Notifications

Clear all

Adventures in DIY Smart Thermostating

scrchngwsl · 2023-04-26T09:21:46Z

Over the past 5 months or so, I've been experimenting with my own "Smart Thermostat" algorithm. There are two parts to this: (1) the algorithm itself, and (2) the implementation using Home Assistant and a bunch of python scripts/cron jobs. However, I'll just stick with the algorithm for now, as the implementation part is boring to me! The goals of this are as follows: Exploit cheap rate electric tariffs (my Octopus Go rate is 7.5p/kWh between 12:30AM and 04:30AM) Run the heat pump harder during the day when it is warmer outside (and therefore more energy efficient for the HP) Have a more consistent flow temperature (which I think makes for less cycling and a more consistent "feel" in the room from the radiators) Have a more consistent room temperature (objectively, as measured by a thermometer) Have fun. (Frankly this is the biggest motivator for me. I just like making stuff work the way I want it to.) I start with a Weather Compensation curve that I've been tweaking for a while. Nothing special here, just something that gives me a solid baseline. However, from this base, I'll be adding and subtracting "offsets" or otherwise modifying the standard WC calculation to achieve the above goals. The first offset is a simple one: add X degC to my flow temperature during the cheap overnight rate period. Currently I use an offset of +5C, but I change this frequently. That raises my house's temperature overnight, which I'd imagine for most people would be kind of annoying, but for me it works for two reasons. First is that we have a baby and we often are awake between 12:30AM and 04:30AM. The second is for efficiency of the heat pump – more on that later. The second modification of the standard WC calculation addresses points 2 and 3. During the course of a 24 hour period, the temperature rises and falls gradually, peaking in, say, mid-afternoon, and troughing some time at night (and usually staying low til morning). This does two things. First, the delta-T between required indoor and actual outdoor temperature changes cyclically throughout the day, resulting in the heat loss rate (and therefore heating requirement) of the house to change throughout the day. Second, the amount of energy required to pump the same amount of heat into the house changes cyclically throughout the day. And to make matters worse, these to cycles are in phase; that is, the heat loss from the house is highest when the amount of energy required to pump heat into the house is also at its highest. The two temperature gradients compound to make an even steeper efficiency gradient for the heat pump. For example, if it's 3pm and the outdoor temperature is 12C, but it's forecast to drop to –2C in the night (a massive swing that happened quite a lot in Feb/March!!), a normal WC curve will run at a relatively low flow temperature (say, 30C) as the outdoor temperature is low. But this leaves the house vulnerable to the cold at night, and when it's –2C outside, the WC curve will tell the heat pump to run at say 45C, struggling to pump heat against two steep temperature gradients (higher flow temp AND lower outdoor temp). Wouldn't it be better to pump more heat when it's 12C, store it in the house for the night, then let the heat pump relax a bit when it's –2C? Normally, the Weather Compensation curve has "Outdoor Temperature" as its x-axis, and "Flow Temperature" as its y-axis. My second modification alters the x-axis, so that instead of looking up the outdoor temperature right now, it instead takes the average outdoor temperature forecast for the next 24 hours and looks that up on the WC curve. In the above example, let's say the average temperature for the next 24 hours was 5C. At 3pm, when it's 12C outside, instead of looking up 12C against the WC curve and instructing 30C for the flow temperature, we instead look up 5C, giving a flow temperature of, say, 37.5C. This results in more heat being pumped into the house than is lost by the house, causing the average temperature of the house to rise a bit. Then, in the night, when it's –2C outside, instead of looking up –2C on the WC curve and getting a flow temp of 45C, we again look up 5C and get a flow temp of 37.5C. This results in less heat being pumped into the house than is lost by the house, causing the average temperature of the house to fall a bit. Overall, over the full 24 hour period, if we make the false but simplifying assumption that the heat loss of the house is linear with the delta-T between indoor and outdoor temperature, there is no net change in the total amount of heat being lost by the house, and therefore no net additional heating requirement for the house. In reality, as the relationship is not linear, there will be some additional heating requirement. However, as the heat pump is operating at a flow temperature of 37.5C for the full 24 hour period, the heat pump uses less energy in total than if it were operating at 30C in the day and 45C at night. I admit I haven't done the ...

Page 2 / 2 Prev

Air Source Heat Pumps (ASHPs)

Last Post by Derek M 2 years ago

18 Posts

6 Users

6 Reactions

2,350 Views

RSS

Derek M

(@derek-m)

Illustrious Member Member

Joined: 5 years ago

Posts: 4427

26/04/2023 6:04 pm

@iancalderbank

You could try checking your version of firmware against others with working systems.

ReplyQuote

cathodeRay

(@cathoderay)

Illustrious Member Moderator

Joined: 4 years ago

Posts: 2199

26/04/2023 7:48 pm

Posted by: @scrchngwsl

↑

Please do post about it on the forum if you ever do this as I'd be very interested to know what your experience is.

