Posted by: @tasos

↑

Every house needs a specific heat power to bring it to a set temperature. This means a specific water temperature. If you lower this temperature to be more efficient at the pump side, you will simply not attain the set temperature.

This is true, but what weather compensation does when set up correctly, and even better when tweaked by Homely or whatever, is match the flow temp to the heat loss which in turn is primarily determined by the OAT. Heat loss 10kW at -2 OAT, the heat pump (ideally, not all do, see many other threads) delivers 10kW using a higher flow temp, heat loss 5kW at 10 degrees outside, heat pump delivers less heat at a lower flow temp, and so on. Cold weather, warmer rads, warmer weather cooler rads, basically (there is more to it, but this is the basic principle) adjusting the heat delivered to match the heat lost, resulting in a steady room temp. 

Posted by: @tasos

↑

I don't think (b) is fixed flow temp. Where did you see this ? If this is the case, it destroys all the benefits of an inverter driven pump.

As I said, the Midea manuals are not easy to make sense of. I got my impression this is the case from the less than perfect terminology in the manuals, plus if I recall what actually happens when you set PRESET TEMP. in the wired controller. The left hand (heating side) then behaves more like the right hand (DHW) side. It displays either the room desired temp or the water flow desired temp, ie a fixed number. The only way to access automatic (weather controlled) variable flow temp is via the WEATHER SET TEMP menu (plus any other third party additions).

That said, I think I also recall you can set different (fixed) flow temps for different times of day. But that in effect is timing, the flow temp for each time period is still fixed.

Could it be that the inverter control is useful, in that it makes control of the flow temp when under weather comp control better and easier?

Posted by: @tasos

↑

@cathoderay I don't think (b) is fixed flow temp. Where did you see this ? If this is the case, it destroys all the benefits of an inverter driven pump.

I think you (for the avoidance of doubt by you I mean @tasos) may be misunderstanding what variable the inverter reacts to.  It changes the energy delivered by the compressor to the water in order to achieve the target flow temperature.  This is valuable on its own because it reduces cycling and enables variable flow temperature to be achieved more efficiently.  However the target flow temperature is still driven by something else - either weather compensation or a fixed set value, or something more complex depending on how commissioned.  

I don't know if Midea supports this, but certainly later Samsung models support a mode where WC is enabled (so its NOT fixed flow temp) but the internal temp sensor switches the heat pump on or off.  This is intended, I think to allow you to operate one of the 'tweaked WC' modes I describe above, where you set the WC curve (roughly) a degree or so above the 'ideal' and then the sensor switches on off (but as a consequence of the WC being almost right, with long on times) to achieve the desired room temp.  As I say above this means the 'heavy lifting' is still done by the WC but its 'tweaked' by the room sensor, thus relieving the installer of the need to get the WC curve set precisely albeit in return for a small (perhaps 3-10%) efficiency penalty, yet preserving most of the comfort benefits of WC.

@jamespa OK. Let me repeat that I started this post because I think the "weather curve" method is odd, outdated and wrong. Here is my reasoning:

1. It does not take into account the achieved indoor temperature and thus it is agnostic to the inhabitants needs. There is no way to make it work so that indoor temperature is constant and at a value that the inhabitant deems comfortable.

2. The PI controller on the other hand does not need any setup apart from the value of the indoor thermostat. It works by adjusting compressor speed so that achieved temperature is equal to desired temperature. How does it do this ? By constantly changing compressor speed (and thus water outlet temperature) as a function of the difference between desired and achieved indoor temperature.

I am not saying commercial heat pumps work this way. I am saying I will find it strange if they don't work this way, since it is technologically possible.

Posted by: @tasos

↑

@jamespa OK. Let me repeat that I started this post because I think the "weather curve" method is odd, outdated and wrong. Here is my reasoning:

1. It does not take into account the achieved indoor temperature and thus it is agnostic to the inhabitants needs. There is no way to make it work so that indoor temperature is constant and at a value that the inhabitant deems comfortable.

2. The PI controller on the other hand does not need any setup apart from the value of the indoor thermostat. It works by adjusting compressor speed so that achieved temperature is equal to desired temperature. How does it do this ? By constantly changing compressor speed (and thus water outlet temperature) as a function of the difference between desired and achieved indoor temperature.

I am not saying commercial heat pumps work this way. I am saying I will find it strange if they don't work this way, since it is technologically possible.

OK so we DO agree on the desirability of changing the flow temperature.  Thats the most important point.

