# https://gridfm.org


## Pages

- [Argonne](https://gridfm.org/argonne/): 3rd In-Person Workshop: Foundation Models for the Electric Grid (11-February 13, 2025) · Indico Skip to main content 3rd In-Person Workshop: Foundation Models for the Electric Grid Feb 11, 2025, 8:30 AM → Feb 13, 2025, 12:00 PM America/Chicago Room 1416 (TCS Conference Center) Room 1416 TCS Conference Center Argonne National Laboratory 9700 S. Cass Avenue Building 240, TCS Conference Center (north. entrance) Lemont, IL 60439 +1-630-252-2000 Kibaek Kim (MCS) Description Foundation Models for the Electric GridFebruary 11-13, 2025 | Argonne National Laboratory We are excited to announce the 3rd in-person workshop, Foundation Models for the Electric Grid, hosted at Argonne National […]
- [Workshops](https://gridfm.org/workshops/)
- [Harvard](https://gridfm.org/harvard/): Building on the successes of four previous workshops on Foundation Models for the Electric Grid at Yorktown Heights, Imperial College, Argonne National Laboratory, and RWTH Aachen, the GridFM community will get together for its 5th event at the Harvard John A. Paulson School of Engineering and Applied Sciences in Boston, MA on March 17th – 19th, 2026. We hope that we all see you there. Registration and a call for contributed presentations/posters will go out at the end of December. Please stay tuned. Tentative Agenda: March 17th Time Topic Speaker(s) 11:00 am – 12:00 pm Arrival/Coffee All 12:00 – 12:15 pm […]
- [/remote](https://gridfm.org/remote/): ====DAY 1==== Antonello Monti is inviting you to a scheduled Zoom meeting. Topic: GridFM Workshop Aachen – Day 1Time: Sep 8, 2025 08:00 AM Amsterdam, Berlin, Rome, Stockholm, ViennaJoin Zoom Meetinghttps://protect.checkpoint.com/v2/r01/___https://rwth.zoom-x.de/j/63032906278?pwd=PsqFMaQb2nGxAMEaVVAzw76CAZNxTN.1___.YzJ1OnN0b255YnJvb2s6YzpnOmY5OTVmZWIyZjIyODQ2Yzg4MWFjODY1YzkxMDU5YjdmOjc6MjMzMTpmMmRlZDkzNjk3ZTZiNDRiY2MzMmQ2MDVlNmVhZGZhNjVmNmY0ODcyNTBjYTk5YmQ0NDAzY2QwZTRjZGNlOGVhOnA6VDpG Meeting ID: 630 3290 6278Passcode: 700445 — Join by SIP• 63032906278@lej.zmeu.us Join instructionshttps://protect.checkpoint.com/v2/r01/___https://rwth.zoom-x.de/meetings/63032906278/invitations?signature=B-yh45b-w5g0DRfb1WqfZMvBhbIiH98qInDXZ9KFjzA___.YzJ1OnN0b255YnJvb2s6YzpnOmY5OTVmZWIyZjIyODQ2Yzg4MWFjODY1YzkxMDU5YjdmOjc6OTU5Yjo4MzU2MzliZjFiOGJmNzNhZTM5MDU0YzBhYjUxMmY2ZGU1NjM0ODA1ZTNmNjcyNzQ0ZDRiOTgyMzdlNDExZDc4OnA6VDpG ===DAY 2=== Antonello Monti is inviting you to a scheduled Zoom meeting. Topic: GridFM Workshop – Day 2Time: Sep 9, 2025 08:00 AM Amsterdam, Berlin, Rome, Stockholm, ViennaJoin Zoom Meetinghttps://protect.checkpoint.com/v2/r01/___https://rwth.zoom-x.de/j/69760266882?pwd=Eu2tpCV3wVBtqroeb37umGXTiW3cqn.1___.YzJ1OnN0b255YnJvb2s6YzpnOmY5OTVmZWIyZjIyODQ2Yzg4MWFjODY1YzkxMDU5YjdmOjc6N2FmYjplYTAyY2NhNDk2NTc4ZDk0YWM1MDA1YWU4NzczMWIxZmFjNjI5NWM1ZmE5MjQ1ZTA4OTI5NTNmMzM2M2Q5YTU0OnA6VDpG Meeting ID: 697 6026 6882Passcode: 218341 — Join by SIP• 69760266882@lej.zmeu.us Join instructionshttps://protect.checkpoint.com/v2/r01/___https://rwth.zoom-x.de/meetings/69760266882/invitations?signature=Jx0H2i-d3ZH6J1vKj-ZSOzQ20ipWaIUA_Ria7OWvIPs___.YzJ1OnN0b255YnJvb2s6YzpnOmY5OTVmZWIyZjIyODQ2Yzg4MWFjODY1YzkxMDU5YjdmOjc6ZGQzOTowYWQ1ODIxODNmNDNiMGMzMTA3ZTllMjIwYWU2YTJmYTY0Mjg4ZDY0YWEzNWYxYWRhZmIzMTk3NzIxZmY3ODczOnA6VDpG
- [/tutorial/anl](https://gridfm.org/tutorial-anl/): Interactive Tutorial: Training a Small GridFM with APPFL on AWS This tutorial will provide a hands-on introduction to using the Argonne’s open-source privacy-preserving federated learning framework (APPFL) for training a small GridFM model on Amazon Web Services (AWS), enabling model trains without data sharing. Participants will learn how to set up APPFL in a distributed environment, configure federated learning experiments, and monitor training performance in real time. The session is designed to be interactive, with opportunities for participants to follow along, experiment with the code, and gain first-hand experience running APPFL on AWS. This work is supported in part by […]
