Yes, the raw data was kindly supplied by CathodeRay, the accuracy of which I am not questioning.
My objective is to try to avoid the computer World adage of 'garbage in - garbage out'. I am therefore trying to ensure that the data going into the modeling tool, is the same as that being seen by the heat pump controller, when averaged over a 1 hour period.
The OAT temperature sensor is an analogue device, so within the heat pump control system will be an analogue to digital converter (ADC). I would hope that the controller is using 'real numbers' rather than integers within its algorithms, since a 1C difference in the OAT value, could equate to a 500W change in thermal energy output.
The fact that the OAT value derived over Modbus, does not indicate a value below 0C, would tend to indicate that it is being stored as an unsigned integer, which cannot hold negative values.
The fact that the OAT value derived over Modbus, does not indicate a value below 0C, would tend to indicate that it is being stored as an unsigned integer, which cannot hold negative values.
I think you have hit the nail on the head. Here is a read request to address 107 (OAT) in debug mode which also shows the actual response received:
MinimalModbus debug mode. Will write to instrument (expecting 7 bytes back): 01 03 00 6B 00 01 F5 D6 (8 bytes)
MinimalModbus debug mode. Clearing serial buffers for port /dev/ttyUSB0
MinimalModbus debug mode. No sleep required before write. Time since previous read: 2158867119.31 ms, minimum silent period: 4.01 ms.
MinimalModbus debug mode. Response from instrument: 01 03 02 00 04 B9 87 (7 bytes), roundtrip time: 0.0 ms. Timeout for reading: 50.0 ms.
OAT=4
Decoding the response:
01 = device address
03 = function code (read register)
02 = two bytes of data follow
00 04 = the data x00 x04 and x04 = decimal 4 => OAT=4 (but see below*)
B9 87 = checksum
@derek-m will be able to confirm, but I don't see any option for a signed integer in that response.
Midea often send values with decimal places as integers as decimal (10,100 etc) multiples of the number and note the fact this has happened in the modbus table eg flow rate of 1.43 will be sent as 143, which then gets divided by 100 to get the correct value. The entry for ambient temp has no such note:
* It is possible the first byte of this pair is the sign byte, 00 = +ve, 01 (or something else) = -ve, but I thought I had taken care of this. The read_register command in the logging script is ' ambient = instrument.read_register(107, 0, 3, True)' where True is shorthand for signed=True (parameters mean: 107: register address, 0 = no decimals, 3 = read register, True = signed=True).
It just may get cold enough tonight to test some of this using the debug script. It's already down to 1 degree from 3 degrees a short while ago.
Midea 14kW (for now...) ASHP heating both building and DHW
I have just thought of a register that I know has a negative number: the left hand end of the WCC OAT is set to -4 degrees C. When I read it without adding signed=True I get
MinimalModbus debug mode. Will write to instrument (expecting 7 bytes back): 01 03 01 0B 00 01 F4 34 (8 bytes)
MinimalModbus debug mode. Clearing serial buffers for port /dev/ttyUSB0
MinimalModbus debug mode. No sleep required before write. Time since previous read: 2164673467.90 ms, minimum silent period: 4.01 ms.
MinimalModbus debug mode. Response from instrument: 01 03 02 FF FC F9 F5 (7 bytes), roundtrip time: 0.0 ms. Timeout for reading: 50.0 ms.
LHWCC=65532
but if I add signed=True I get
MinimalModbus debug mode. Will write to instrument (expecting 7 bytes back): 01 03 01 0B 00 01 F4 34 (8 bytes)
MinimalModbus debug mode. Clearing serial buffers for port /dev/ttyUSB0
MinimalModbus debug mode. No sleep required before write. Time since previous read: 2164766748.60 ms, minimum silent period: 4.01 ms.
MinimalModbus debug mode. Response from instrument: 01 03 02 FF FC F9 F5 (7 bytes), roundtrip time: 0.0 ms. Timeout for reading: 50.0 ms.
LHWCC=-4
This is what the minimalmodbus documentation has to say:
"Some manufacturers allow negative values for some registers. Instead of an allowed integer range 0 to 65535, a range -32768 to 32767 is allowed. This is implemented as any received value in the upper range (32768 to 65535) is interpreted as negative value (in the range -32768 to -1)".
