Datasets:
PV system capacity and generation data
Hi, many thanks to the ReadMe explanation! But I am bit confused when exploring the data on PV system capacity vs generation power. Here quotes your ReadMe:
- Metadata.csv: 'kwp: The capacity of the PV system'
- Min.barquet: 'generation_wh: the generated power (in kW) at the given timestamp for the given system'
In browsing data, generation_wh can be 90 times or more than kwp for the same solar panel(ss_id). Any relationships missing in the unit transfer or else?
Hi there
Sorry about the confusion.
- This is the capacity of the system. bare in mind this can be a bit un reliable.
- Yea, sorry this is confusing.
generation_whis a unit of energy. So its the the amount of energy produced in a 5 min period. If you divide it by 12, you get the power (in watts). Then divide by 1000 to get to kw. Hopefully then most of the data is below the capacity_kwp of the system, however I'm sure there are still some anomalies.
Does this help?
@peterdudfield
Perfect! Thank you for helping.
Now I use 5 minute.barquet data and scale generation_wh with 12*0.001. The result is <= to capacity with very few anomalies.
Hope my model is happy now.
Excellent, tjhats great news. It would be great to hear what you are training model for?
Hello @peterdudfield and the Open Climate Fix team,
I just wanted to take a moment to thank you for providing such great data. You've been instrumental in supporting my thesis on solar energy predictions and I couldn't have asked for a better source.
One quick thing I wanted to ask regarding the netcdf file: I saw Peter`s reference to converting generation_wh to kW (divide by 12 and then divide by 1000). Super helpful tip, thanks! When I calculate the numbers this way (with subsequent normalization by the installed power which is also given in kW), the maximum values seem to be a bit too low and often don't reach more than 0.1 compared to the installed power. However, if I skip the divide by 12 step (and in this case assume watts instead of watt-hour for generation_wh), the numbers look more like I would expect, although sometimes they exceed 1, which again confused me a bit.
So I thought I'd better check with you again. Many thanks in advance and best regards,
Felix
Hi,
any update on this?
Thanks!
Hi @felitsch - sorry for the slow response.
I wonder if there is a differences between the parquet and the netcdf file.
What units does it say in the netcdf file?
I wouldnt' worry if there are some outliers over 1, this can either be noise in the data, or the capacity was inputted incorrectly. For example the tilt and orientation are perhaps not so accurate as they are measure by the installation engineers