- [/regulation/general](https://gridfm.org/regulation-general/): Empowering AI in the Electricity Grid: The Role of Public Authorities Goal of this session is to understand how the work of public authorities can actually help in developing initiatives such as GridFM. For this reason, first of all the speakers will present the current portfolio of activities in the different agencies and then an open panel will discuss the way forward.The session is coherently structured in three parts 1) Part 1: Pitch presentation from Prof. Heymann on the challenges related to regulation in AI 2) Part 2: Current activities in the agencies. Each agency will make an overview presentation […]
- [/data/anl](https://gridfm.org/data-anl/): Considerations for Multi-task Data Generation in Grid Foundation Models This talk discusses several practical considerations in the development of datasets for the multi-task training of power grid foundation models. Tasks considered in this talk encompass a range of regression, classification, and optimization tasks including optimal power flow, reliability assessment, state estimation, and unit commitment. The talk focuses on dataset construction, data ingestion and normalization for multi-task training, and issues pertaining to time-scale selection.
- [/usecase/dis](https://gridfm.org/usecase-dis/): Advancing GridFM from Distribution Grid Applications Distribution Grids present interesting challenges for GridFM given the number of nodes involved in the calculations. The recent trend to include low voltage feeders inthe different aspects of grid operation is creating a new level of challenge.At the same time Distribution Grids are probably also the portion of the infrastructure with more issues in terms of data availability.In the discussion we will try to address the following points:1) What are the key use cases for distribution where GridFM can make adifference?2) How do we get the data we need for training?3) How do we […]
- [/transients/rwth](https://gridfm.org/transients-rwth/): Dynamic simulation for grids with high penetration of renewables: the role of shifted frequency analysis As outlined in a recent IEEE report, grid with high penetration of renewables create new challenges for dynamic modeling. The classical separation between electromechanical and electromagnetic transients loses its role and it becomes more difficult to have an adequate description in the time domain without reaching a huge level of computational complexity. Shifted frequency analysis emerged as a possible compromise to have a flexible model able to capture a variety of dynamics in a single representation. The presentation will introduce the modeling approach and its […]
- [/data/dis](https://gridfm.org/data-dis/): Addressing Data Limitations in Power Grid Research with Synthetic Datasets Data-driven methods rely critically on the availability of large, high-quality datasets. While such data is abundant for domains like text and image generation, information on the operation of critical infrastructures is rarely public, and when available it is often aggregated and incomplete. In this talk, we present a method for generating large-scale synthetic datasets of power injections in a transmission grid model of continental Europe. The approach combines structural information on the grid – its line admittances, the location, type, and capacity of generators – with publicly available aggregated load […]
- [/data/potsdam](https://gridfm.org/data-potsdam/): Enhancing Power Grid Resilience to Extreme Weather by Deploying Machine Learning Models Mehrnaz Anvari – Fraunhofer SCAI Societies are undergoing rapid changes in energy generation and consumption. ENTSO-E predicts that electric energy will constitute up to 50% of total energy use by mid-century, up from the current 20%. This increase in electrification will necessitate significant advancements in sectors like transportation and heating. To ensure a resilient society, a robust power system is essential due to emerging dependencies. The sixth IPCC Assessment Report underscores the intensification of weather extremes, such as severe wind conditions, which may exploit vulnerabilities in the power […]
- [/usecase/for](https://gridfm.org/usecase-for/): GridFM for forecasting (power flow, loads, grid state) In this brainstorming discussion, we will focus on how to extend current version of GridFMs for modeling grid dynamics. We will review use cases, data requirements and modeling needs across different time scales to collaboratively build a shared understanding. The goal is to identify and prioritize the development activities needed to advance GridFMs in the near future.
