Hi,
Does GivEnergy have a dynamic timed charge function that pulls weather forecast for the next day and sets the minimum in the battery to an appropriate level?
Like a weather compensation function for amount of charge
Thanks
s_r_thomas it doesn't but that would definitely be a great alternative to having to check the forecast each day and try to estimate the best overnight timed charge percentage. I did read somewhere they were looking into it though, so fingers crossed!
I do think some people in this forum have managed to test a similar functionality using APIs - you may want to try that if that's your cup of tea.
Huntzie I saw the same from one of the GE developers. However, I'd be surprised to see it come to anything because of the complexities of individual setups. Imagine the discontent if users start losing money because the GE software isn't charging batteries when it "should" be!
I've developed my own code to reduce overnight charge based on the midnight Solcast forecast.
I have the forecast solar data in home assistant and use this to guide how much I charge overnight. I would say it's almost spot on about 80% of the time but can occasionally be really wrong. I also look at the weather forecast to see if the forecast solar still looks correct, but I also try to take account of lots of other factors which an algorithm wouldn't be able to such as am I working at home or in the office, is it a weekend or a weekday, am I going to be cooking at home or eating out etc. (bit sad really!). I also stay very much on the cautious side because if I charge overnight and it turns out the next day that I didn't need to then each overnight kWh cost me 7.5p but if I undercharge and the battery runs out then each kWh at peak rate is 31p.
judgepd I am so simple by comparison with so many here. I allow my battery to charge to 100% at octopus Go off peak which of late has been 4 to 5 units each at 7.5p. If it turns out that the solar production is good the next day I am exporting at 4.1p so chipping away at the overnight cost. The notional target is to cover the charging cost and then the 25p standing charge. I would much rather pay 7.5p for a unit that I am storing than 31p for a unit that I actually need. If we get a good summer and I feel that the solar can be guaranteed I can see me dropping down to 50% which will reduce my nightly cost by 30p or so. But I am still in the fortunate position that every day my GE system is significantly reducing my home running costs in comparission with my grid reliant years and that is good enough for me.
It is not just the accuracy of the solar forecast you need to be wary of. If the battery SOC figure has drifted significantly above reality then the overnight "charge to" will be lower than expected and undercharging is a real possibility.
I'm lucky that I get fit payments so my payment is the same whether the solar is exported or stored in the battery. I could make it much simpler like you say and just set it to 100% every night rather than trying to save a few more units at 7.5p...but where's the fun in that. 😂
Good to see how people are using the system. I'm still at design stage and installer coming to quote on Friday.
I'm keen on the api side stevelewis maybe you could share? Do you have a github?
Most of my data I keep local and store in influxdb with a custom grafana dashboard. Helps me bring all the data together instead of inside half a dozen different apps!
stevelewis Imagine the discontent if users start losing money because the GE software isn't charging batteries when it "should" be!
Tesla Powerwall owners say Hi 😃
browellm If Mr Musk can predict the weather, when the washing machine gets put on and also if/when the wood burner will be lit the next day, he's a better man than I am 😆
Github location: https://github.com/salewis38/palm/tree/main
The code is stand-alone and can run on most things, including a tablet, RPi or even a router running OpenWRT. It does a few things within a 1-minute loop structure... the GivEnergy and Solcast classes set the SoC, the parameters may need adjustment for your system - depends on power consumption profile, PV setup, battery size, etc.
The code sets the overnight charge based on time of year, day of week and generation forecast for the following morning. It's "roughly right" rather than "precisely wrong". Improvement suggestions welcome :-)
stevelewis fab I'll take a look!
heres another you might like to look at https://github.com/lockedj/solarbatteryoptimize it has been working well for the last and a half.
locked great thanks very much... That readme fits my setup exactly, even down to the synology running docker.
s_r_thomas ping if you have any questions. With the old APIs being deprecated I will be updating it to use the new V1 APIs over the next month