Like most conferences, FOSDEM was entirely virtual in 2021 and 2022. Having attended every edition since 2017, last year’s return to meeting in-person was a success and now I can confirm it has completely recovered. It’s back in full swing in 2024, I’m not giving a talk this time but definitely enjoying the vibe. The smell of chips, waffles and chocolate is strong as ever. And of course, there is a huge quantity of high-quality beer in Brussels. The first stop on Friday night for many attendees, myself included, is at the Delirium where you can easily bump into old acquaintances and other pink elephants.
Now let’s take a look at the schedule. Topics keep evolving gradually over time as does the tech world, in particular it’s now the second FOSDEM in a row with an Energy devroom which is where I spent the best part of Saturday to recharge my batteries. It’s also the main topic I want to cover with this post. Other devrooms I would recommend are the new AI & ML obviously and always Community which is a great source of inspiration about how to advance open source adoption in general.
Lots of Energy
Electricity has obviously been around for a very long time. However, it tends to lag behind the development curve followed by other grid-based technologies such as software-defined networking. I found myself among a majority of attendees who did not work directly in the energy sector or didn’t know too much about all the standards and details but clearly shared a growing interest. So it’s kind of early days for this industry which is about to entirely reshape itself, the open-source way.
Electric vehicles need charging points, renewable energy sources require more flexibility in the grid, remote living areas make a great case for decentralised power stations, climate change awareness is rising… There are lots of compelling reasons for building a clean, inclusive, high-tech energy grid.
Before diving into particular talks and topics, here are a few key things that I learned and struck me as really important for anyone joining this field like me:
- Open-source software has a big potential for defining electricity grids in general and managing substations in particular to convey energy efficiently - see the IEC 61850 standard.
- Open-source hardware is also picking up, in particular with projects such as OwnTech.
- A big presence at FOSDEM was LF Energy. It’s a bit of a maze at first glance with lots of projects which may contain a tad too much Java code to my taste, but it’s early days and parts of it are really inspiring such as CitrineOS.
- There were lots of stickers everywhere about Climate Triage which appears to be a great resource for navigating the open-source energy world.
- Machine learning is of course gaining territory here as well. A great example is Quartz Solar which aims to improve solar forecast and help reduce usage of backup power generators based on fossil fuel.
It’s my first blog post on energy and I must say it feels like the very first steps of a long journey!
Among all the buzz, here’s a handful of summaries I cherry-picked from all the talks I attended:
With my university background in electronics, I was truly delighted to discover OwnTech which is a project backed by the CNRS to produce open hardware for managing electric loads. All the code and design documents can be found on the LAAS GitLab.
There are two main boards which have only been available in universities for early experiments so far but production has just started now:
- TWIST: a power
- DC only for now, up to 110V / 300W
- can be grouped to drive higher power loads
- target price: ~300€
- SPIN: a controller
- running the Zephyr RTOS
- using RS485 for fast communication
- target price: ~35€
This should lay some solid foundations for prototyping which will help accelerate innovation in this field. It’s definitely worth keeping on everyone’s radar.
Starting from the problem statement that about 25% of the electric vehicle charging stations in the USA aren’t operational (as in most places), CitrineOS is a great step forward in improving the whole situation there. It’s a thing called a Charging Station Management System (CSMS), which implements the OCPP protocol. It has also been adopted as a LF Energy project.
Why is it great? It was designed with particular requirements in mind, for example it uses a messaging system rather than some clunky database-driven synchronisation which greatly helps coordinate the charging stations with a central backend. And of course, it’s open source with all the benefits that brings.
Some technical software bits:
…and it’s looking for new contributors!
This was another great talk. A picture is worth a thousand words:
This open-source project is essentially providing an IEC 61850 Substation Configuration Language editing tool. It’s deployable as a web application to be more easily adopted in environments where IT security rules don’t allow third-party binaries.
It has a very modular plugin architecture using Lit. You can find out more on GitHub and it was so rich I can only recommend watching the whole talk once the recorded video is available on the FOSDEM website.
Created by Open Climate Fix, the Quartz Solar project provides an open-source implementation for a solar forecasting service to anticipate energy production coming from photovoltaic (PV) panels. As solar energy is not constant, the grid relies on backup sources or “spinning reserves” which are typically fossil fuel generators such as gas turbines. Those would be normally running at say, 50% of capacity and brought up on a cloudy day to compensate for the lower PV output. Predicting with more accuracy when the sun is going to shine means the spinning reserves can be dialed down more and for longer periods of time to maximise the PV utilisation instead - hence reducing the fossil fuel usage.
This is where Quartz Solar literally shines, using machine learning models trained on satellite imagery such as EUMETSTAT and PV measurements gathered all around the UK (Open Climate Fix is based in London). The models are now mature enough to be used virtually anywhere in the world, so electric vehicle charging stations and even individuals may start making use of them locally with their own panels. The models themselves and the datasets used for the training are all available on Hugging Face. This of course results in a growing community of users and energy producers.
Results are very promising already: the UK’s National Grid solar forecast horizon has now increased from 2 hours to 6 hours. This has enabled true real-time grid balancing between PV farms and spinning reserves. Actual solar forecast capabilities of Open Quartz are actually up to 48h, and more research is underway to push the boundaries even further.
To close the conference’s main track on Sunday, a discussion panel was formed to launch the Open Renewable Energy Systems (ORES) project. The main message was that there is now a clear opportunity to drive a revolution in how energy is being distributed using open-source software. Defining open standards, building more flexible and modular systems to reach out to isolated communities, creating microgrids and enabling individuals to generate and trade their own clean energy locally are some of the new challenges to overcome in the next decade or so.
Renewable energy is a lot more diverse than traditional fossil-fuel sources and can often be deployed with small decentralised generators. This also means it can be produced closer to where the consumers live to extend electricity coverage and reduce the cost of transporting electricity. Doing so requires more flexibility for balancing loads and power sources at short notice, and interoperability is critical for this to work. Some early experiments have shown that it’s possible for citizens to produce their own electricity and sell their surplus to their neighbours without infringing regulations - but standards, tools and infrastructure to facilitate adoption still need to be created.
There’s a bright future ahead of us!
Last modified on 2024-02-03