But it doesn't (stay at very nearly the same temp, it drops at times, sometimes by several degrees for periods) and the general rule of thumb is I believe 10% energy difference for every 1 degree of IAT difference. I think we need to look closely at what the actual IATs are actually telling us.
Sorry but you aren't looking at the data correctly, you need to multiply temp by time. To achieve the claimed 20% saving overall (and using the 10% rule of thumb which is good at the temperatures in question) the temperature would have to drop by 2C for the whole 24 hrs, 4C for half of the 24hrs, or 8C for a quarter. I cant find anywhere in the data that it does anything close to that. If you can please point me at it.
There is also the thermal mass question (perhaps it is relatively huge?)
Indeed and that does explain why the reaction time to any perturbation may be quite extended. However this, if it were the only factor at play and if the 20% saving is correct, we would lead to the expectation of a progressive reduction in IAT as the trial progresses through several days, due to the ~20% deficit overall in energy supplied, with and the IAT settling at a new norm very roughly 2C below the IAT during the control period. Neither is observed.
Of course the thermal mass could be so huge that the progressive reduction is insufficiently large to be observed given the duration of the experiment, but that is, so far as I can see, inconsistent with the fact that the temperature drop while the heat pump is switched off is observable.
Actually we do have a reasonable estimate from now three more or less independent methods. I think we can say with a reasonable certainty that its somewhere between 10,000 and 20,000 kWh/C. Thus your daily apparent deficit of 6-7kWh would lead to a progressive drop of ~0.6-0.3C/day, which would mean that the progression should be observable during the experimental period.
Perhaps you aren't so used to dealing with systems that, given some observable (or derivable) facts, can be pretty accurately described quantitatively as well as qualitatively by known equations and numbers - my understanding is that human systems don't (yet) fit well into that category. However many engineering systems do and, whilst heating systems are quite complex, the underlying physics is very simple and extremely well established. So we should be able to get reasonably close to explaining whats going on inside the 'black box' and, if we cant, its because we lack relevant data. Until we can its simply not safe to 'extrapolate' to another system.
I'm quite suspicious that the diurnal range has a part to play in this. The control period has, as I have already pointed out, relatively few periods when the OAT matched the OAT during the setback. Furthermore the control period has greater (downward) diurnal range. Its entirely possible that the heat pump was playing 'catch up' through much of the day, thus increasing the energy demand at any given OAT (well intake air temp in reality) relative to that which would be required in a steady state. This could even be the case during the subsequent descent. I cant help feeling that the PHE may also play a part, as might solar gain, vampire loads (there remains the unexplained offset in the best fit equation which is the equivalent of a vampire load, even if its not actually a vampire load). More data may eliminate (or confirm) one or more of these.
This post was modified 1 year ago 5 times by JamesPa
4kW peak of solar PV since 2011; EV and a 1930s house which has been partially renovated to improve its efficiency. 7kW Vaillant heat pump.
CathodeRay's data provides the OAT and the IAT at hourly intervals. The model needs to adjust the energy output to represent what the heat pump controller would be doing, so that the IAT achieved by the model at the end of each 1 hour period matches the IAT value from the raw data. This would therefore provide a measure of how much thermal energy the heat pump must provide in that 1 hour period to achieve the required IAT. In an actual heat pump operating in WC mode, I believe that although the required LWT is set by the WC curve parameters and the OAT value, the actual thermal energy output is derived by a combination of the LWT-RWT DT, and the flowrate, the feedback being in the form of changes to RWT. This capability is not built into the spreadsheet, so to simulate the loading effect as energy is absorbed by the heat emitters, the WC offset parameters basically tells the simulation to output more energy. By varying the WC offset, the Energy Supply to the heat emitters is varied which in turn changes the calculated finish LWT value. I appreciate that it is a little complex, but it was the best way at the time that I could thing of how to vary the heat pump loading. If anyone has a better idea I am always open to suggestions.
...