Thus 65535 is -1, 65534 is -2, 65533 is -3 and 65532 is -4 (and 65532 decimal is hex FF FC as seen in the first response).
But this, while helping me to understand things, doesn't clear up the OAT missing sub zero values...
Midea 14kW (for now...) ASHP heating both building and DHW
The WC Modeling Tool infers that Energy Supply will be a product of LWT and the capacity of the heat emitters, such that at a LWT of 50C, a greater quantity of thermal energy will be supplied than if the LWT is 30C.
It may first be useful if I explain some of the terms I have used and how they inter-relate with each other. Heating Demand is the same as Energy Demand and is derived from the Calculated Heat Loss for the building. As OAT reduces, Energy Demand increases. Energy Supply, as described above, can be higher or lower than Energy Demand, though there is some inter-dependency through the process. If Energy Supply is kept constant by maintaining LWT and DT at their present values, but then Energy Demand is increased, then the net effect would be a reduction in IAT until balance is restored.
To calculate the Duty Cycle, or likelihood of cycling, the Modeling Tool takes the Energy Supply value, the LWT value and the OAT value, and calculates the minimum energy output at which the heat pump can operate continuously. This value is derived from the manufacturer's operating data table.
If the Energy Supply value is greater than the minimum energy output value, under the given operating conditions, then it is assumed that the heat pump should be operating continuously. If the Energy Supply value is less than the minimum energy output value then the Duty Cycle is given by Energy Supply divided by minimum energy output.
Obviously this assumes that the heat pump in question can achieve the turn down specified in the manufacturer's data.
Since Energy Supply, LWT as well as OAT, are all used in the determination of the minimum energy output, which itself is not fixed, but varies dependent upon operating conditions, the actual OAT at which cycling is likely to commence is within a range.
A further factor is the desired IAT, since if this is reduced, heat loss will reduce, which in turn reduces Energy Demand, and thus Energy Supply, requiring a lower LWT, with the likelihood of increased cycling. Simples. 😜
The Modbus registers are two byte 16 bit long. When displayed in binary the left-hand end is normally denoted to be the Most Significant Bit (MSB), and you have guessed it, the right-hand end is the Least Significant Bit (LSM).
If Bit 15 is a one, that would probably indicate that the following bits constitute a negative number.
Actually I have just remembered that you need to use 2's compliment to extract the value. Using 2's compliment the 1's become 0's and the 0's become 1's, if I remember correctly.
My observation about the discrepancy of the onset of cycling between the model and the measurements was actually not questioning either. The fact the (real) heat pump cycles at all at 4C suggests that something either temporary or permanent is awry. Temporary factors could include solar gain, cooking or something else which temporarily reduces the demand. Permanent could simply be that it is a bit oversized. Perhaps the discrepancy is due either to one of those temporary factors or due to a mismatch between the modelled house loss and the actual house loss.
My first instance would be to check the (real) power output at which the real thing started cycling. If this is consistent with the manufacturer data then then next question becomes, why is this so low in the specific conditions. Only then would I think you are in a position to examine why the model hasn't reproduced this. But all of this is likely obvious to you anyway so I apologise if Im telling you something you already knew.
Indeed. The apparent cycling seen at 4 degrees and below isn't modulating cycling, it's defrost cycles, and by doing them, the heat pump shoots itself in the foot. Data from the last 24 hours:
You can tell they are defrost cycles because the LWT falls below the RWT. Energy in soars, COP falls, and energy out fails to keep up with the demand, and the LWT/RWT average instead of straddling the Set LWT falls below it, leaving the IAT a degree or two below the desired IAT. My heat pump is actually slightly undersized for these conditions: despite claims that there is plenty of headroom, there isn't. Actual output as buried in their engineering data by Midea shows that their 14kW heat pump running in ambients around zero and (set) LWTs around 50-55 degrees becomes a 11.5kW heat pump. At -2 OAT, my calculated heat loss is 12.4kW or thereabouts. The heat pump is fighting a losing battle.
