 Good morning. Good afternoon everybody. Happy to see you all here. I see some familiar faces and also some new faces, so that's good. Let's start with a quick introduction. Let's see if the techniques is working today. It seems not to work. Then we do it the old way. Hi, I'm Jonas van Bogert, Solution Architect at Alieander and one of the members of Alieander's open source program office. And I'm here today with Nico. Yeah, just like Jonas, I also work at the open source program office and besides that recently I've been a DevOps consultant and for those who are not familiar with Alieander, let me give you a quick overview. Alieander is the largest distribution system operator in the Netherlands and we develop and operate energy networks and to give you an idea, we are nearly we're nearly responsible for six million customer connections, manage over 90,000 km of electricity grid and manage over 40,000 km of gas grid. And so our key task to make sure the households and companies stay warm and have access to energy at all times. And today we'll tell you more about the new challenges that the energy transition poses to distribution system operators like Alieander and how open source projects like OpenStaff and the PowerGrid model help Alieander to tackle these challenges. I will start off with a quick overview of those challenges. Next we'll dive into what FighterWall OpenStores plays in realizing new digital solutions and new digital capabilities. After that we'll introduce the OpenStaff project and next I will give the mic to Nico and he will introduce the PowerGrid model and he will also end the presentation with a short overview of the other open source projects we are involved in. And hopefully there will also be room for some questions at the end. So let's get started with the key challenges a distribution system operator is facing today. I see the slide is not turning up that well, so you have to do it with my story. In 10 years time the way we live, work or travel won't be the same. And this is necessary since fossil fuels are becoming more scarce and have a growing impact on our environment. And you see that everyone in the world new initiatives are emerging to conserve energy and make it more sustainable. And we see that the way we generate, distribute and use energy is changing rapidly. Fossil fuel power stations are closing down in order to reduce CO2 emissions. We see enormous increase in the use of wind energy and solar energy. And also the number of electric cars is growing rapidly. And distribution system operators like Allende face major challenges to support this transition. In the Netherlands we have one of the most reliable electricity grids in the world. However, this grid is not designed with the rapid developments that are happening today. The demand for electricity is growing so fast that it's driving the grid to its limits. And we already see that in many places in the grid, there's currently no room for extra electric power for new business consumers or business consumers that want to expand their businesses and are in the need of new and extra electric power. Even though we as Allende have significantly increased our investments in expanding the grid in the last few years. And this requires new smart solutions to better use the capacity available in the grid. And the goal of a couple of these smart solutions is to stay a supply and demand in periods when the maximum grid capacity is reached. And this can be due to too much supply. Think of a day in a summer where there's a lot of sun and wind. But it can also be too much demand. Think of a winter period where there's nearly no sun or wind. And this steering of supply and demand is also often referred to as congestion management. And congestion management can be in the form of direct control. Such as the curtailment of the excess of renewable energy. Or in the form of a market-based mechanism where pie signals are used to steer and incentivize grid parties to adjust their supply or demand. And when we as Allende are active managed congestion, we can create extra room in the electricity grid for new business consumers and business consumers that want to expand their businesses. And to perform active congestion management, new digital capabilities and solutions are required. And here open source plays a vital role. By using open source, participating in open source projects and open sourcing internal projects and being a member of Eleve Energy, Allende aims to benefit from the power of open source calibration to realize these new solutions and capabilities for active congestion management. And there are multiple projects Allende is involved in, but there are two projects in particular I would like to highlight today. And that's the open staff project and the power grid model project. Both projects are developed internally at Allende and recently be open sourced. So let's get started with the open staff project. Open staff is a Python package which is used to make short term forecasts for the energy sector. And it's specific, it's making forecasts of the load on the grid for the next hours to days. And why is this important? If Allende wants to perform active congestion management, it needs to know what the load on the grid will be so that it can date a man if the maximum grid capacities reach. Also it needs to know this in advance so there's sufficient time to take action when it's needed. And besides active congestion management, open staff is also an enabler for grid safety analyzers and data mining grid loss. So how does open staff work? Open staff provides a fully automated machine learning pipeline which produce forecasts for the loads on electricity grid for the next hours to days. And for those who are not familiar with a machine learning pipeline, let me explain. A