I don’t disagree with anything you have said here. I did argue that HACS is not wrong.
But don’t let anyone else tell you what you want to do is wrong.
I still believe that HACS is a large/strong part of HA. Just like other open source projects that I run are a strong part of my separation from the huge data grabbers.
If you can/want to run HA without HACS, then go for it. I get it!
Hi all. I’m another person who’s moved to BJReplay and it mostly just worked.
I don’t have the forecast lines in my Energy tab though. Before I troubleshoot deeply, I was wondering if that was typical from others who had moved, or whether I’ve just broken something in my install.
Thanks
Graham
If you’ve uninstalled one Solcast Integration (original one) and installed another (the BJReplay) one then you’ve removed the Forecast from the Energy Dashboard when you uninstalled the and need to re-add it once you’ve setup the “new” Integration…
Even tho we see Solcast as one Integration, HA sees them as unique (which I suppose technically they are)
GitHub - BJReplay/ha-solcast-solar: Solcast Integration for Home Assistant walks you through how to add to the Energy Dashboard if you need it
I’ve finally installed the open-meteo integration and when I set the DC efficiency factor to 0.9, the forecasts are almost on par with the solcast one. I’ve already moved all my automations which were using solcast to open-meteo and I’ll ditch solcast soon.
Since the non-commercial use of open-meteo allows up to 10k calls per day, 5000 per hour and 600 per minute, I can’t see a reason to use Solcast anymore.
The only downside is that if you have more than one string, you have to create helpers to combine the values for both arrays if you want to see it combined, but that’s just a few clicks and you’re set.
Thanks to @elRadix for letting us know about it!
open-meteo was wildly innacurate for my location sadly - which was a pity. Even with tweaking it was insisting a day with 21Kwh from Solcast would be 35Kwh!
the author of the Open-Meteo Solar Integration was looking for help with identifying the over-forecast issue and proposed a plan here
Not sure if anyone jumped on board but maybe worth sharing the info/offering to help if you wanted to help improve the accuracy of the forecast over there
I think the main reason it is off for me is that it doesn’t take cloud cover into consideration.
Yup…got the impression in the other thread that developer was willing to investigate and work with people to try and improve the forecast vs actual that’s all👍🏻
I’m having a play with open-meteo but I have four strings and two inverters. In my case I need to combine the forecast (you can see this under Developer > States), not just some sum totals…that is this data:
'2024-06-21T11:15:00+12:00': 981
'2024-06-21T11:30:00+12:00': 963
'2024-06-21T11:45:00+12:00': 943
'2024-06-21T12:00:00+12:00': 908
'2024-06-21T12:15:00+12:00': 869
and
'2024-06-21T11:15:00+12:00': 1933
'2024-06-21T11:30:00+12:00': 1937
'2024-06-21T11:45:00+12:00': 1940
'2024-06-21T12:00:00+12:00': 1926
'2024-06-21T12:15:00+12:00': 1904
As my array is oversized in peak generation I need to add the values together for each period and then either return the sum or a max value of 5000W (the inverter limit). The output should be:
'2024-06-21T11:15:00+12:00': 2914
'2024-06-21T11:30:00+12:00': 2900
'2024-06-21T11:45:00+12:00': 2883
'2024-06-21T12:00:00+12:00': 2834
'2024-06-21T12:15:00+12:00': 2773
Any thoughts on how to do this?
I think this is cerainly doable by defining a template helper which will combine the values and in case it’s over 5000, it just returns 5000. If I was doing it, I would ask chatgpt, as it’s faster than if I would to come up with it myself, I’m no template guru 
There’s already some work done around combining strings in the actual Open-Meteo Solar Forecast topic so you’re prob better asking/posting over there rather than in this Solcast topic…you’ll most like get a better answer over there and it’s kinda off topic in here at this stage too…
If that’s true it’ll never be accurate in cloudy areas like the UK. Cloud cover is hugely important to solar forecasts here.
Hi All, I have some data-inconsistencies I’m hoping you can help me with.