AFAIK commercial heat pump controllers absolutely don't work in the way you describe in (2) based on room temperature (certainly none of the 10 or so whose manuals I have studied do).

I think we can assume that at least some of the manufacturers are familiar with PID controllers, and indeed incorporate them into their systems

Furthermore boilers have featured weather compensation (in addition to the smart room temperature control loops) for several decades (It was invented in the Netherlands I believe).

So we have to ask why, because they cant all be stupid.

I think there are perhaps several reasons.  History and simplicity of control loops may well be part of this.  However there is a more fundamental reason I think.  Room temperature is a rearward looking indicator (ie it tells you in arrears) of the energy necessary to keep the house 'system', which reacts very slowly, at a constant temperature.  As such it represents a large (and somewhat unpredictable) delay in the feedback loop and, as you doubtless know, delays in feedback loops can cause oscillation.  Furthermore the integral of the IAT offset vs target is not necessarily the amount of energy needed to make up the energy difference, because the OAT may have varied quite significantly in the interim and unless we also take that into account we are ignoring the main variable.

OAT on the other hand is forward looking indicator (ie it tells you in advance), and the integral of the OAT differential is the amount of energy needed to make up the energy difference, because we are assuming we want to maintain the IAT constant (or nearly so, and certainly more constant than the variations in OAT)

Thus a feedback mechanism (to control the variable - FT- which we both agree needs to change) based at least in part, on OAT, does have some distinct advantages that are not possessed by an IAT based feedback mechanism.

I suspect one has to look back at the papers around the time WC was invented to see the full analysis, but its not difficult to see why WC has come about and has become commonplace in boilers (albeit not in the UK, but in the more enlightened of our European neighbours) as well as essentially mandatory for heat pumps, and why this may well be a  good approach.

Since we agree on the desirability of adjusting the flow temperature, and since commercial heat pumps (AFAIK see above) only do this based on WC (or some slightly more sophisticated variant thereof), we are in any case left, I think, with the conclusion that WC (or an alternative more sophisticated flow temperature controller) should be enabled, even if it is supplemented by something else.

To your point numbered (1), many people, particularly those whose house has a high thermal mass in relation to loss (eg a typical eco house with a big slab and low loss) apparently find that WC alone gives good stability, so your statement is not universally true.  Where it isn't they may be prepared to tolerate the small loss in efficiency incurred by (eg) enabling WC and tweaking it using the inbuilt sensor or a dedicated controller. 

@jamespa Classical PI controllers need tuning i.e. commissioning. This is difficult to achieve, unless well-trained. This is why weather curves were employed. However, fuzzy PI controllers, known for some time, do not suffer from this requirement. I believe they are implemented in other machines (e.g. washers). Why they are not implemented in heat pumps, escapes me.

The problem with delays that you mention has to do with delay in information not the system itself, meaning that what the temperature sensor measures, is the current temperature. There is delay in changing this temperature, but this is taken into account in the controller. Every system has inertia, this does not prevent PI controllers to perform well. In any case, weather curve based controller, faces the same obstacles and performs worse (regarding the desired indoor temperature).

As a final point, outside temperature measurement greatly enhances the performance of a PI controller (faster response), by adding a feed forward logic, but this is not needed.

Posted by: @tasos

↑

@jamespa Classical PI controllers need tuning i.e. commissioning. This is difficult to achieve, unless well-trained. This is why weather curves were employed. However, fuzzy PI controllers, known for some time, do not suffer from this requirement. I believe they are implemented in other machines (e.g. washers). Why they are not implemented in heat pumps, escapes me.

They may be for all I know!  Perhaps 'adaptive' mode (or other manufacturers equivalents) use such controllers.

Posted by: @tasos

↑

 

The problem with delays that you mention has to do with delay in information not the system itself, meaning that what the temperature sensor measures, is the current temperature. There is delay in changing this temperature, but this is taken into account in the controller. Every system has inertia, this does not prevent PI controllers to perform well.   

AFAIK the problem with delay occurs if there is a lag between stimulus (in this case the energy delivered by the heat pump), and the response of the system (in this case the house warming up).  In the case of a house there certainly is.  Furthermore that delay is material in relation to the time which the external conditions (which are the principal driver of the house response) change, and the scale of the change of the external conditions is much bigger than any response we desire in a reasonably well controlled house.   I am perfectly prepared to believe that a controller based on IAT alone might be able to deal with this, whether it can while keeping the level of stimulus needed within reasonable engineering bounds where the emitters are of the nature used in our space heating systems, I do not know.  I find it fairly difficult to believe that, if it is possible, it would not require a differential component as well as PI, but am open to being proved wrong.