machine learning pipeline is a way to codify and automate the workflow it takes to produce a machine learning model and deploy it. A machine learning pipeline consists of multiple steps from data extraction and data processing to model training, optimization and deployment. And open staff in particular encapsulate all the learn best practices of producing a machine learning model for load forecasts on the electricity grid. And it also allows users to execute this pipeline at scale. At input, open staff can use multiple data sets for multiple data sources. Also it's possible to add new data sources or new data sets. Key input data for Allende at this moment is historical load and generation measurements, weather data and data for market parties. Open staff performs validation on this input data and combines the historical load and generation measurements with external predictors like weather data and market prices. In the next step, Open staff uses this input data to train a model and any skiescape learned compatible machine learning model can be used. The current model that we use is inspired on the extreme gradient boosting algorithm. And in the Allende use case, we train models for each substation. And these models are after training optimized and deployed to provide forecasts on the load on these substations. And these forecasts are renewed every 15 minutes. And also these forecasts are by Open staff continuously evaluated so that in case that the forecast underperforms, it will be re-trained and re-optimized. And in the final post processing step, both the models and the forecast will be stored and made available for applications and users. So to give you an example, you see an image, I hope it's a little bit visible, where you see one of the forecasts that's performed for Allende substations. As you can see, we make forecasts for different time horizons for one hour ahead, four hours ahead, 24 hours ahead and more. And this substation specific is an area where there's a lot of wind production. And as you can see, there are multiple times that the load on this substation is negative, which means that there is times when there's more energy produced in this area behind this substation than consumed. And in addition, Open staff can also provide forecasts for the amount of solar energy and wind energy generated. And since solar energy and wind energy have by far the greatest uncertainty, it helps have insight in how these components contribute to the total load in that substation. As already mentioned, Open staff by itself is just a Python package. And there are multiple ways to implement Open staff in your application or solution. In this slide, we show you one example of how Open staff can be implemented. This implementation is also inspired on the implementation we have currently at Allende. It's based on an open source stack and organized in a microservice architecture and optimized for cloud deployment. And since last year, Open staff project is hosted under the ElevEnergy Brumbella. ElevEnergy is an open source foundation focused on the power system sector hosted within the Linux Foundation. ElevEnergy provides a neutral collaborative ecosystem to build shared digital investments that will transform the world's relationship to energy. So how does ElevEnergy help us in the Open staff project in particular? There are four things that stand out for us. First, Open staff provides us a neutral ecosystem to host open source projects and collaborate with other organizations both within and outside the energy sector. Secondly, ElevEnergy also provides us with the know-how on how to best leverage and adopt open source batch practices. And third, ElevEnergy also helps us with IP and legal related questions. And fourth, ElevEnergy is also a great network organization to get in touch with other organizations working on open source within the energy domain. Today, ElevEnergy has over 50 members and this includes distribution system operators and transmission system operators like RGE, Tenet, Energynet and Statenet, but also several universities and large corporates like Microsoft, Google and GE. And currently, Open staff is one of the 18 projects that is hosted within the Linux Foundation of Linux Foundation Energy. And these projects have a really broad scope. It includes projects in application domain like Open staff, but it also includes projects in the edge domain like Flashpower. So back to Open staff. Although Aleander is still the largest contributor of Open staff, the project community has become more diverse. Last year, RT, the French TSO and also a member of ElevEnergy started to contribute to Open staff and will also plan to replace their legacy forecasting solution by a solution based on Open staff. Another contribution by Open staff is made by the Aachen University. Aachen University implemented the sport of the Sonya Prolof model in Open staff. And the Prolof is a probabilistic load forecasting model that can also be used to forecast the load on the electricity grid. With Prolof included, Open staff currently has two models and sports two models ends up to the user to decide which model they want to use in which use case. And we would love to see the Open staff community grow further and become more diverse. So how can you get started with the Open staff project? If you want to know more about the Open staff project or read the documentation, please visit the Open staff project website on the ElevEnergy website. If you want to get started right away, you can also visit the Open staff Github community through the link on this slide and also don't hesitate to raise any questions on the Open staff mailing list. And now it's time to introduce our second and newest Open source project, the Power Grid model. And to tell you more about our Power Grid