Grid Singularity and rebase.energy (Rebase), two open-source tech startups, collaborated in the framework of an EU-supported AI4Cities project to integrate a new, custom photovoltaic (PV) tool which allows users of the Grid Singularity Exchange to benefit from location-based PV generation data when simulating different configurations of energy assets and considering the optimal energy setup of a microgrid or an energy community. The custom PV tool is connected to energydatamap, which generates data profiles based on the asset’s capacity, location, related weather data, azimuth, and tilt.
Grid Singularity’s vision since the beginning has been to develop the most optimal energy exchange, and this is what we have done — our Exchange consists of an open source codebase and interactive interface for simulating and operating interconnected, grid-aware energy marketplaces, with the aim to create and operate peer-to-peer energy markets. By simulating a “digital twin” of your house and community, you can assess the financial and social benefits of local energy trading, as well as the impact of specific actions or individual assets in the community. Our modular Exchange and real-time results allow you to build and deploy your energy community in line with the market mechanisms allowed in your country, whether that’s collective energy sharing schemes, peer-to-peer trading, or something in between. Based on our experience and user feedback, we continuously work to roll out new features. That’s why we teamed up with a long-term partner of ours, another open-source energy startup rebase.energy, to apply for the AI4Cities grant, which was awarded to us in 2021 🏆 With this funding, we integrated our partner’s highly useful online tool — the energydatamap (Figure 1) with the Grid Singularity open-source platform, enhancing the value provided for simulating and operating local energy communities.
To give some more context, AI4Cities is a three-year (2020–2022) EU-funded project, bringing together six leading European cities seeking artificial intelligence (AI) solutions to accelerate their carbon neutrality in terms of CO2 emissions reductions and other climate commitments. Is it more obvious now how we fit into the picture? 😁
PVs — again?
With PVs being (literally) a hot topic at the moment, our custom PV tool has come at the perfect time to enable users to better simulate their PVs and take a more active part in the energy transition.
Most energy simulation tools like PV*Sol online use generic PV template profiles, typically representing the standard PV profile based on the geographic location and associated weather profile. Some others, like Sisifo, require more technical knowledge as they ask for detailed system parameters. The custom PV tool, therefore, brings enhanced results with minimal user input.
Currently, the Grid Singularity Exchange allows users to add and assess the performance of energy assets in their simulated homes and communities. These assets, including consumption profiles, solar panels, and batteries, are represented by digital trading agents and enable users to configure and optimise a local energy market structure digitally before bringing the energy communities to reality. They can also use the simulation to assess any further investment in energy assets. When adding solar panels (PVs) to the simulation, users can now choose between a) template PV profiles (Figure 2), b) custom profiles based on the geographical and weather profile of 6 different cities developed in the collaboration we describe here, or c) uploading their own PV data. Each PV component can represent a single panel, array of panels or an entire solar park, with the possibility to add these to a single home or a community.
This new feature allows Grid Singularity Exchange users to model PV production representing their specific geographical location and weather profile without uploading any personal data, better estimating the benefits of future PV installation (or adding renewable capacity) for their home and community. The user can adjust multiple parameters, such as the capacity and the orientation of the asset (Figure 4). What happens “behind the scene” for this process is that the Grid Singularity Exchange sends an application interface (API) request to energydatamap to generate the equivalent profile depending on the users’ building/community’s location (Figure 5).
While we were working on the new custom PV tool, the partnering rebase.energy team was busy further developing Enerflow, an open-source toolkit to calculate the optimal investment for each asset based on estimated operational cost based on an asset’s location and other inputs. It is based on Pyomo, an open-source toolkit, which is able to solve a wide range of optimisation problems (including integer linear problems that use only integer values as investment and operation costs input, and mixed-integer linear problems that use a mix of integer and continuous values as inputs). This toolkit is combined with one of the best-performing open-source lean programming solvers, the CBC solver to obtain the most accurate solution for the optimisation model. With this cutting-edge technology, Enerflow allows energy modellers, building managers, and households to optimally size the installation of additional energy assets (PV and batteries) with the objective of reducing CO2 emissions while improving the Return of Investment (ROI). Pretty cool, right?
We think so. But if you’re like us, you’ll want to know this process in more detail — what is involved and how does it work to benefit cities in developing renewable energy communities exactly? Let’s get to know the steps required by examining what happened in the city of Helsinki, which has the goal of becoming Carbon Neutral by 2035.
Keep reading for a sneak peek into the cities’ storyline:
First, cities need to vote and adopt new energy directives. Once directives have been adopted, cities establish energy project committees to work on and achieve these long-term goals.