Posted by: @tasos

↑

As a final point, outside temperature measurement greatly enhances the performance of a PI controller (faster response), by adding a feed forward logic, but this is not needed.

Noted, but apparently all heat pump manufacturers and many boiler manufacturers currently think its useful.  As I don't design heating controllers I don't, beyond what I have said, have an insight into the reasoning, which may have many engineering niceties, may be cost based, maybe something else entirely, or more likely a combination of all of these.

For the benefit of those who aren't so interested in control theory as you and I, I think its important to reiterate what I believe we agree on, namely that varying the flow temperature  of a heat pump is, generally, an important part of controlling it efficiently and weather compensation (or variants thereof), whilst arguably not perfect, is the mechanism which heat pump and boiler manufacturers provide to do this.  Operating on a fixed flow temperature will, in almost all cases, certainly cost you quite a lot of money and reduce comfort.  

@jamespa There are numerous, very detailed, simulators that show the capabilities and benefits of each approach. Further, the delay you mention, is actually the time constant of the system and is much less than the rate at which outside temperature changes.

However, consider this: why don't, inverter driven, air-to-air heat pumps (i.e. classical air conditioners) use weather curves but work based on their internal thermostat ?

Posted by: @tasos

↑

Further, the delay you mention, is actually the time constant of the system and is much less than the rate at which outside temperature changes.

That may be the case in your house but its definitely not the case in mine nor in many others reported here.  Even with my current gas boiler (with mostly oversized radiators for the flow temperature - because my boiler only has manual WC) the house takes an hour or more to change in temperature by a fraction of a degree, whereas in that time the OAT can change by as much as 2 degrees and not infrequently changes by a degree.   As a concrete example when it recently got suddenly very much colder, the house took nearly 2 days fully to respond.  It was actually rather interesting watching the energy consumption and sensing both the IAT and the consistency of IAT across the room during this period.  This wasn't (principally) a function of the heating system, it was (principally) a function of the house, but of course its the house temperature we are trying to control (and the house temperature, albeit generally at one point only, that the control loop measures) so the house itself is an integral part of the control loop.

Posted by: @tasos

↑

However, consider this: why don't, inverter driven, air-to-air heat pumps (i.e. classical air conditioners) use weather curves but work based on their internal thermostat ?

 

As I say above I don't design controllers for central heating systems so I don't know.  Nor do I know if some A2A systems do feature some sort of OAT loop.  However, if I were to speculate, I would suggest that contributory factors (and I would expect there are others) may be

1. As you say A2A are classical air conditioners mostly used originally for cooling.  People done much care whether its 20 or 25 C indoors if the outdoor temperature is 35, they only care that its much less than 35.  The precision of the control is (or was) less important
2. Aircon emitters are frequently well oversized.  In my limited experience (3 systems in a semi-commercial environment) the oversizing is by a factor of 2 or more.  So they can and do have an effect more quickly than wet heating systems where the emitters are generally not oversized materially
3. Supplementary to 2, because aircon emitters blow air down into the room, the effect they produce is much more quickly sensed, well before the fabric heats up, so its les important to heat the fabric to temperature quickly

This is only speculation, as I say I don't design controllers for central heating systems!

Of course the question could equally be turned the other way round, and perhaps the 'balanced' way to ask it is 'why do air to air heat pumps (aka air conditioners) have control loops that are (usually) governed principally or solely by IAT, whereas air to water heat pumps and many boilers have control loops governed by principally (or possibly solely) by OAT?'.  

Posted by: @jamespa

↑

Of course the question could equally be turned the other way round, and perhaps the 'balanced' way to ask it is 'why do air to air heat pumps (aka air conditioners) have control loops that are (usually) governed principally or solely by IAT, whereas air to water heat pumps and many boilers have control loops governed by principally (or possibly solely) by OAT?'.  

But this is the whole point of our lengthy and interesting conversation. At this point, I owe to apologize if my English sounded a bit hostile, this was not at all my intention. Also I am not very familiar with the acronyms you use, so I may have misunderstood some points.

I will take this further with the manufacturers to see if they can give me a convincing explanation. By the way I recently installed a Midea R290 10kW heat pump, replacing my old oil boiler, so there is much to think about. Thanks for bearing with me.