If there is anything that you do not fully understand then please feel free to ask. That applies to everyone.
Thanks. So to check my understanding (please confirm if I have got it correct!): the modelled LWT is adjusted from its default value (set by an assumed WC curve - presumably based on the @cathoderay set up) by an amount shown in the second row above each table, in order to fit the model output to the measured IAT at the end of each period. This adjusts the (modelled) energy delivered.
If this is correct then just two more questions
- is the COP you then use based on the default LWT or the adjusted LWT?
- is the net result to tell us how much energy must have been delivered (according to the model) to achieve the measured IAT?
Many thanks, I'm looking through the data to see what it tells me.
CathodeRay's data provides the OAT and the IAT at hourly intervals. The model needs to adjust the energy output to represent what the heat pump controller would be doing, so that the IAT achieved by the model at the end of each 1 hour period matches the IAT value from the raw data. This would therefore provide a measure of how much thermal energy the heat pump must provide in that 1 hour period to achieve the required IAT. In an actual heat pump operating in WC mode, I believe that although the required LWT is set by the WC curve parameters and the OAT value, the actual thermal energy output is derived by a combination of the LWT-RWT DT, and the flowrate, the feedback being in the form of changes to RWT. This capability is not built into the spreadsheet, so to simulate the loading effect as energy is absorbed by the heat emitters, the WC offset parameters basically tells the simulation to output more energy. By varying the WC offset, the Energy Supply to the heat emitters is varied which in turn changes the calculated finish LWT value. I appreciate that it is a little complex, but it was the best way at the time that I could thing of how to vary the heat pump loading. If anyone has a better idea I am always open to suggestions.
...
If there is anything that you do not fully understand then please feel free to ask. That applies to everyone.
Thanks. So to check my understanding (please confirm if I have got it correct!): the modelled LWT is adjusted from its default value (set by an assumed WC curve - presumably based on the @cathoderay set up) by an amount shown in the second row above each table, in order to fit the model output to the measured IAT at the end of each period. This adjusts the (modelled) energy delivered.
If this is correct then just two more questions
- is the COP you then use based on the default LWT or the adjusted LWT?
- is the net result to tell us how much energy must have been delivered (according to the model) to achieve the measured IAT?
Many thanks, I'm looking through the data to see what it tells me.
Yes, you are correct. Whilst it is not what I believe happens in the real World, in CathodeRay's system he already has his version of AA mode varying the required LWT, and I also suspect that in the real World the actual LWT gets pushed up by the increasing RWT, which I seem to remember is evident in the 'minute' raw data previously supplied. So using changes in LWT to control heat pump energy output is not too far from the real World.
The simulation uses the LWT value to select the correct location within the manufacturers data tables to extract the likely COP value, based upon the LWT and the energy output values.
Here are the PHE temps I recorded yesterday evening. They were taken using a IR thermometer in contact with black masking tape on the relevant pipe about 3" from the PHE entry/exit point. These are not precision readings, just best available readings, and the reading is quite sensitive to the angle of the sensor to the pipe - did my best to keep it constant. Readings were taken every 20 seconds or so. The first two batches happened to coincide with the nadir of an off period in the cycle (heat pump was cycling at the time), the third with the middle of an ascending period in the cycle. First the raw plot, then the third batch crudely overlaid on the minute data plot for the relevant time:
Midea 14kW (for now...) ASHP heating both building and DHW
The third period, when the heat pump was actually running, appears to show
relatively little temperature drop across the PHE between primary in and secondary out: good
but also
relatively little temperature drop across the PHE between primary in and primary out, even though there is a significant temperature drop between sec out and sec in - bad
Basically it appears that your heat pump is continuously circulating hot water through the PHE, and only giving up a small proportion of the available energy to the secondary (and thus onward to the emitters). Essentially the PHE is out of balance.
