Teach-a-thon: Showcase of input generation procedures

I think we can all agree that performing energy system analyses is a blast! But producing the inputs for our models is generally not. Nevertheless, how can you model an energy system without, for example, RES generation profiles, demand profiles, or storage options? Most importantly, we want to ensure that the products of our analyses are comparable to those of our peers and our former selves. It’s therefore of the utmost importance that we take great care regarding how we generate the inputs we need (ensuring reproducibility and consistency as much as possible). Fortunately there are at this point a number of easily-followed recipes and open-source tools available which we can use to lighten this burden.

Some examples of common model inputs (Please suggest others to add):

  • Regional capacity values for RES technologies (most notably wind and pv)
  • Sites of future RES installations
  • Managing large weather datasets
  • Time series generation values for RES technologies
  • Time series demand values (electricity, heat, hydrogen)

I propose an interactive teach-a-thon in which those of us possessing experience with these procedures and tools can exemplify their use. The format could be one where several people each have a chance to showcase their experience, followed by a short question and answer round. In the end, we can have a group discussion regarding lessons learned. From this, we can collectively identify common methods and perhaps even identify roles which future tools could fill. At the very least, such a program could surely teach each of us something new, since no one is an expert in everything!

In the examples listed above, I have highlighted the first three points as I feel that I can offer expertise in these areas. So I would of course offer to be one of the people filling the ‘teacher’ role for some of the time. However to prevent this introduction from becoming too long, I will explain each of these points in the comments below.

15 Likes

RES Land Eligibility:

In regards to the first item, determining the wind energy, PV, or other RES capacity potential within some region is conceptually fundamental to designing an energy system for that region. This is tied to determining the ‘Land Eligibility’ for the technologies in mind and is constrained by a number of criteria; such as terrain slope, distance from settlements, and nature conservation. This is inherently a geospatial question, which is commonly evaluated using GIS applications like QGIS and ArcGIS. But performing land eligibility analyses with these tools is cumbersome and highly reliant on manual effort. I therefore created a tool called GLAES to perform this task programmatically; which is already available on GitHub as open source software. In just a few minutes I can show how one can compute the land eligibility of any RES technology for any regional definition within Europe, which can then be used directly to compute a reasonable technically-installable capacity of the technology in question within that region.

RES system siting:

Sometimes we know that a certain capacity of some RES technology should be placed within some region. But we don’t know precisely where it should go. This is important since weather is fickle, both temporally and spatially. By building on my previous point of land eligibility, I would like to show further functionality of my GLAES model which can be used to identify potential installation sites and then use multi criterion decision analysis to decide which of those sites should be selected. In the same way as before, this is not a method I have developed myself, but is rather a practice I commonly see in literature which I’ve simply made more accessible.

Weather data management:

In order to perform RES generation simulations, estimate heat demands, or other such application, researchers commonly look to using the outputs of climate model simulations (like MERRA, ERA5, COSMO-REA2, ect.). But there is a problem with this…weather datasets are huge. Like, really huge. So when we want to extract weather information from some source, the situation can become unwieldy quickly. Luckily, the climate modeling community shares in our woes, and has for this reason built an amazing tool called the Climate Data Operators (CDO) which can do a lot of the heavy lifting for us (although you will still need several terabytes of hard drive space if you want to do evaluations at the European level).

In the progress of my research I have become very familiar with the CDO tool and would be happy to summarize its usage. I can also point out the lessons I’ve learned, and can lastly show some tricks I’ve picked up in dealing with such large datasets.

Looks good @s.ryberg!

RE your two outstanding points:

RES timeseries generation: perhaps @stefan.pfenninger could do a run-through of using renewables.ninja?

Demand generation: Fits in with another do-a-thon, the existence of which suggests that it’s a big stumbling block, in need of a manageable solution…

The glaes package is very nice! Thank you very much!

It would be great if you can attach the notebooks you used.