Home automation - how intelligent is it?

20 comments started 2024-03-29 last 2024-03-29
Home Automation
#1 Midlands_Solar

I dabbled with Home Assistant and wondering if it’s really intelligent to take away some of the manual settings I’m having to make around battery charging off peak

Presently I’m checking the forecast and if there’s good sunlight forecasted early the following morning, i’ll scale back the amount of charge I give the battery, and let it take the load for the immersion heater to then be topped up in the morning and early in the day. to then take the house load in the evening

My question is can Home Assistant do this for me? Or is there always going to be an element of manual intervention needed?

Today for example I charged to 85% and the battery took the water heating load from 04:00 -05:15 leaving me about 15% charge left early morning after the charge schedule finished prior to the water being heated

The weather today is however changeable, and so I’ve my fingered crossed the battery will reach a decent level of charge to take the house load from about 18:00, but it’s somewhat guesswork a this time

V
#2 Vestas

Midlands_Solar GivTCP has solar forecasting built in, just need to setup a Solcast account and enter the API.

D
#3 DD

Midlands_Solar Today for example I charged to 85% and the battery took the water heating load from 04:00 -05:15 leaving me about 15% charge left early morning after the charge schedule finished prior to the water being heated

I think that's the first thing to address: you should be heating the water directly from the grid during the cheap rate, not from your battery. You should configure the battery charging timer to use the entire cheap rate period, to prevent any discharge - use the charge power and limit to get it to stop actually charging at whatever level you want, but don't allow it to discharge.

#4 Midlands_Solar

My logic here was charging the battery from 00:00 - 04:00, our water heating comes on at 04:00 - 05:30, the battery takes the load from 04:00 - 05:30 and then is topped up in the morning and in the day, so saving 1.5hrs from the grid at the off peak rate

D
#5 DD

Midlands_Solar But your cheap rate runs until 7am ? Just heat the water directly from the mains. No need to discharge the battery to get off-peak prices.

And if you are then topping up the battery during the day, at peak rate, then you are effectively forcing the water to be heated at peak rate, even though it happens during the off-peak period. (What you are doing is equivalent to charging the battery to 15%, then heating the water at cheap rate, though with battery round-trip loasses, and then waiting until peak rate to charge up the battery.

If you just charge the battery to whatever you need and also heat the water overnight from the grid, then you don't need to use any daytime rate to end up with hot water and a charged battery.

#6 Midlands_Solar

Sorry I meant I’m using solar in the day to top up the battery, the aim of the game is to reduce or negate as much a possible any pull from the grid at any time 🙋🏼‍♂️😉

Yesterdays tweak and using the battery to heat water has seen my overnight pull (granted at the cheap rate) to literally drop by 2/3rds this morning


D
#7 DD

Midlands_Solar I dabbled with Home Assistant and wondering if it’s really intelligent to take away some of the manual settings

Home automation isn't intelligent as such: it just automates the rules you set. So if there's stuff you are doing every day, it can do that for you. You don't need to use Home Assistant to perform automation - I do it with simple scripts via the GE API. Just simple things like choosing what charging power is needed to charge the battery slowly overnight, and and what rate to export excess power in the evening to get the battery down to around 10% by 2230.

Solar forecastiing is harder to encode rules for: you are probably making judgement calls you don't realise. (Things like whether you're likely to need more or less hot water next day, all those sorts of things.) The tools in Home Assistant will have parameters that have been found to work well most of the time for most of the people, but that's not intelligence from HA itself.

Predbat is one of the plugins that is available for HA. It may look intelligent, but it's simply optimising profit and loss using simple rules (and a lot of parameters) to micromanage the inverter.

D
#8 DD

Midlands_Solar Yesterdays tweak and using the battery to heat water has seen my overnight pull (granted at the cheap rate) to literally drop by 2/3rds this morning

Sure... but it would use probably 10% less if you just heated the water directly from the grid, rather than charging the battery then discharging it into the water. (Round-trip conversion losses - converting AC to DC to store it in the battery, then DC back to AC to send to the water heater.) Plus its extra unnecessary cycling on the battery.

