Practical approaches to solar generation in 30 minute intervals (for financial modelling)

[zalazala]

Hi everyone! My background is in biomass and biogas and I am getting stuck with solar. I had designed a financial model for a 24/7 operating manufacturing plant in Singapore. Electricity will be generated by solar during the day and re-purchased from the grid during the night.

Electricity purchase / sales prices are determined per 30 min intervals and vary a lot (anywhere from 0.04 - 0.20 cents per interval). I have annual prices from the past 10 years, which I can use for different scenarios. The size of PV is 2.7MWp and it is oversized to generate the entire “requirement” of the plant’s 24/7 electricity demand, with excess sold to the grid. Electricity demand is quite constant during the entire time of operations.

Here comes the challenge:
What would be the most elegant solution to get an estimation of my generation in kWh (per 30 min interval?) to which I can attach the 30 min prices to estimate my income on annual basis. This would need to feed in the financial model. The national research institute has published a solar generation profile, to be used as an estimate of generation, but it comes with 0 explanation how the numbers were obtained and I find it too risky/ questionable to use.

So far, I had played with pvLib and got 30 min GHI and POA profiles (I really don’t know what to do with that). I installed PVsyst and Heliosolar but best I could get was hourly generation.

Really looking forward to get some guidance. Developers usually use annual guaranteed output of PV I understand.
Maybe I am making the entire thing quite complicated

[robbie.morrison]

I’ll take a stab. Did you catch the talk today on the atlite python library listed here and also on GitHub:

• https://forum.openmod.org/t/4rd-online-lightning-talk-mini-workshop/2574/2
• https://github.com/PyPSA/atlite

atlite converts weather timeseries to renewable generation timeseries at the resolution of your input data.

[zalazala]

Thanks Robbie,
I did! I was secretly hoping for some more tips beyond the workshop (which btw was excellent!).
I managed to get it work so all good now,

[jack_kelly]

Maybe also check out Renewables.ninja

[robbie.morrison]

merra2ools

See also merra2ools:

R-tools to process MERRA-2 data (NASA) with focus on renewable energy

Among other features, the merra2ools package contains (and I realize the OP question was 30 minutes resolution):

• functions and methods to evaluate solar photovoltaics hourly output/capacity factors

This data is also more favorably licensed than Renewables.ninja, the latter carries a Creative Commons non‑commercial (NC) restriction for data from the web‑interface.

And just noting that consulting services, including consulting by universities, normally classes as commercial (Hirth 2020:10).

• Hirth, Lion (1 January 2020). “Open data for electricity modeling: legal aspects”. Energy Strategy Reviews. 27: 100433. ISSN 2211-467X. doi:10.1016/j.esr.2019.100433. Open access.