Why is the setpoint way over the set temperature?

[Woody]

Good explanation. Much clearer than the official one I read in the Residio mail after asking for the beta. Thanks!

I'll set load scaling to 'Normal', give it a few days and see what happens. BTW, in my system there are 6 HR92's and 7 HR80's. Does that influence load scaling at all?

[DBMandrake]

Good explanation. Much clearer than the official one I read in the Residio mail after asking for the beta. Thanks!

I'll set load scaling to 'Normal', give it a few days and see what happens. BTW, in my system there are 6 HR92's and 7 HR80's. Does that influence load scaling at all?

I don't think the number of radiator valves make a difference to the learning time.

The individual radiator controllers all have some intelligence built into them (a self tuning PID controller) and it learns over time the response characteristics of the room - how much over the air heat demand and water flow causes the room temperature to rise at a certain rate, how early before the target temperature it has to start backing off to avoid overshoot etc...

Load scaling throws it a curve ball where the heat demand request from the individual radiator controllers is intelligently "scaled down" from what they actually asked for. So the adaption process is largely about the radiator controllers learning that they need to send a higher heat demand than they used to to get the same result. (Also the load scaling itself performed in the controller has its own separate adaption process)

So typically the initial reaction to enabling load scaling for the first time would be for some of your rooms to undershoot their targets by maybe a degree or so and then gradually adjust to get it right. Similar to what you might see if you got a sudden cold snap where the targets might be undershot for a day or so.

[Woody]

After only two days load scaling behaves exemplary. No more 100% boiler requests just to heat up a bathroom. Out of my 11 zones only one still undershoots. This is always a difficult room (office) that has a lot of loose stuff in it (projects, measuring equipment, PC's, tools, materials, and what not) that all have to heat up in the morning. So I'll forgive Evohome that one. I am nicely surprised :-)

[DBMandrake]

After only two days load scaling behaves exemplary. No more 100% boiler requests just to heat up a bathroom. Out of my 11 zones only one still undershoots. This is always a difficult room (office) that has a lot of loose stuff in it (projects, measuring equipment, PC's, tools, materials, and what not) that all have to heat up in the morning. So I'll forgive Evohome that one. I am nicely surprised :-)

Glad to hear its working well for you as it does for me.

On the room that undershoots do you mean it just doesn't hit its target temperature in time but does reach it later ? (In which case increasing the maximum optimal start lead time might help - I have mine set to 4 hours to allow for a couple of rooms which take a long time to heat up)

Or do you mean it reaches a temperature that is a bit below the target for the room on time, and maintains that lower temperature steadily for a long time? (This is what I would call an undershoot) In that case, it may still eventually adapt, although it sounds like the heat output in that room might be a bit marginal relative to the work to be done to heat up the room.

What is the heat demand reported for this undershooting room while it has stabilised at a below set point temperature ?

[Woody]

It reaches the target temp later and then more or less keeps it:

Undershoot.jpg

The optimal start time is set to 3 hours. By the graph it looks like the office heats up nicely for 2 hours after which it takes another hour for the last .5C to end up at the target temp. It is set to reach the target temp at 08:00. I think it might need another couple of days to get it right.

I haven't looked at the heat demand yet; will take a look next Monday.

[DBMandrake]

Ok. That looks the same as what I see when I get a sudden cold snap - it undershoots the set point by maybe 0.5C, remains steady at that value for quite a while and then gradually creeps up to the desired set point over a couple of hours. After a day or so of the new cold weather conditions it has fully adapted and doesn't undershoot.

My understanding is that this is caused by the tuning of the PID controller in the HR92 having adapted to the previous conditions (the degree of heat loss through the walls due to outside temperature thus the amount of heat and valve position required to reach the set point) and when the operating conditions change dramatically (and enabling load scaling would count as a dramatic change in operating conditions from the point of view of the HR92 which is unaware of it) so it will take time for the Integral and Differential factors to adjust themselves to match.

What it also shows is what a good job it was previously doing of predicting how early to start closing the valve to reach the set point quickly and without any overshoot. (As the undershot temperature is so flat) When it gradually creeps up over another hour or two this is the integral action of the controller gradually increasing the valve position and heat demand due to a sustained undershoot.

I suspect another couple of days and it should finish adapting.

[Woody]

This weekend the temperature in our living zones (three zones that are connected and share the same temperature setting) was 0.5C too low, and noticeably so. Increasing the setpoint increased the temp, but still with keeping a -0.5C difference to what was set.

The office is still heating up atm; the setpoint is right, the room temperature runs behind. It needed to heat up 4.5C and be done at 08:00, but at 09:15 it still needs to increase 1C. The office asks for 17% heat but the wireless relay box translates this to 5%. I guess this is load scaling learning.

I'll give it a couple of more days.

One thing I would like to do is eventually replace the on/off relay with the R8810 but I'll postpone that until my current setup is settled.

[bruce_miranda]

Are you absolutely sure you don't have one of those uncontrolled boilers that will try and ramp up to 90C when used with the R8810? Remember that load scaling doesn't work with the R8810.

[DBMandrake]

This weekend the temperature in our living zones (three zones that are connected and share the same temperature setting) was 0.5C too low, and noticeably so. Increasing the setpoint increased the temp, but still with keeping a -0.5C difference to what was set.

The office is still heating up atm; the setpoint is right, the room temperature runs behind. It needed to heat up 4.5C and be done at 08:00, but at 09:15 it still needs to increase 1C. The office asks for 17% heat but the wireless relay box translates this to 5%. I guess this is load scaling learning.

I'll give it a couple of more days.

One thing I would like to do is eventually replace the on/off relay with the R8810 but I'll postpone that until my current setup is settled.

Try switching Advanced Load Scaling to "Partial" mode. This reduces how much the heat demand is scaled down. I found the "Normal" mode scaled the load down too much on my system and I saw similar results to you until I did so.

If you check the System Info page where the heat demand is shown before and after changing it you should see the heat demand to the boiler increase significantly.

BTW Load scaling doesn't operate with an OpenTherm bridge, so having it "settle" won't make any difference to fitting the OpenTherm Bridge - it will be learning all over again as OpenTherm control is very different to TPI.

[Woody]

Try switching Advanced Load Scaling to "Partial" mode. This reduces how much the heat demand is scaled down. I found the "Normal" mode scaled the load down too much on my system and I saw similar results to you until I did so.

If you check the System Info page where the heat demand is shown before and after changing it you should see the heat demand to the boiler increase significantly.

BTW Load scaling doesn't operate with an OpenTherm bridge, so having it "settle" won't make any difference to fitting the OpenTherm Bridge - it will be learning all over again as OpenTherm control is very different to TPI.

Set it to partial and yes, the boiler heat demand increased immediately. Looks like it now translates the heat request of the office (84%) 1:1 to the boiler again.

About OT and load scaling, does that mean that OT is supposed to do the load scaling always? My experience is different as, when I had the 8810 controlling the boiler, it overheated every now and the due to a large heat request for only a small room. Haven't seen that since installing the BDR and load scaling. Kinda silly if I could only prevent that by installing an on/off relay IMO.