model project, here's Nico. Yeah, thank you. Let's see that on the left is on the right. Yeah, so I'm going to tell you about Power Grid model. And this is a software solution we typically use for power system modeling. More historically, ever since we've been using this kind of software, this will be run at specific points in the process where, for example, we want to do an expansion on the grid because the capacity is not sufficient. So what we take as input parameters is the current or the to be grid configuration. We take information about what do we know about the consumers or the connections, what do they use? Do they have specific usage patterns? And we run some simulations to see is the plant grid sufficient? Will it stay within grid code? Is the voltages, will they be within the grid code? If there's a short circuit, how will that be handled? That's all designed and verified tested by our engineers. And then eventually, they'll be executed. As you can see, this will be sometimes maybe once a year, sometimes a couple of times a year, but not so frequent. Now, in the last decade, as Jonah said, the world has changed. So starting with, I think, most important for Alayanda was this holiday park experiment. So there was a holiday park where they added solar panels on the roof. And with this access solar generation, we also saw issues with voltages. So we added the battery to control and reduce the load on the grid, also collect more energy during the day and maybe optimize it. So when the sun isn't shining, we can use that energy. And there was a control system in place. But as it was very uniform, we had similar houses with the same panels with similar orientation. Every house was measured. It was quite easy, relatively, to write the software and it was a custom solution. Now, later on, we did a less controlled experiment in a neighborhood with a similar situation with lots of solar panels. The grid wasn't sufficient. We added the battery. But in this case, we didn't have measurements real time measurements of the houses. So we only had a measurement at the substation with the transformer and one at the end of the feeder with the battery. And we need to constantly monitor to see what are we going to do because maybe there's clouds in between the wind, the consumption might change. And a new control system was developed that included some power system analysis. So this is no longer the process where maybe once a year you run a computer program by human, but rather this is an automated process where you have to do the calculation continuously. And this was the beginning of the power grid model. So in general, we saw an increasing need for more calculations. So we had to do more, we had to incorporate a larger part of the grid to come to reliable results. We had to do different types of simulations, a larger variety of things to calculate. And we had to do it more frequently. So on multiple axes, we had an increased need for these calculations. At the time, the idea was maybe to centralize everything with grid, but eventually we came to a more fundamental approach where the power grid model came from. Looking at the ecosystem at that time, and this is a re-interactive table, but the commercial offerings, they were good in all the features, but they didn't integrate as well. They were designed for humans, desktop applications. There were open source solutions at the time already that might be, that would integrate as well better, but they weren't as optimized, mostly, and the performance was lacking. So we'd say that the power grid model is, it doesn't cover all the scenarios, but for us it's good enough and it integrates well and it performs well. So going from that battery experiment in the neighborhood, that code was originally written in R. Later on we did a rewrite in the last couple of years to make it fast in C++ and Python. But more importantly, I think, is all the development that happened in between, where we had scientific research to come up with new models to linearize this typical way of calculating. Normally you do a Newton-Raphson method, and if you know a bit about math, but you iterate between the voltages and the current, and eventually the solution converges, and when you're in an error bound you say, okay, this is the solution. But it takes a couple of iterations to come to the conclusion, so it's not as fast. If you can linearize that problem, you're there sooner. But who knows, you introduce an error if you linearize it. So with this scientific proof, we validated that this linear method actually works for us, and for most of the cases is good enough, and then we get a major speed boost. So a lot of people work to make this a reality over all those years. Now this is a benchmark. You can find up the notebook and check it out yourself in comparison to another open source project that is there on this field. It's PandaPower, also a great project. If you take that Newton-Raphson iterative method as the base, the Newton-Raphson method that is here in a power grid model is 25 times faster. If you then use the linear method, you can get up to 70 times. And if you then normally would say there's a three-phase system in the grid, we simplify it. It's all the same. But if you want to have a more detailed analysis and you want to actually calculate those three feeders separately, if you have maybe some households connected to one and some to the other, and that's relevant, then of course the problem space becomes larger. In that case, we can close to 500 times faster with the linear method for this larger grid. So this is immense. It's not like 500%, it's 500 times. So we're now to the point where the actual data conversion loading the grid and getting the results out is the bottleneck and no longer the calculations. And going back