The OAT can be seen falling briefly below zero between 0700 and 0800 this morning. This settles the missing sub zero OAT values question: there aren't any; when it goes below zero, the OAT values go below zero.
Something else also caught my eye: the steep rise in OAT as the setback starts. It looks improbable. But what I think we are actually seeing is a lowering of the OAT by the heat pump in the pre setback period, and then once the setback has started, the air around the heat pump rises to match the wider OAT. Once the setback is over, and the heat pump starts up, it again causes the OAT in the vicinity to fall again.
Midea 14kW (for now...) ASHP heating both building and DHW
Indeed. The apparent cycling seen at 4 degrees and below isn't modulating cycling, it's defrost cycles, and by doing them, the heat pump shoots itself in the foot. Data from the last 24 hours:
-- Attachment is not available --
You can tell they are defrost cycles because the LWT falls below the RWT. Energy in soars, COP falls, and energy out fails to keep up with the demand, and the LWT/RWT average instead of straddling the Set LWT falls below it, leaving the IAT a degree or two below the desired IAT. My heat pump is actually slightly undersized for these conditions: despite claims that there is plenty of headroom, there isn't. Actual output as buried in their engineering data by Midea shows that their 14kW heat pump running in ambients around zero and (set) LWTs around 50-55 degrees becomes a 11.5kW heat pump. At -2 OAT, my calculated heat loss is 12.4kW or thereabouts. The heat pump is fighting a losing battle.
The OAT can be seen falling briefly below zero between 0700 and 0800 this morning. This settles the missing sub zero OAT values question: there aren't any; when it goes below zero, the OAT values go below zero.
Something else also caught my eye: the steep rise in OAT as the setback starts. It looks improbable. But what I think we are actually seeing is a lowering of the OAT by the heat pump in the pre setback period, and then once the setback has started, the air around the heat pump rises to match the wider OAT. Once the setback is over, and the heat pump starts up, it again causes the OAT in the vicinity to fall again.
Thanks, that seems like a good analysis. My car does something similar to your heat pump. If I drive then park, when I return to the car the reported OAT is higher than when I left. After a couple of minutes driving it falls back again. Its all to do with the location of the sensor of course. Some heat pumps use a separately located sensor to drive the WC and to me this seems the sensible approach, albeit at a small additional cost.
Im sorry to hear your pump is slightly undersized (unless of course that was the design intent). Having said that you will likely be benefitting from the undersizing during most of the season!.
Unfortunately some manufacturers sticker capacities bear little or no relationship to their actual capacity. Vaillant's 7/7.5kW R290 unit is nearer 8.5kW (for conditions in the South East of England) and their next mode up has a similarly conservative sticker capacity. Daikin's sticker numbers (which, in fairness, they dont actually claim represent the capacity) are way over the actual (in their two most common ranges they appear to have taken one basic design and tweaked the control to 'peak' the output at different FT points). Midea's are no better it seems. Samsung and Mitsubishi clearly 'clamp' theirs, the capacity tables show far too many values exactly equal to the nominal for it to be anything other than a firmware limitation. Obtaining and reading the capacity tables is an essential design step!
Im sorry to hear your pump is slightly undersized (unless of course that was the design intent). Having said that you will likely be benefitting from the undersizing during most of the season!.
It was more design accident that design intent. Originally I was going to get a 16kW model, but a whole series of events led to a rush on Midea heat pumps, and when delivery day came along, there weren't any 16kW versions left, but they did have a 14kW one. I knew this was marginal but (a) I was at the end of a very long process that was doing my head in and (b) it was 14kW or nothing, as the grant window was about to close, and the lead time for 16kW models was months away. It has since transpired that the Midea 12 14 and 16kW heat pumps are identical from a hardware point of view, with output controlled by software, which in turn is controlled by a trio of dip switches. All I have to do is flip the dips, and I have a 16kW heat pump. I have been very tempted, but haven't done so, because of both the running costs implications (as you note, in one way I benefit from a lower output heat pump) and warranty implications (what if it isn't just a case of flipping the dips?).
It seems to be standard practice in the heat pump industry to quote performance based on ideal conditions, and ignore the fact that when the mercury goes down, heat pump performance falls even faster. This is in my view another MCS/regulatory failure - manufacturers should be required to highlight performance in arduous winter conditions, when output really matters, not on a sunny day in April.