Plus it's an additial 3kWh that's no longer available in the battery for the rest of the day, though of course since I don't think you are being paid for export, you'll be wanting to ensure there's space in the battery for solar.

(Looks like you're charging the battery back up again from the grid, after water heating is done ?)

#9 Midlands_Solar

DD yes that’s correct it’s charging from around 07:30 on a good solar day, current stroll no SEG so looking to make further efficiency gains until I’m able to make £ from exporting

I think despite the round trip efficiency losses I’m still better off doing what I’m doing for the moment

That’s interesting on HA and Pred Batt, I was hoping it would make more decisions based on the SolCast APO plugin, but looks like there are too many variables, and it would take a lot of effort to get it close to the decisions I’m making currently, with good results

D
#10 DD

Midlands_Solar no, it looks like it's charging from the grid around 6am, at just over 1kW. But if you're happy with your setup, that's fine.

#11 Midlands_Solar

DD yes that was me making a change early this am to top up the battery off peak, would have probably been fine as the battery at this rate if charge will be charged by around lunchtime

V
#12 Vestas

Midlands_Solar I think despite the round trip efficiency losses I’m still better off doing what I’m doing for the moment

You're not. The round trip losses will effectively push the cost of heating your water up by 15-20%.

#13 PianSom

Midlands_Solar I dabbled with Home Assistant and wondering if it’s really intelligent to take away some of the manual settings I’m having to make around battery charging off peak

DD Predbat is one of the plugins that is available for HA. It may look intelligent, but it's simply optimising profit and loss using simple rules (and a lot of parameters) to micromanage the inverter.

I'm not sure what intelligence is if it's not taking in information and using rules to produce an output!

Just to give a bit more colour, Predbat takes in as input

  • your system setup (inverter(s) size, battery(ies) size, etc), often grabbed using the GivTCP integration
  • your import and export tariffs (for many using the excellent BottleCapDave Octopus integration)
  • a solar forecast, typically from Solcast
  • your historical usage over the recent past, hour by hour, from HA
  • details of your electric car, if any
  • any other over-riding factors (eg Octopus Power-ups, Octopus Saving Sessions (it will auto-opt-in to these), holidays)
  • a whole raft of parameters, which are used to change the weightings of the above in its determination of its outputs

The outputs are

  • predicted load (based on historic usage patterns)
  • predicted solar generation (based on Solcast, with the ability to be more conservative)
  • a plan of when to charge/discharge in order to maximise income/minimise cost over the next [24] hours
  • the ability to take control of your system and implement that plan (for Givenergy and a few other systems)

It's a complicated beast, and is in active development. Tuning the parameters to get the result that works best for an individual case takes time and effort. But it is pretty impressive.

#14 Midlands_Solar

Vestas my kWh usage is down by 10kWh overnight, so despite the loss I’m still better off

#15 Midlands_Solar

PianSom thank you

I’ll likely look at it a HA in the next couple of month, I built one on an earlier edition Raspberry PI but it wasn’t playing nicely, so decided to shelve it and will buy a v5

V
#16 Vestas

Midlands_Solar I've had it running OK on a RPi3 so I'd concentrate on getting GivTCP/HA/Solcast working before you buy more hardware.

D
#17 DD

PianSom I'm not sure what intelligence is if it's not taking in information and using rules to produce an output!

That's all any traditional programming is doing. Nothing intelligent about it.

T
#18 TX200

Midlands_Solar

To avoid the round trip efficiency loss, set the timed charge to end at the same time as your off peak ends

BUT set the charge to percentage on the timed charge to something other than 100% - say 75% or whatever suits,

That will leave space in the battery for early solar charging and ensure water is heated from the grid not the battery, thus saving in the long run.

#19 Midlands_Solar

TX200 yes that’s what I’ve done, I set the percentage charge to between 70-85% based on weather conditions the following day, this gives enough overhead to charge the battery to full in the day from the solar array, for discharge for the house in the evening

#20 PianSom

DD
It’s also all that brains do!