- [/transients/delft](https://gridfm.org/transients-delft/): Learning for Power System Dynamics: The Generalization Challenge Jochen Cremer – TU Delft Fast, accurate evaluation of power system dynamics can enable new use cases for managing the variability of modern grids. However, currently these use cases are not realized as simulations are slow. ML models were studied in the past as fast surrogates for such dynamic simulations. However, to this date, ML models often fail when disturbances fall outside their training data. This talk explores why extrapolation, especially for discrete events, is so challenging, and how transfer learning helps (and where it falls short), pointing toward new paths for […]
- [/usecase/state](https://gridfm.org/usecase-state/): Advancing GridFM from Power Flow to State Estimation To date, GridFM has demonstrated remarkable efficiency in Power Flow (PF) estimation, achieving orders-of-magnitude reductions in computational cost. In this breakout session, we will explore extending GridFM to State Estimation (SE). Leveraging a masked autoencoder approach holds promiss to accurately reconstruct missing grid topology and effectively denoise sensor data even under non-Gaussian noise. We will outline a study to test this capability for both transmission and distribution grids, using realistic parameters under quasi-static and/or dynamic operation scenarios. The discussion will also cover: i) available datasets relevant for SE and ii) architectural enhancements […]
- [/gridfm/eth](https://gridfm.org/gridfm-eth/): The role of physics-informed neural networks in power systems analysis Anna Varbella – ETH Zurich We investigate Physics-Informed Neural Networks (PINNs) as unsupervised tools for solving the optimal power flow in power systems. Starting from the fully supervised PowerGraph benchmark, we systematically explore three approaches: fully supervised, fully unsupervised PINNs, and hybrid methods, examining the advantages and limitations of each. Our analysis reveals that supervised approaches encounter challenges with mixed feasible/infeasible data, while pure unsupervised methods may lack the operational precision required for practical applications. The hybrid PINN framework shows promise as a balanced solution, capable of learning from mixed training […]
- [/gridfm/harvard](https://gridfm.org/gridfm-harvard/): PowerAgent: A Roadmap Towards Agentic Intelligence in Power Systems Qian Zhang – Harvard University Recent advances in Artificial Intelligence (AI)—particularly the emergence of general-purpose AI agents capable of tool use, reasoning, and task orchestration—offer a new direction for enhancing grid flexibility and resiliency. We introduce the concept of the PowerAgent: an AI-enabled, context-aware assistant that leverages foundation models, standardized tool interfaces, and structured workflows to support grid operation and planning decisions. We discuss the conceptual architecture, implementation pathways, and system-level benefits of deploying Power Agents in power grid operations, with an emphasis on augmenting operator capabilities, improving situational awareness, and […]
- [/gridfm/anl](https://gridfm.org/gridfm-anl/): Toward a Unified Graph Learning Framework for Power Grids: Progress on Argonne’s GridFM Research We present recent progress on Argonne’s GridFM, a framework for training heterogeneous graph neural network (GNN) models across multiple power grid applications using a multi-task learning approach. GridFM integrates multimodal datasets from five key grid applications (ACOPF, SCUC, distribution system state estimation, reliability assessment, and transient stability analysis) to enable shared and task-specific representations. We highlight model architecture design, data pipeline development, and preliminary results showing cross-task generalization and scalability. This work aims to build a unified foundation for GNN-based surrogates in power system analysis and […]
- [/data/opensynth](https://gridfm.org/data-opensynth/): OpenSynth – Grid ready synthetic demand data Gus Chadney – Centre for Net Zero Granular demand data is a critical component of grid modelling and a natural partner for GridFM, key to understanding load conditions and modelling different scenarios. We present Faraday, a generative algorithm to produce high-fidelity, synthetic load profiles that capture a wide range of consumer behaviours, and OpenSynth, an open source community to promote and accelerate adoption of this data as well as other AI ready datasets such as RTE’s grid topology dataset. By leveraging this synthetic data, GridFM can be effectively trained to understand complex grid […]
- [/transients/opal-rt](https://gridfm.org/transients-opal-rt/): Accelerating Transient Simulation – Could AI Push While Grid Evolution Pulls? Ravinder Venugopal – OPAL-RT The evolution of the grid in the last decade, with the introduction of widespread inverter-based resources and load increases from electrified transportation in addition to a rapidly growing number of data centers, has driven the need for both a larger number as well as more detailed power-system transient simulations. Grid operators and equipment manufacturers face challenges in completing comprehensive studies while meeting tight project deadlines. Conventional offline simulations, while very accurate, run slowly on currently available computing platforms. Real-time simulation approaches reduce simulation time by […]
