get that, it runs at a fixed LWT as currently fixed by the set LWT set by the WCC. When the WCC changes the set LWT, a new 'fixed' LWT comes into effect. Maybe more clearly expressed as "your [my] ASHP runs to the current set LWT"?

Posted by: @bobtskutter

↑

The most recent trend you posted shows the outside air temperature lags the change in LWT setpoint by about 1hr

I'm not sure about this, consider this recent 12 hour period when there was a reasonable variation in OAT over the period. The Set LWT seems to be close to being in step:

Posted by: @bobtskutter

↑

...this is most likely due to an averaging function being used in the controller to smooth out sudden changes in outside temperature.

You must have a modbus table to be able to get the data out of the Midea controller.  Look for "corrected air temperature" or "adjusted air temperature", that's what will be used for control.

One thing that is weird is the ASPH shuts down when the RWT matches the LWT setpoint.  Are you confident you're reading the correct LWT setpoint from the modbus table?  Is there an "adjusted LWT setpoint"?

I do have the modbus tables, but can't see anything that looks like a corrected or adjusted OAT (but that doesn't mean it doesn't use one internally, it may do so but not make it available over modbus). Note also the OAT is in fact the Air Intake Temperature (AIT) because of where it is (inside the unit). 

Some annotated scanned pages from my modbus tables. Address 136 is the Set LWT (I'm pretty sure), T1 is the LWT, T1S is the Set LWT, calculated from the 'hydraulic module curve' which in my case is curve 9 which is the custom WC curve, with the end points readable and write-able at addresses 265 and 266. 

I do have an additional 'auto-adapt' script running which checks the actual IAT against the desired IAT every hour and if there is a discrepancy it adjusts the curve (via address 265 and 266). The adjustments are small, +/- 1-3 degrees, and there are checks to make sure the LWT is never set above 60 degrees.

Shutting down when RWT gets to set LWT - this isn't always the case (see plot above) and I wonder if it may be an artefact. What the control logic appears to do (when in 'normal' cycling mode) is allow the LWT to overshoot so the mean LWT is about at the Set LWT (don't know if this is by design or by accident). If it stopped at the Set LWT the mean actual LWT would be less than the Set LWT. It just so happens (this is conjecture) that this tends to mean it goes off at about the point the RWT gets close to the Set LWT. (Note: defrost cycles and steady state running are different, actual LWT gets Set LWT but doesn't overshoot).

That said, I think someone, very possibly @jamespa, might have once said the controls target the RWT? Not sure. JP will no doubt tell us, one way or the other.      

Hello.  OK, I've plotted your PEX data in the attached spreadsheet.  The data shows your PEX is pinched between Prim_in and Sec_Out.  That  means the flow to your radiators is as hot as it can get.  You need more flow in the secondary circuit.

 I'll look at the pump data in a minute.

It's possibly just a fluke that your ASHP is shutting down when RWT = Set_LWT.  It could be when LWT = Set_LWT+5 and you're getting a 5C DT between RWT and LWT.

Interesting way to fix the PEX being piped up incorrectly.  You've reversed the flow through your secondary circuit.  Do you have TRV's?  Are they designed for flow in either direction?

Bob

@bobtskutter - thanks will look at the spreadsheet tomorrow (supper calls!).

Posted by: @bobtskutter

↑

It's possibly just a fluke that your ASHP is shutting down when RWT = Set_LWT.  It could be when LWT = Set_LWT+5 and you're getting a 5C DT between RWT and LWT.

Yes, that is what I was suggesting (an artefact).

Posted by: @bobtskutter

↑

Interesting way to fix the PEX being piped up incorrectly.  You've reversed the flow through your secondary circuit.  Do you have TRV's?  Are they designed for flow in either direction?

Indeed. Simple, but effective. There are no in use TRVs (all fully open), I don't think the circuit itself cares which way the water goes round it. 

OK, I calculate a flow of about 1.9m3/hr in the secondary circuit.

This is the Grundfos web page for the pump data:

https://product-selection.grundfos.com/uk/products/up-oem/upm3-oem/upm3-auto-59C93602?pumpsystemid=2573797517&tab=variant-curves

You said CC3, that gives a pump differential pressure of about 3.8m, or 37kPa (assuming it's water), it would be slightly higher for glycol with a higher density.

Ignoring pressure drop in the pipework, each radiator would "see" 37kPa pressure drop across the inlet and outlet connections.

Looking at a K3 radiator from stellrad, a 37kPa pressure drop would require a flow OVER 1m3/hr through one radiator.

It appears as though you have a very high resistance to flow on the secondary circuit.

