Hard fail if max forecast above 100 GW

[peterdudfield]

It would be good to add a hard fail in the forecast is above a crazy amount like 10% above Nationla capacity

The check could happen here

if da_abs_national.max() > 1.1* national_capacity:
   raise Exepection('The maximum of the forecast is above 10% of the national capacity ({national_capacity})'

[dantravers]

Maybe 30GW is enough, and would catch more possible error cases?

[peterdudfield]

Perhaps it could be > 50% above national capacity, so like 17*1.5 GW. This means it wont fail in the future when capacity grows

[dantravers]

Yes, good idea if we can index to national capacity. Might only need a 10% buffer, as we never see forecasts close to MWp capacity.

[Sukh-P]

Great check to add in! Would it still be good to have a hard max limit constant? I guess the current check assumes that the national capacity (which I assume we get from PVLive also?) is also correct, which might not always be the case

[braddf]

This is a good point @Sukh-P – either as a hard max constant or as a warning constant, maybe both make sense if we're worried about hitting a manually configured limit?
Capacity doesn't change that quickly, so could be notified when we get a number 10% below the hard limit or something.

[peterdudfield]

Ah, so we could put in an extra limit that stops a 100 GW as well?
We can put a warning too at 90GW, but that'll be a while a way

[braddf]

Still seems a little high to me – if we set the warning/limits lower and need to bump this once a year when it reaches the warning capacity, I think that's fine, so more like 25GW warning, 30GW limit? As 40-50GW would take us to 2030 ish at current estimates I think!
(There are also other things like the UI that would like to know when the capacity does meaningfully increase which would be helpful to be aware of)

[dantravers]

Great check to add in! Would it still be good to have a hard max limit constant? I guess the current check assumes that the national capacity (which I assume we get from PVLive also?) is also correct, which might not always be the case

Yes, this occurred to me also - if the capacity figures were missing for some reason, we might trigger false positives and incorrectly fall back to a lower quality forecast.

[peterdudfield]

Ok, so perhaps

• warning if forecast is above 30 GW
• hard fail if the forecast is above 100 GW.

For missing capacity, i.e capacity is nan, then we currently fail. I think I would rather keep it like this for the moment. If it starts happening then we put in a back up capacity amount. This would raises risk of not knowing we are on a backup capacity, and therefore producing lower forecast, vs robustness of the service.