This would almost certainly result is a vampire load in addition to the pump, but I cant currently work out how to estimate it, may well account for the failure to deliver sufficient energy to the emitters and would most likely increase cycling above what would occur if it could be better balanced. I haven't yet worked out how it relates to the question posed in the thread though, for that I need more thinking time. Reducing the pump speed on the primary would likely help, but of course as its fixed speed it may then be insufficient at low temperatures (on the other hand it may make the system work better at low temperatures because its better balanced).
This post was modified 1 year ago 2 times by JamesPa
4kW peak of solar PV since 2011; EV and a 1930s house which has been partially renovated to improve its efficiency. 7kW Vaillant heat pump.
Here are the PHE temps I recorded yesterday evening. They were taken using a IR thermometer in contact with black masking tape on the relevant pipe about 3" from the PHE entry/exit point. These are not precision readings, just best available readings, and the reading is quite sensitive to the angle of the sensor to the pipe - did my best to keep it constant. Readings were taken every 20 seconds or so. The first two batches happened to coincide with the nadir of an off period in the cycle (heat pump was cycling at the time), the third with the middle of an ascending period in the cycle. First the raw plot, then the third batch crudely overlaid on the minute data plot for the relevant time:
I would suggest repeating the test with a more reliable method of measuring the temperature, since if the readings were taken at 20 second intervals, I cannot see the temperature of the pipe changing by several degrees in such a short period of time.
It is always important to try to ensure that the most accurate and consistent readings are obtained before starting to analyse the results.
To be honest I was not going to spend any further time analyzing the data from your system, but your comments about your particularly bad day in late November intrigued me and piqued my curiosity.
I therefore put the hourly raw data into the spreadsheet to see what results could be obtained.
The first thing that I noticed was the profound effect that the heat pump appeared to be having upon the OAT sensor reading, rather than a few degrees difference, there appeared to be deviations of up to 6 degrees. Not only that, but, as suspected, the deviation appeared to vary with heat pump loading, the harder the heat pump was working, the larger the temperature deviation. Not an ideal situation.
I carried out the previous 3 table analysis with the results shown in the attached spreadsheet as tables 1,2 and 3.
Because of the large variance in the OAT sensor readings I decided to perform the task you hate the most, some 'whatiffery'.
Table 4 shows the possible effect of removing the shrubbery from around your heat pump and achieving correct airflow through and around the outdoor unit, though still allowing the IAT to fall from 19.2C to 18C. The predicted results would be a 16kWh reduction in electrical energy input and a marked improvement in COP.
Obviously you may not be too happy with the IAT falling from 19.2C to 18C, so Table 5 shows the predicted result of still removing the 'cold well' effect, but also maintaining the IAT at a more reasonable 19.2C. Still a reduction of electrical energy consumption of 5.3kWh and an improvement in COP.
Table 6 now shows the effect of removing the PHE from your system, which I had assumed was reducing the water temperature at the heat emitters by 5C. The results predict a total electrical energy reduction of 12.6kWh by removing some shrubbery and removing the PHE.
As an added bonus if the heat pump is not working so hard then it is likely it will not need to defrost so frequently.
Obviously this is all hypothetical, but I would urge you to consider what would appear to be tangible facts, if you wish to improve the overall efficiency of your heating system.
Is there any update on the estimated 5% energy saving on a night time setback?
Open query: do night time setbacks tend to happen during the coldest period of a 24 hour period, ergo the most inefficient time to use a HP?
Does latent heat mean setbacks can start earlier than bedtime?
How short or long is an efficient recovery period?
How much extra energy needs to be used in the recovery period to only produce a 5% saving?
The attached charts may help to see perceived benefits of a setback. The last chart shows the first hour of recovery on our system. The recovery show that it starts with a high deltaT of DT7 or 8 for first 40 minutes then modulation to around DT3 or 4 and with a flowtemp. Of 38c. The OAT is 4C at 6am. First step of IAT is 18.5C.
Is there any update on the estimated 5% energy saving on a night time setback?