Rest assured I will, but don't hold your breath, it will probably be next autumn when the days start getting cooler again before I do this in earnest, though I have to say today was blooming cold for late April.

Posted by: @scrchngwsl

↑

That way you're not limited to a straight line or a x^(1/a) curve or whatever.

I agree, see also my comments on the other thread (<=link) about the potential of using a fixed LWT and then setting that fixed LWT from a DIY programmed weather curve.

Midea 14kW (for now...) ASHP heating both building and DHW

👍

ReplyQuote

Derek M

(@derek-m)

Illustrious Member Member

Joined: 5 years ago

Posts: 4427

26/04/2023 8:45 pm

@iancalderbank

If you look at the manual for the Samsung Modbus card, there is a register which needs to be set to enable write operations to be performed. I would suggest that you check this if you have not already done so.

ReplyQuote

scrchngwsl

(@scrchngwsl)

Reputable Member Member

Joined: 3 years ago

Posts: 91

Topic starter 26/04/2023 9:42 pm

@iancalderbank @derek-m - Wow, I can't believe I never thought of using my PV array data to compensate for solar gain! Thanks, will investigate this. Presumably the amount of energy absorbed by the house is proportional to the amount of energy converted by the panels, with some adjustment for the angle of the sun throughout the day. Setting up a fake array in HA for each orientation of my house might provide the right correction to the reads from my actual SE-facing array.

@iancalderbank - I'm in total agreement, I would much prefer a local-only solution, as I'm dependent on the good will of Mitsubishi to keep their servers running in exactly the same state for as long as I have the heat pump.

ASHP: Mitsubishi Ecodan 8.5kW
PV: 5.2kWp
Battery: 8.2kWh

ReplyQuote

iancalderbank

(@iancalderbank)

Noble Member Contributor

Joined: 3 years ago

Posts: 643

27/04/2023 8:59 am

Posted by: @derek-m

↑

@iancalderbank

If you look at the manual for the Samsung Modbus card, there is a register which needs to be set to enable write operations to be performed. I would suggest that you check this if you have not already done so.

is there ? which one? I've read that manual dozens of times, nothing jumped out at me like that. if you mean 64 "restrictions on the remote control ", yes, I already checked that (you already asked me to , a month or so back....)

please bear in mind (as per the other thread) I have already documented that I have run code (not written by me, I hasten to add) that successfully writes to a bunch of registers that enable some "more registers" (that aren't enabled as standard for whatever reason). I'm successfully reading those "more registers" thus proving that the "write" that enables them has worked.

ref firmware - that is a possible line of attack, but if you google "samsung heat pump firmware" there is nothing that jumps out, it doesn't look like anyone else has been down that road. just a bunch american discussion where "heat pump" means air conditioner. I am not even sure how to find out or tell what firmware its running or is the latest. certainly not something I'd attempt to go into myself. when I get my usb-modbus adapter I'll be trying it from a laptop, hopefully will get a better process of stepping through what does and doesn't work.

My octopus signup link https://share.octopus.energy/ebony-deer-230
210m2 house, Samsung 16kw Gen6 ASHP Self installed: Single circulation loop , PWM modulating pump.
My public ASHP stats: https://heatpumpmonitor.org/system/view?id=45
11.9kWp of PV
41kWh of Battery storage (3x Powerwall 2)
2x BEVs

ReplyQuote

Derek M

(@derek-m)

Illustrious Member Member

Joined: 5 years ago

Posts: 4427

27/04/2023 9:51 am

Posted by: @iancalderbank

↑

Posted by: @derek-m

↑

@iancalderbank

If you look at the manual for the Samsung Modbus card, there is a register which needs to be set to enable write operations to be performed. I would suggest that you check this if you have not already done so.

is there ? which one? I've read that manual dozens of times, nothing jumped out at me like that. if you mean 64 "restrictions on the remote control ", yes, I already checked that (you already asked me to , a month or so back....)

please bear in mind (as per the other thread) I have already documented that I have run code (not written by me, I hasten to add) that successfully writes to a bunch of registers that enable some "more registers" (that aren't enabled as standard for whatever reason). I'm successfully reading those "more registers" thus proving that the "write" that enables them has worked.

ref firmware - that is a possible line of attack, but if you google "samsung heat pump firmware" there is nothing that jumps out, it doesn't look like anyone else has been down that road. just a bunch american discussion where "heat pump" means air conditioner. I am not even sure how to find out or tell what firmware its running or is the latest. certainly not something I'd attempt to go into myself. when I get my usb-modbus adapter I'll be trying it from a laptop, hopefully will get a better process of stepping through what does and doesn't work.

Please forgive me for having forgotten one of your previous posts from one month ago, and for me having not read back through all your previous post to ensure that I had not forgotten anything.

I'm afraid that being old and decrepit I easily get confused, and mix up the messages from the numerous people, whose problems I am trying to help resolve.

I appreciate I will only get a three out of ten, must try harder. 🙄

👍

ReplyQuote