v2021.12.5-beta1 - Ke compensation

[bigbread59]

I would like to give a small contribution in the "Ke compensation" introduced in v2021.12.5-beta1.

I found an inconsistency between the one written in
"Beta version What's new"

It is written:

An external component E is added to the PID output:
...
E = Ke * (target_temp - outdoor_temp)
This compensation will act like the integral of the PID

It seems to me that the contribution of "E" is similar to the proportional action of the contribution "P" rather than the integral "I".

Unfortunately the effect of the temperature outside the building does not propagate instantaneously inside the building but with a delay due to the thermal resistance of the walls and the thermal capacity of the internal environment (such as an RC electrical circuit).

Therefore I would like the opinion of expert peaple and open the discussion whether it is better to use:

E = Ke * (target_temp - outdoor_temp)
or:
E = last_E + (Ke * (target_temp - outdoor_temp) * dt)
similar to integral action which will be delayed as needed.

The problem in this case is that the "E" term will grow indefinitely until the outside temperature is lower than the inside one, so a way should be found to avoid this.

[ScratMan]

Hello, when said it works as integral, it means by adding a permanent offset to the PID output.
The difference with integral is that I is automatically tuned by the system, so the offset will depend on the current temperature ; while E does not depend on the status of the regulation: colder outside = faster losses = need more power.

The E part is directly inspired from boilers linear regulation of water output temperature based on outside temperature.

[bigbread59]

@ScratMan
Hi,
Thanks for the clarification!
Now your implementation becomes clear and I think it can be very useful.

I'll try it.

[ScratMan]

To test it, you can look at your integral part once settled in the PID, for example mine was moving between 20 and 30 %, and then look at the target_temp - outdoor_temp difference (for me today it's around 22°C - 3.5°C = 18.5°C), and then compute Ke so that the integral can flow between 5 to 15%, that means the 18.5°C difference should lead to E = 15% on my example, it gives a Ke = 15 / 18.5 = 0.81

[bigbread59]

@ScratMan
Thank you,
I just restarted the PID clearing the integrating part of the PID.
I'm waiting for PID stabilization, after that i'll do the Ke compensation.

[ScratMan]

Working fine in my bathroom...

[bigbread59]

I will keep Zwave products in mind for "safety critical" functions/applications.

[ScratMan]

Yeah, ZigBee is good for "standard" home appliances, like shutters, lights, or sockets.
For openings detection, alarms or heating I prefer the flexibility, security and robustness of Zwave.

[bigbread59]

I agree

[bigbread59]

Which Z-Wave gateway do you use?
Do you have some suggestion?

[ScratMan]

I'm using an Aeotec stick on a Raspberry pi.

[nagyrobi]

Question:
The doc says The Ke gain depends on the insulation of the building, on recent buildings with good insulation, a gain of 0.6 is recommended.
What's a good value to start with poor uninsulated older buildings? What are min/max values? Is it a linear value?
Like 0.01 poor insulated and 0.99 best insulated? Or reverse?

[nagyrobi]

Anybody a hint please?

[ScratMan]

Sorry for late reply, to determine the value, please check the pid_i value when settled, and use method described in this post above to compute your base Ke.
#50 (reply in thread)