Open query: do night time setbacks tend to happen during the coldest period of a 24 hour period, ergo the most inefficient time to use a HP?
Does latent heat mean setbacks can start earlier than bedtime?
How short or long is an efficient recovery period?
How much extra energy needs to be used in the recovery period to only produce a 5% saving?
The attached charts may help to see benefits. The last chart shows the first hour of recovery on our system. The recovery show that it starts with a high deltaT of DT7 or 8 for first 40 minutes then modulation to around DT3 or 4 and with a flowtemp. Of 38c. The OAT is 4C at 6am. First step of IAT is 18.5C.
-- Attachment is not available -- -- Attachment is not available -- -- Attachment is not available --
My reading is this (others may disagree)
There are currently no experimental results from which we can deduce any conclusions applicable generally, or indeed conclusions which can safely be applied to any system other than the one in question. Thus we are currently currently reliant on modelling alone. As @derek-m has done most of the modelling I will leave him to comment on his reading of the output from that.
Thank you for posting some experimental data, however measurements from a single day sadly tell us nothing with any degree of certainty; there is far too much variability due to other factors to reach a sound conclusion. If we wish to reach sound conclusions based on experiment, data would need to be collected over several weeks and in carefully recorded conditions, both with and without setback. If you are prepared to do this and post the results it would be very helpful indeed. Sadly heating systems are subject to several 'external influences' which make it impossible to reach robust conclusions on what is likely a small effect based on limited data. In summary this is a hard problem!
4kW peak of solar PV since 2011; EV and a 1930s house which has been partially renovated to improve its efficiency. 7kW Vaillant heat pump.
Is there any update on the estimated 5% energy saving on a night time setback?
Open query: do night time setbacks tend to happen during the coldest period of a 24 hour period, ergo the most inefficient time to use a HP?
Does latent heat mean setbacks can start earlier than bedtime?
How short or long is an efficient recovery period?
How much extra energy needs to be used in the recovery period to only produce a 5% saving?
The attached charts may help to see perceived benefits of a setback. The last chart shows the first hour of recovery on our system. The recovery show that it starts with a high deltaT of DT7 or 8 for first 40 minutes then modulation to around DT3 or 4 and with a flowtemp. Of 38c. The OAT is 4C at 6am. First step of IAT is 18.5C.
I fully agree with the assessment by James, home heating is a complex subject. Not because the Physics are particularly difficult, but because of the number of variables involved, many of which are not being measured, and also cannot be controlled.
So far there is no definitive answer, since the results I posted recently show minimum energy reduction or increase, under varying conditions. The problem is that the system from which the raw data was obtained, does not appear to be performing at optimum capability, so it tends to negate the usefulness of the results obtained. KevM provided some data a while back which I hope to use in the near future to perform more tests, but as James pointed out, ideally we could do with suitable data from a number of sources, so that proper testing and comparison can be performed.
I forgot to mention. One thing I may be able to do (when time, and she who must obeyed, permits), is carry out some 'whatiffery' that you may wish to specify, such as what effect would a 5 hour setback likely have under various weather conditions.
To be able to do such tests I would require the following details of your home and heating system.
Calculated Heat Loss.
Size and type of heat pump.
An assessment of the Thermal Mass/Heat Energy Capacity.
Whether the system contains a PHE, Buffer Tank, Low Loss Header and the like, with details of how these items affect the thermal energy transfer from heat pump to heat emitters.
The desired indoor air temperature, and any required variations on an hourly basis.
The hourly expected outside air temperature on which the assessment is to be performed.
Anything else I can think of later.
Oh, and your bank account details, including PIN and passwords. 😋
I seem to have most of that list but some extra work on existing thermal mass is needed. If I manage to get some energy data sorted I may come back with the above information. After which my bank account won’t be worth much at all...😉
Thinking about installing a heat pump but unsure where to start? Already have one but it’s not performing as expected? Or are you locked in a frustrating dispute with an installer or manufacturer? We’re here to help.