I have two strings setup in SolCast. One points mostly south, one points mostly west.
SolCast is properly showing this in the individual Sites. If you look at the forcast chart, the West chart peaks later in the afternoon:
My first observation is that the Solcast integration is not properly merging the two strings.
Solar noon is at 1:40pm in Atlanta today, but SolCast is forcasting ‘Peak Time Today’ at 1:00pm
First, the time for ‘Peak time’ is oddly a round number. I would think that should be closer to solar-noon.
Actually, it should be well-past solar noon on account of my west-facing panels.
I am not seeing a West facing tilt in this chart, and I do dramatically overproduce in the afternoon.:
Total forecasted energy does seem to include my full array, these numbers feel right: (currently 4:oopm)
And what I think is a separate issue, the forecasts are grossly inflated when I look at the forecast in ApexCharts:
Apex Charts are configured with:
- entity: sensor.solcast_pv_forecast_forecast_today
name: Solar Forecast (D1)
extend_to: false
color: grey
opacity: 0.3
stroke_width: 0
yaxis_id: kWh
show:
legend_value: false
in_header: false
data_generator: |
return entity.attributes.detailedForecast.map((entry) => {
return [new Date(entry.period_start), entry.pv_estimate*2];
});
Does anyone have tips for improving the accuracy of SolCast forecasts?
I was curious about the *2 multiplier in the data_generator, but removing that shows the forecast dramatically underestimated.
I would like to 1: ensure the west-facing panels are considered, it appears they are not, and 2: correct the overestimation in the Apex-Chart graph.
The *2 is because detailedForecast is kWh per 30 min period
Unfortunately it appears that this was a leaving present from Oziee. The BJReplay fork incorporated Oziee’s latest code changes and one of those was the measurement type of the PV today sensor. Changing this has zapped historical forecast history in the energy dashboard.
So far haven’t been successful in finding a way to recover this
If you have removed the original Integration (and deleted it’s entities as a by product) and installed the new one (with new entities, irrespective if same frontend name or not) then you have removed the historical data from your database…
It’s not an “issue” with the integration, it’s how Home Assistant works unfortunately 
This might be a way for you to “recover” your old forecast data if you have access to the old data via backup
(not sure if data is stored as deleted in HA DB or not so maybe it’s there either)
have a look at this thread but general gist would be to create a blank template senor and then import your historical data against it using the linked Integration…
edit: had a read through of the linked GitHub issue linked two posts back and the import Integration is mentioned as an option in there already…not harm having it mentioned here too I suppose
There’s an FAQ been added to the BJreplay code for this issue Troubleshooting FAQ #donotreply · BJReplay/ha-solcast-solar · Discussion #38 · GitHub
One person has tried reloading history from a backup and using the suggested integration, it hasn’t worked so far, but if we find a way of doing so we will update the discussion item.
By the way, the BJreplay fork is alive and well, a number of recent new improvements have been made, importantly increased caching of solcast data sets so the integration doesn’t crash if solcast gives a 429 busy error on startup, and automatic retries when 429 errors occur, @autosteve has been doing a great job
Makes sense. Although I’m totally confused as to why the Apex Charts grossly over-estimates my production while the one from Energy page grossly under-estimates my production.
Hi All,
Billy from Solcast here, of whom some of you may have reached out to via support.
Though this might not be the best place to communicate, I just wanted to let you know that we are looking to be more involved and gather information on how better to serve the community.
In the coming weeks, I’d like to learn as much as possible about how Solcast/solar data is being used and how to improve our services to hobbyist/home users.
For the initial feedback, I have created this google form about possible subscription options should we decide to go down that path: https://docs.google.com/forms/d/e/1FAIpQLSefZqRZ9z2f9pjl4Tc3zpscPVSBzhsPbErdyAwzui2xXqRz4A/viewform
If you have any other feedback, please feel free to add it at the bottom of the form.
And I would like to thank the community and especially Oziee, for all the hard work and effort to get the Solcast integration to where it is today.
Cheers.