Obtaining and reading the capacity tables is an essential design step!
Exactly, but almost no one does it! I suspect most Freedom supplied installers rely on Freedom's heat pump calculator which can get it wrong, and there is no way the average householder in the market for a heat pump is going to dig out the data from obtuse engineering handbooks, nor should they have to, even if they could make any sense of it. I only learnt (entirely thanks to this forum) about the engineering data after my heat pump had been installed...
Midea 14kW (for now...) ASHP heating both building and DHW
Indeed. The apparent cycling seen at 4 degrees and below isn't modulating cycling, it's defrost cycles, and by doing them, the heat pump shoots itself in the foot. Data from the last 24 hours:
You can tell they are defrost cycles because the LWT falls below the RWT. Energy in soars, COP falls, and energy out fails to keep up with the demand, and the LWT/RWT average instead of straddling the Set LWT falls below it, leaving the IAT a degree or two below the desired IAT. My heat pump is actually slightly undersized for these conditions: despite claims that there is plenty of headroom, there isn't. Actual output as buried in their engineering data by Midea shows that their 14kW heat pump running in ambients around zero and (set) LWTs around 50-55 degrees becomes a 11.5kW heat pump. At -2 OAT, my calculated heat loss is 12.4kW or thereabouts. The heat pump is fighting a losing battle.
The OAT can be seen falling briefly below zero between 0700 and 0800 this morning. This settles the missing sub zero OAT values question: there aren't any; when it goes below zero, the OAT values go below zero.
Something else also caught my eye: the steep rise in OAT as the setback starts. It looks improbable. But what I think we are actually seeing is a lowering of the OAT by the heat pump in the pre setback period, and then once the setback has started, the air around the heat pump rises to match the wider OAT. Once the setback is over, and the heat pump starts up, it again causes the OAT in the vicinity to fall again.
Interesting to compare the same time for mine. One thing that stands out is that your flow temps drop a lot further than mine during the setback.
My observation about the discrepancy of the onset of cycling between the model and the measurements was actually not questioning either. The fact the (real) heat pump cycles at all at 4C suggests that something either temporary or permanent is awry. Temporary factors could include solar gain, cooking or something else which temporarily reduces the demand. Permanent could simply be that it is a bit oversized. Perhaps the discrepancy is due either to one of those temporary factors or due to a mismatch between the modelled house loss and the actual house loss.
My first instance would be to check the (real) power output at which the real thing started cycling. If this is consistent with the manufacturer data then then next question becomes, why is this so low in the specific conditions. Only then would I think you are in a position to examine why the model hasn't reproduced this. But all of this is likely obvious to you anyway so I apologise if Im telling you something you already knew.
The reason that I mentioned the cycling was because of the fact that it makes doing an accurate comparison between theoretical and actual that more difficult.
The excellent raw data, at 1 minute intervals, clearly shows that rather than being kept constant, the LWT is performing a 'sawtooth', as the heat pump heats the water from the high 20's / low 30's to the high 30's / low 40's, then stops running whilst the water dissipates the thermal energy and cools.
Averaging the data over 1 hour periods, leads to the possibility of inaccuracies, which could accumulate in the final results. It is therefore important to try to minimise any errors.
I have various theories as to why there may be a divergence between actual and theoretical, the primary one being that in the same manner as most homes in the UK, the data was collected during a period of 'human infestation' 😋, which of course will have upset the balance in the 'actual's' favour. Obviously removing all the humans from the equation would not be practical or even desireable, so making an allowance within the Modeling Tool may be the best option. I believe that the average daily electrical energy consumption (unless used for heating), is in the region of 8kWh to 10kWh, most of which may probably be converted into thermal energy within the home.
What are your thoughts? Should an hourly additional heating value be added to the Energy Supply figure within the Modeling Tool? If so, should this be spread evenly over the 24 hour period, or when humans are likely to be more active?
The fact that the OAT reading appears to increase by approximately 5C during the setback period, would make me suspect that your heat pump may be sitting in a 'cold well'. Where is it located?
Could you please provide a copy of the raw data for analysis.