Are all the valves open?  (Even the isolation valves on the pump).

Is there a mesh filter on the inlet to the PEX?  (if there is, is it on the actual inlet, you reversed the flow at commissioning time).

You said the PEX was to isolate the primary from an old secondary circuit (oil boiler I believe).  It could be the PEX has plugged up with dirt on the secondary side and that's limiting the flow.

According to Grundfos, there is also a CC4 curve:

https://api.grundfos.com/literature/Grundfosliterature-6154149.pdf

Bob

I found an install manual for an MD16IU-008DW unit.  Can you find the value of dT1SH?

It might be in "ServiceMan / 3. Heat mode settings"

The manual says "the unit will switch off when T1 > T1S+dT1SH"

Bob

@cathoderay &@bobtskutter

This is getting interesting!

I was convinced from much earlier posts that the early hypothesis of the PHE being inadequate to transmit enough heat to the secondary was the key. Obviously your Midea’s control is too simplistic to cope with any load limitations.

But I recall JamesPa saying trvs were not all unidirectional some had arrows on top to show the right flow direction. Only ones which display <=> did both. Again something new to learn.

I really hope you get a solution!

Posted by: @bobtskutter

↑

This is the Grundfos web page for the pump data:

I don't think that is the same pump. Grundfos operate one of the most impenetrable identification systems known to man for their pumps, and their manuals are written by AI bots running on chips a few capacitors short of a full circuit. Any manual for a central heating pump that open with "This appliance can be used by children aged from 8 years and above..." tells you right away that you aren't going to get much sense out of it. 

More pragmatically, the problem is a lot of Grundfos pumps look very similar, but are in fact very different. The pump I have has two green LEDs and three yellow/amber ones, and has only CC1 to CC3, no CC4. I have this from a "Grundfos UPM3 AUTO 25-70 130 ZZZ Data Booklet":

 and as you can see, there is no CC4. A "Grundfos UPM3 AUTO 25-70 130 ZZZ Quick Guide" I have shows something similar:

I think you are looking at a pump designated AUTO L whereas my pumps is designated just plain AUTO. Go figure...

Posted by: @bobtskutter

↑

OK, I calculate a flow of about 1.9m3/hr in the secondary circuit.

I am not sure how you got this flow rate, maybe the heat transfer equation (kw = flow rate x delta t x sp.ht)? If so, to solve the equation for flow rate, you need to know the other three. We do know vaguely (the PHE temp measurements are not precise) the delta t (but it changes constantly) but how do we know the kW? However, if it is about 1.9m^3/h, then it is greater than the primary flow, which is mostly fixed at about 1.4 something, with periods when it drops to about 1m^3/h.

Posted by: @bobtskutter

↑

You said CC3, that gives a pump differential pressure of about 3.8m, or 37kPa (assuming it's water), it would be slightly higher for glycol with a higher density.

Ignoring pressure drop in the pipework, each radiator would "see" 37kPa pressure drop across the inlet and outlet connections.

Looking at a K3 radiator from stellrad, a 37kPa pressure drop would require a flow OVER 1m3/hr through one radiator.

Doesn't a simple sanity check tell us something is wrong here? If I have ten radiators, then that needs OVER 10m^3/h! 

Posted by: @bobtskutter

↑

It appears as though you have a very high resistance to flow on the secondary circuit.

Are all the valves open?  (Even the isolation valves on the pump).

Is there a mesh filter on the inlet to the PEX?  (if there is, is it on the actual inlet, you reversed the flow at commissioning time).

You said the PEX was to isolate the primary from an old secondary circuit (oil boiler I believe).  It could be the PEX has plugged up with dirt on the secondary side and that's limiting the flow.

All valves including the various isolation valves - pump, magnetic filter etc are fully open apart from the lockshields used to balance the system.

There is a magnetic filter, but so far as I know no other filter on the secondary circuit. Last time I cleaned the magnetic filter, at the stater of this heating season, it was clean. The secondary circuit is the one the old oil boiler used but I am pretty sure it is about as clean as can be because it had a pretty serious heads down no nonsense power flush shortly after the heat pump was installed (because there were concerns it might be clogged up) and it emerged apparently as clean as a whistle.  

Posted by: @bobtskutter

↑

I found an install manual for an MD16IU-008DW unit.