Energy committees get to work and create projects harnessing third parties’ advanced tools 😉. They look for energy retrofit/upgrade opportunities in the city’s building portfolio or enable other large building owners to find clean energy projects. After that, energy specialists come into the picture and assess the benefits of new investment opportunities in the cities’ building portfolio and new business models driven by energy communities regulations. For this, the Rebase platform comes in handy. Once we have simulated energy assets (e.g. solar PV, batteries, heat pumps) we can plug it in to right size these assets and place them most effectively. In this way, the Grid Singularity Exchange becomes an enhanced tool that allows engineers and project planners to simulate Local Energy Markets (LEM) in a community, enabling them to experiment with multiple asset configurations. When the best solution is found, energy specialists have more information to help them to decide to invest and deploy the relevant projects and new assets in reality, where regulation allows. The final objective is to deploy and run the energy communities in real-time.
At the end of this AI4Cities project, a prototype was developed of a simulation of an optimised local energy community, which consisted of the following steps:
- Model the energy community in the Grid Singularity Exchange: go to the Grid Singularity map and start building the community to create a digital twin of reality (more information on how to do this can be found here)
- Run the simulation of the local energy market (with enabled peer-to-peer trading and interaction with the grid) for this community
- Export simulation data: two files are generated from the modelled community using two data APIs that we have developed in part in the framework of this project:
- Grid Data API: This API provides information about the structure of the user’s energy community, listing all the different configurations of energy assets in the modelled energy community, including their location (latitude and longitude).
- Historical Data API: This API provides historical consumption and generation profiles of assets in modelled buildings.
- Upload the generated files to Enerflow: based on the community configuration and historical consumption data, the Enerflow tool will optimise the added PV assets to propose the best setup for each building in terms of energy use and costs. The results generated by Enerflow will show the best setup for each building.
- Update the configuration of the simulated energy community with the new PV setup: add PV installations for each building based on the Enerflow results using the Custom PV Tool.
- Simulate the optimised version of the local energy market: rerun the simulation on the Exchange with the now optimised energy community setup
- Repeat steps 3–6 to optimise the setup of the local energy market and thereby your investment in the right capacity and allocation of renewable assets like solar panels.
How it works in practice:
In essence, what we did was create a digital twin of a small four-building cluster in Helsinki and model it as a local energy market in the Grid Singularity Exchange. These four city-owned buildings, located in the northeast of the city, had no generation assets and thus were only consuming electricity. Their profiles were downloaded from the Nuuka API, an open application interface where hourly-resolution energy data from different city-owned buildings in Helsinki can be requested.
Once this community was configured, we ran the simulation using one week of historical data. The initial results showed that the community had a high solar panel investment potential since the consumption pattern (Figure 6) increased during the day, when the sun is shining, and reduced during the nights when there is no solar generation.
Two data files were generated using the Grid Singularity Exchange data APIs (Grid Data API and Historical Data API), and these were then uploaded to Rebases’s Enerflow to calculate the optimal PV sizing specifications for each of the four buildings.
Lastly, we integrated the optimal PV results from Enerflow into the custom PV tool and ran the simulation for the final time. The results achieved by the modelled Helsinki energy community are as follows:
- self-sufficiency (49.5%)
- self-consumption (88.3%)
- reduced peak congestion (21%)
- reduced electricity bills (52%) were achieved if PVs were added to 3 of the 4 buildings, with one week of local trading simulated (please note that PV amortisation was not taken into account)
To simulate or not to simulate — there is no question
All in all, rebase.energy and Grid Singularity have completed the development of an AI algorithm toolkit and the technical design to integrate Rebase’s open-source custom PV optimisation into Grid Singularity’s open source local energy market Exchange. Taking part in this project and creating the custom PV tool has only laid ground for us to keep expanding and pushing boundaries.
In future iterations, the AI algorithm toolkit will also be integrated in the data and algorithm marketplace, allowing the provision of valuable data / algorithm services, achieving full functional alignment of these two complementary open-source solutions for energy optimisation.
For cities and prosumers, we provide the means to join the new local energy economy, and be rewarded for energy data they choose to input or to optimise how their assets perform.
For aggregators and service providers, we have created a nascent ecosystem of dataset access, algorithms, with reporting to inform their interactions with the energy market. Incentive structures are designed to reward participants for valuable contribution and incentivise AI innovation, improving metrics for communities (e.g. energy bill, self-sufficiency) and grid management. Through the data and algorithms marketplace and the open source toolkit, we democratise AI algorithms and promote openness and transparency.
So, what are you waiting for — start your simulation today! And for those of you hungry for more, take a look at the AI4Cities Phase 2 prototype demo
This article was authored by Christopher Dietrich with contributions from Sebastian Haglund El Gaidi, Ilias Dimoulkas, Natalija Bytniewski, Grace Mullin, Luis del Aguila Escamilla, and other colleagues at Grid Singularity.