Do-a-thon: Comparison of variable renewable energy generation models

Proposal for:
Do-a-thon

Session Title
Comparison of Variable Renewable Energy System (VRES) input generation models

Session Description
Variable Renewable Energy Systems (VRES) such as onshore wind, offshore wind, rooftop PV, and open-field PV will play an important role in our future energy systems, and as such, they are central in our current energy system design (ESD) activities. However, in order to effectively incorporate VRES technologies into these ESD models, it is necessary to create realistic VRES generation time-series profiles to then serve as ESD inputs. Since VRES performance is derived from the design and configuration of these systems (i.e. the capacity, rotor diameter, and hub-height of a wind turbine, or the tilt, azimuth, and chemistry of a PV system) in addition to transient weather phenomena, determining these input VRES generation profiles is far from a trivial exercise.

Fortunately, there are many openly available weather datasets and VRES simulation tools available to help us generate generation profiles for our ESD activities. Many of which are being developed by openmod’ers! I think it’s time we spoke about the pros and cons of each of these tools, compared them against each other, and, if enough common ground can be found, think about the possibility of merging similar functionalities. We can also spend some time discussing “best practices” when generating VRES profiles for use in ESD analyses.

Examples of Models I have in mind are:

  • Atlite for wind, PV, and hydro generation
  • gsee for PV
  • vwf for wind
  • windpowerlib for Wind
  • PVLib for PV
  • SAM for…everything
  • Also the simulation code for wind and PV which I wrote for my PhD thesis will be put on GitHub in the next week or so

If anyone else has other models in mind, then I think we should also consider them as well.

Anyhow, for this do-a-thon short introductions could be given for each model by those who are familiar with them. Afterward, we could compare the models against each other by performing some on-the-spot simulations, and perhaps also rate them against measured generation data. If anyone has open validation data they would like to share, this would be greatly appreciated (I will bring some). I can also bring weather extracts from MERRA2, ERA5, and COSMO for us to play around with.

Would you like to be responsible for this Session?
Yes

Do you need any special infrastructure for this Session?
Projector, Each person with his/her own laptop

Do you have any recommendations who could be part of this Session?
Anyone interested in this topic is, of course, welcome to join. Moreover, I’ll mention those which I know have had a hand in the development of some of the models I mentioned above. Apologies if I forgot someone.

4 Likes

Hey Severin,

I think it is a good idea.
I am not able to attend the workshop in January, but could assist you with some thoughts.
I am myself culpable because I created such a tool (https://github.com/tum-ens/renewable-timeseries). When I started, I was not aware of any tool that would assess any shape of regions and provide its potential and representative time series, so I created my own tool and published it in GitHub.
However, I tried to include some correction based on the Global wind Atlas, to match IRENA full-load hours if needed, and to use historical time series / EMHIRES for the regression.
I realized quite late that I could use PVLib to avoid dealing with the physics with solar radiation… but it was instructive nonetheless.
Now that I have done that, I realize that there is not such a thing as the perfect time series. If you use reanalysis data, you will necessarily deviate from measured data (either from power plants or even weather stations, in some regions of the world there is a non-negligible bias and/or time delay). And if you only rely on measured data, you cannot get a full coverage. If you do a regression to match historical generation time series, you might not be able to estimate how the time series should look like if new sites for power plants are added in your region… So there will always be a certain variety in tools. But we definitely should consolidate what exists, and make it visible so that others do not have to reinvent the wheel.

Cheers,
Kais

2 Likes

Hi @kais_siala You have a nice full load hours (FLH) solar map of Australia in 2015. Have they or do researchers need to consider the impact of smoke haze from bushfires in these projections? Does that degradation come in through the historical weather data anyway? But surely many more bushfires in the future? Just thoughts, R.

References

Doherty, Ben and Helen Davidson (6 December 2019). “Australia fires: five blazes merge north of Sydney as conditions forecast to worsen”. The Guardian. London, United Kingdom. ISSN 0261-3077.

Hi Kais,

I think I went through the same experience during my work as well :slight_smile:. Plus your need to “assess any shape of regions and provide its potential and representative time series” is more or less exactly my use case as well. It would be really interesting to see how we’ve both addressed the same problem.

It’s a shame you cannot join, however, If you would like, I would be happy to discuss the issue with you beforehand (perhaps over skype?) so that your perspectives are still “included” in the do-a-thon?

Very sad what is happening over there.
I haven’t considered the impact of smoke haze yet. We have to see whether it is limited in time or whether it is here to stay…

Sure, just write to me (preferably per email) and we can arrange for a skype call.
Cheers

Small update: The wind and PV simulation code I’ve written from my PhD is up on GitHub. It is called RESKit, and is at the moment super hacky.

The documentation is sparse, and I have many small things to clean up or otherwise take care of. So don’t mistake it for a final version. Nevertheless, the core functionalities are reliable and I’ve added a bunch of examples which more or less show how things should be working.

Text and images licensed under CC BY 4.0Data licensed under CC0 1.0Code licensed under MITSite terms of serviceOpenmod mailing list.