I’ve not used auto adaptive. Is it true that it uses the weather compensation curve as it’s source of information and some how enhances it by reading room temperature and modifies the flow temp if it needs to, to help bring room up to target temp?
Any other information on it would be welcome. Eg does it help for a radiator system? Does it automatically select the main controller thermostat? Does it affect COP etc? Sounds useful.
It has just occurred to me that if a Mitsubishi video states that the auto adaptive feature uses the weather compensation curve as it’s base source of information then one possible cause of problems for some installations might be that the compensation curve might not have been modified to match the heat loss of the home. Having said that there seems to be various other settings and factors that may affect the “learning” of this tool.
@kev-m I've double checked and MELCloud returns "DipSwitch5": 2 which would suggest that it is turned on.
Anyone run with this for a while and seen it adapting to be a bit more controlled in how it tries to meet room temperature?
Would be really good if someone had graphs to show this happening to give an idea of how long it takes and what it does.
I've started it on our office a few days ago and am using MELCloud to schedule a setback of the temperature at 16:30 and a few increases in temperature in the morning to get back up starting at 06:30. Daytime temp is 21C nighttime and weekend is 19.5C.
Here is a graph of what happened when is switched from WC to Room Temp (Note the actual value for energy consumption is too high as I wasn't converting from the Compressor Frequency properly at the time):
And here is what has been happening since (Note the last who on periods now are using the correct multiplier to get the consumption value):
Not that all my data is pulled from MELCloud and subject to its estimations and low frequency of updates.
I did adjust the day time temperature up a little yesterday.
I think our Ecodan 8.5kW is maybe oversized for the very well insulated and airtight office so I would expect some on off behaviour in milder weather but currently in the auto adapt mode it flow temps are spinning up to the mid 40Cs before dropping down for a long time before the next big spike.
I have the adaption interval set to 40mins currently. Haven't really gotten my head around whether this could be too long or too short. There probably is low thermal mass in the building due to Internal Wall Insulation and stud dividing walls but very low heat loss.
I'm afraid I'm at the same stage as you; I'm only just trying Auto adapt. I know @mjr has more experience so maybe they could comment?
It looks like your office is well insulated and doesn't lose heat very fast. How does the comfort (room temperature) compare with the different approaches?
I would think the longer interval times (your 40 minutes) are aimed at ufh rather than radiators.
It does look from the first graph that 35 deg or so is enough to heat your office. You could limit the maximum to somewhere in the 35-40 range, which would stop the 40 plus degree spikes, heat the office slower and maybe let the ASHP run longer? I assume you have the FTC manual.
@kev-m Yea, I might try reducing the maximum flow temperature (although don't want to do this too much as it may need to go higher when really cold).
Interesting about the Interval value, maybe I should try lowering this for our more reactive building and system.
In terms of comfort, the intermittent blasts of heat are noticeable, especially when there is a large gap between them. When on WC the heat was more even.
I'm going to leave things for another couple of days to see if it starts to dial down a bit before I fiddle with any settings.
Anyone run with this for a while and seen it adapting to be a bit more controlled in how it tries to meet room temperature?
Well, when I started this attempt, it was keeping temperature controlled but still cycling a bit and nowhere near achieving the published efficiency (gold line on the top graph):
I've been running it a few weeks and here's the sort of efficiency it achieves now, often pretty much right on the book value:
Interesting about the Interval value, maybe I should try lowering this for our more reactive building and system.
Yes, if your building is reactive, you may find that a shorter interval works best. I'd still be surprised if many homes react quickly enough for effects to be noticeable in 10-20 minutes which I suspect is needed for the default 10 minutes to work.
Also @mjr how are you getting on with auto-adaption? Is it still working efficiently now that you have changed the interval? Is it working better than any of the other control methods you have tried?
"Better" in that it's more efficient and less confusing to other inhabitants than having the controller flip between thermostat and curve modes. If only the controller allowed more than four step changes in temperature, there would be little for the RPi to do except monitoring and handling override requests... but for now, it also ramps the temperature up step by step in the morning.
I’ll let it run for a bit longer. Think I’ll dial the interval down to 20mins and see what happens. Also might reduce the max flow temperature down to 45 degrees. There are a few sun zero days coming up so would be good to see how it performs then.
I’ll let it run for a bit longer. Think I’ll dial the interval down to 20mins and see what happens. Also might reduce the max flow temperature down to 45 degrees. There are a few sun zero days coming up so would be good to see how it performs then.
Personally I would suggest increasing the time interval to 60 minutes, since this is probably the more stable starting point.
I’ll let it run for a bit longer. Think I’ll dial the interval down to 20mins and see what happens. Also might reduce the max flow temperature down to 45 degrees. There are a few sun zero days coming up so would be good to see how it performs then.
Personally I would suggest increasing the time interval to 60 minutes, since this is probably the more stable starting point.
Surely 60 mins will see a fast -reacting home overshoot and undershoot repeatedly?
I’ll let it run for a bit longer. Think I’ll dial the interval down to 20mins and see what happens. Also might reduce the max flow temperature down to 45 degrees. There are a few sun zero days coming up so would be good to see how it performs then.
Personally I would suggest increasing the time interval to 60 minutes, since this is probably the more stable starting point.
Surely 60 mins will see a fast -reacting home overshoot and undershoot repeatedly?
The functionality of the Ecodan controller appears to be somewhat similar to that of a PID Controller (Proportional Action, Integral Action, Derivative Action).
Weather compensation acts in a similar manner to Proportional Action, or Gain as the Americans prefer to call it. By varying the slope of the WC curve it adjust the response of the controller to variations in outside air temperature. This helps keep the indoor temperature fairly constant when the settings have been optimised.
It would appear that the Integral Action function, referred to as Reset by the Americans, is being performed by Auto Adaptation. In a PID Controller, Integral Action senses the error between the temperature setpoint and the actual temperature, and then adds to, or subtracts from, the controller output at a rate dependent upon the controller response settings.
The way that I interpret Auto Adaptation is as follows:-
If the setpoint is at 21C, but the actual room temperature is 20C, the controller would be measuring an error of 1C. The LWT is 35C.
The Auto Adaptation function may be programmed to add say 10% of the error value to the controller output (which is setting the required LWT), over a period of time set by the Interval Timer within the controller. If the Interval Timer has a setting of 10 minutes, then over a 10 minute period the controller output would be increased by 0.1C. So after 10 minutes the required LWT has been increased to 35.1C.
This rate of increase in the required LWT will continue while the error is 1C. It could take quite some time before the increase in LWT actually causes the indoor temperature to start to increase, and if the overall response is quite slow it could cause the LWT to exceed that required for an indoor air temperature of 21C. The controller response is therefore too fast for the home in question.
If the Interval Timer is instead set at 60 minutes, the response of the controller will be much slower, and hence is less likely to cause a temperature overshoot and temperature cycling.
This is my assumption, since I cannot look at the software within the controller, but your own experience would seem to provide confirmation, since increasing the Interval Timer on your own system has indeed slowed down the response and reduced the cycling.
I therefore suggest that when setting up Auto Adaptation, start with the Interval Timer set to 60 minutes, then reduce the time setting until the indoor temperature starts to cycle, the timer should then be set to the next higher value.