This is not in fact my heat pump (which is a MHC-V14W/D2N8-B), but there is a lot of overlap between the different Midea models, similar wired controllers and 'For Serviceman' manuals, and mine does have a dt1SH setting, and it seems to do the same thing, though the wording is more obtuse (this is from the manual I have, with a note on the values (Default 5, mine 5) previously added by me): 

which is I think a long winded way of saying it will turn off when the actual (T1) and set (T1S) LWT delta t is 5 or less. If the current LWT/RWT delta t happens to be 5, then it will turn off when the RWT reaches the Set LWT, but not because of that. The fact it may appear to do that is just a coincidence occasioned by the settings and current values.     

Posted by: @judith

↑

This is getting interesting!

I was convinced from much earlier posts that the early hypothesis of the PHE being inadequate to transmit enough heat to the secondary was the key. Obviously your Midea’s control is too simplistic to cope with any load limitations.

But I recall JamesPa saying trvs were not all unidirectional some had arrows on top to show the right flow direction. Only ones which display <=> did both.

Indeed. My PHE has been getting it in the neck from day one. Yet you may recall Freedom (in their 'Quick Installation Manual') said my system must have one, otherwise my warranty is toast. Another problem has been we don't actually know anything about it except (a) it was supplied by Freedom as part of their 'installation kit' and (b) its brand (Secespol) and approx outside dimensions (about 100 x 100 x 440mm). Oh, and of course, it must be fitted.

I have a Henry VIII policy towards TRV valves, off with their heads!  

Hello,
I'm not very good with using quotes from forum posts, so I'll try to use headings insted.

How did I calculate the secondary flow rate?
If you ignore energy accumulation within the metal work of the PEX and assume the same fluid on both sides with no phase change:
Energy reduction on hot side = Energy increase on cold side
Energy = M x Cp x DT
Energy_hot = Energy_cold
M_hot x Cp x DT_Hot = M_cold x Cp x DT_cold
Therefore:
M_cold = M_hot x DT_Hot / DT_cold
If I assume the fluid is water then density = 1, so:
flow_cold = flow_hot x DT_Hot / DT_Cold

Your most recent data doesn't have a primary side flowrate, so I'll assume it's 1.5m3/hr, which seems a typical number in the data you've posted so far. The temperatures you've measured are repeatable - you've recorded roughly the same temperature 5 times. So I'll use the average value of the temperatures.

1.5 x 5 / 4 = 1.875m3/hr

System pressure drop is high?
I looked up what I thought was the pump curve for your Grundfos pump. The chart tells me how much head the pump will generate for a given flowrate and speed.
As the flow in the secondary circuit goes up, the head required to "push" the liquid through the circuit goes up. The pump will keep increasing flow until it's discharge head matches the head required to push that liquid round the loop. The pump can only operate on it's pump curve and you told me it's curve was CC3, so that tells me the pump is operating at 1.9m3/hr and 3.8m head.

The next step is to ask "what flow give me 3.8m head through a k3 radiator?". The answer to that is "a very large number". So that means your secondary circuit requires a large amount of head (or "pushing force" if you like) to get the water to flow through it. Something is wrong.

I think you've looked at the wrong pump on the Grundfos website.
Ah, OK. I might have done. Do you have a pump curve for your pump, there might be one in the manual? However, normally central heating pumps are the same curves for the same pump model number. The letters denote some clever options, such as digital control, pulse width modulation control etc.

Lock Shield valves
You said you've balanced the system with the lock shield valves. I undestand why you've done this, but it's added extra head to the secondary circuit and that could be what's limiting the total flow through the secondary side. Can you open all the lock shield valves? That will minimise the head "loss" in the seondary circuit and allow the total flow to increase. That would then prove that you can get more than 8.4kW out of your 14kW heat pump.

I don't have any detailed information on the PEX.
You do. You've told me the vendor and the dimensions. Please can you confirm if those are the dimensions of the PEX or the outside of the insulation. I can lookup the details from the vendor if I know the physical dimensions. Do you know what size the connections are? 3/4", 22mm, 1", 28" - doesn't need to be acurate, just roughly.

More to follow, but I'll put that in another post.

Bob

Hello again,

I wonder what's going on here?

The system seems to have landed at an equilibrium point. LWT = Set_LWT and the compressor has slowed down. I wonder why the compressor then speeds up and the pump slows down. That's just wierd. The ASHP was working OK, then it stopped behaving. Which is where you started this whole thread from.

I'm convinced the reason you're radiators aren't getting hot enough is because your PEX is pinched and can't transfer any more energy. But that doesn't explain why your ASHP compressor suddenly speeds up.

You said you've got a script which automatically adjusted the weather compensation curve. That sounds like you've created some sort of "room demand override" - which is very cool. But could that be making the compressor speed up for no apparant reason?

Regards
Bob