This may be old news, so apologies if I am going over old ground here, but I was at a conference today where a new startup company was presenting on their new zwave based intelligent learning heating control system. It appears to use zwave motion sensors and a control unit to learn household behaviour to run the heating in the most efficient way. In isolation, its not of great interest to us yet as they have not opened their API (though plan to I was told). Anyway, what got my attention was it turns out they have repurposed Danfoss TRVs for their system and have reverse engineered them somewhat to allow them to integrate them into their system. As a result they had some insights into the behaviour of the units that I thought might be of interest to people.
Apparently the key to some of the unpredictable behaviour is the learning feature. The way this works is that in the first week of being switched on (if you set the time) on Thursday at 11 (AM or PM I cant remember which) the unit will attempt to learn the exact point at which hot water begins to flow through the valve. There are 3 temperature sensors in the unit, a high and low sensor and room temp sensor (which as we know, is not something you can see). The issue we have in UK and probably other countries is that the design and self learning feature is based on the origin countries system of always hot water in the system. So, if the unit attempts this when the heating is not switched on, it will fail to set itself properly. Danfoss say that after a week it will stop trying, but it strikes me that if the unit does not set itself up properly, this could cause unpredictable behaviour further down the line. I think it would be worth someone testing this theory:
- Ensure that the heating is on when the unit is first powered on
- Set the unit to 11 on Thursday to see if this is indeed when it wakes up to try and learn… u might have to try AM and PM.
I have no idea what impact (if any) this might have on it’s responsiveness, and I don’t have one myself, but am trying to find a suitable TRV solution so would be very interested in the results!
Incidentally, I asked how they get around the delay times and they account for this by sending the setpoints in advance, as I think others have done.
The CEO of JSJS who make LightwaveRF was also there and talked about a new TRV they are due to release this month. I can see its mentioned on their website and you can sign up for pre-orders. I wonder if this with the LightwaveRF integration in Vera may provide an alternative solution?
http://www.lightwaverf.com/PreOrderHeating