in time, like 10 years ago, this was unimaginable that we'd be at this stage. So what it can do, it can do the power system analysis, the power flow calculation in a number of ways. It can also do state estimation. So if you have real-time measurements, maybe you have some measurement errors, you have lacking measurements, it can deal with that as well. And also it can do data conversions from a number of formats. And most notably, I think, is the SIM format, which is used in industry as a standard for modeling these grids. So it fits in the larger open source, open standard ecosystem. In the meantime, because of all the history internally, we use it in a variety of software, over 10 projects at the moment, whether it's used by humans, if it's used by computers, it's such a convenient building block because it's just a Python library, so everybody can include it. All the data scientists know how to use it. And also now it's part of the GridCal software. Recently there was a LinkedIn post last week. So in that way, it's already adopted in other solutions outside of Alliander. And that brings the circle back around because now it's also included in the desktop application we sort of started off with. So if you want to know more about PowerGrid model, get involved with the team, they're really happy to help because they see the value, they know the value of this project and they want to make it a success and get it integrated in other solutions. So here's the contact information. Also notice the link below. If you're interested in getting in contact for a webinar or a workshop, there's a form, you can leave your details there and the moment something pops up or you can schedule something, we can get in touch. So this was another of those projects that we took from Grid in Alliander outside. There's some other projects we're involved with. I can tell a bit about all of them, but for example, take the GXF project. This is, you could think of it as a generic IoT platform for data communication, but it is used because it's so secure and scalable. We use it internally for controlling public lighting and reading out all smart meters. So it has to be secure and reliable and it is. We have a weather provider API that's also open because we use a lot of weather data, but for data scientists, it's not always convenient to access that data because do they download all the data sets on their own computer? Sometimes the formats change. You want to combine multiple data sets. We host that internally so our data scientists and applications can get just the data they need in the format they want. We work with a compass project together with RTE and other companies where the idea is that if we can, we have substation automation, so all the control systems, controlling switches in a substation that need to be configured rather than configuring everything by hand and going over the configuration. What if we can generate that from the data we had, like a topology and a network planning? And then if something changes, we can automatically update that. So really streamline the whole automation. So on multiple fronts, we're really trying to be as innovative as we can and also do this collaboratively. Now, if you want to know more about the bigger picture and what we're doing with open source at Allende, here's some contact details and of course we're happy to discuss further on that. And wrapping this up, we hope to have shown you that the challenges, you read about it every day on the topic of energy that the energy industry is now facing. What that means for us as a grid operator and how we are using open source as a means to come to a solution with innovation rather than just, well, doing a conventional method because we need every innovation we can get. I was going to say something else, I think. But yeah, so if you're interested, you have the contact details and we're happy to answer questions. But do you do anything around natural gas? Yeah, we're also involved with gas. Yeah, that's a concern of us. Mostly we're now focused on electricity. Some might be applicable to gas, but no, at the moment it's focused on electricity. Yeah, but I don't see why we cannot focus if the opportunity is there for gas. Maybe to add on that. So I think it's in the scope can be both gas and electricity. You see at this moment there's more innovation and digitalization efforts going into the electricity grid or the electricity domain because there the challenges are the biggest. So you see a lot of, at least in the Netherlands, you see that the gas consumption is going lower. So there we have less struggles to grow in the need to fill the gas dispersion and transmission needs in comparison to electricity because a lot of gas energy consumption is moving to the electricity domain and that's really creating pressure on the electricity grid and that's why we're seeing the most digital investment in that area. And also there we see now at this moment the most opportunities for open source collaborations. That may change in the future. More questions? I see there's no more questions. So as already Nico mentioned, I think it's a challenge we are not faced by ourselves but I think we are facing as a world. So I really want to invite everyone who are interested in this domain to reach out to us or to Eleventy in general. We mentioned a few projects but I've already mentioned there are 18 projects in total and also more on the pipeline. So if you have interest in this domain or you see chances to collaborate, we invite you all and also if you speak others on this summit related to energy, please refer them to Eleventy or to us and we're happy to connect and explore synergies and new opportunities and hopefully make the step to a more sustainable and cleaner world a bit faster. Thank you everybody.