Investigate modelling of solar generation given user input

[cormac85]

Can we use public geo data? Or do we need to rely on user intra-day input?

Acceptance Criteria

• Given sensible user input (e.g., a set of example days of intra-day electricity generation & consumption, or a latitude & longitude + intraday consumption example data)...
• ... the model simulates an entire year of intra-day generation & consumption data

Notes

• The example input data could give a sensible way to internalise cloud cover & irradiance data without going mad with solar irradiance & stochastic modelling etc.
• On the other hand, the variance in days might be much higher in some regions than others so assumptions of variance made internally to the model might not generalise well.
• To do some Fisher Price irradiance modelling based on latitude & longitude will be quite a task, a few thoughts:
  1. Calculate the solar curve at the lat/long.
  2. Assume a cloud cover % (user input?) and apply some stochastic modelling to take "bites" out of the generation (random binary "is cloudy", random % drop)
  3. Use your own solar data to validate model (using testthat?)

[cormac85]

Some interesting links:

1. How to calculate solar irradiance: https://www.pveducation.org/pvcdrom/properties-of-sunlight/calculation-of-solar-insolation
2. Another model for solar irradiance: https://www.sciencedirect.com/science/article/pii/S221260901400051X