- [/gridfm/ibm](https://gridfm.org/gridfm-ibm/): GridFM – Proof of Concept Empowering Diverse Downstream Tasks Etienne Vos – IBM Foundation models (FMs) have proliferated and transformed many industries over the past couple of years. GridFM represents an opportunity to apply the same FM paradigm to the electric grid by leveraging from abundant power system data available, and adapting (fine-tuning) it to various “downstream tasks”. In this presentation we will report on our progress towards developing GridFM, showing generalizability across network topologies and topology perturbations, to serve regression tasks such as power-flow and state-estimation. We will introduce our recent release of the GridFM “Starter-Kit”, consisting of the […]
- [/transients/inesc](https://gridfm.org/transients-inesc/): The Iberian Peninsula Blackout: Opportunities for Foundational Models Francisco Fernandes – INESC TEC This presentation explores the recent blackout event in the Iberian Peninsula as a case study of the growing complexity in operating power systems dominated by renewable energy sources (RES). The event exposed critical limitations in current operational strategies under high-RES penetration, particularly in managing uncertainty and system flexibility. We examine where foundational AI models can offer meaningful support, including in forecasting, system awareness, and real-time decision-making. Special attention is also given to strategies for generating high-quality, diverse datasets required to train such models effectively.
- [/powermarkets/mercuria](https://gridfm.org/powermarkets-mercuria-2/): Energy Commodity Trading: AI for Power Grids and Powering AI Karthik Mukkavilli – Mercuria In this panel presentation I will briefly provide: 1) an overview of some key drivers impacting the landscape of power trading; 2) how AI and big tech companies are impacting power markets; and 3) potential AI vendor opportunities for power trading and more broadly energy commodities.
- [/powermarkets/mercuria](https://gridfm.org/powermarkets-mercuria/): Energy Commodity Trading: AI for Power Grids and Powering AI Karthik Mukkavilli – Mercuria Energy Group In this panel presentation I will briefly provide: 1) an overview of some key drivers impacting the landscape of power trading; 2) how AI and big tech companies are impacting power markets; and 3) potential AI vendor opportunities for power trading and more broadly energy commodities.
- [/regulation/nerc](https://gridfm.org/regulation-nerc/): A NERC perspective on AI regulations  Mark Lauby – NERC While Artificial Intelligence (AI), Machine Learning (ML), and Data Science have been studied and developed for decades, technological growth and public attention have recently ballooned. This has led to a tremendous amount of research and new solutions in the marketplace, affecting nearly every aspect of work and personal life. Innovations and discourse continue to evolve rapidly as excitement grows and investment and research branch into new areas.The BPS is the planet’s most complex machine in transition to the largest computer (a system involving complex humans and complex systems with complex […]
- [/tutorial/ibm](https://gridfm.org/tutorial-ibm/): GridFM starter kit In this workshop, we will introduce and walk attendees through gridfm-datakit and gridfm-graphkit, two Python libraries that consolidate previously scattered state-of-the-art practices in data generation and modeling for power systems. Developed to support the growing use of learning-based methods for tasks such as power flow and contingency analysis, these tools address the lack of standardized datasets and model implementations that have hampered comparability, limited scalability, and confined research to overly simplified grids and settings. Our goal is to establish gridfm-datakit as the standard for model training and benchmarking, and gridfm-graphkit as the foundation for future neural power […]
- [/usecase/expansion](https://gridfm.org/usecase-expansion/): Data Center Interconnection and Grid Expansion Hendrik F. Hamann – SBU/BNL/IBM In this brainstorming discussion we will focus on opportunities of using GridFM ‘s acceleration (in particular for DC power flow) for identifying optimal locations of adding new data centers and/or connecting new generation to expand the electric grid before engaging in time-consuming interconnection studies.
- [/usecase/inesc](https://gridfm.org/usecase-inesc/): The Iberian Peninsula Blackout: Opportunities for Foundational Models This presentation explores the recent blackout event in the Iberian Peninsula as a case study of the growing complexity in operating power systems dominated by renewable energy sources (RES). The event exposed critical limitations in current operational strategies under high-RES penetration, particularly in managing uncertainty and system flexibility. We examine where foundational AI models can offer meaningful support, including in forecasting, system awareness, and real-time decision-making. Special attention is also given to strategies for generating high-quality, diverse datasets required to train such models effectively.
- [/usecase/alliander](https://gridfm.org/usecase-alliander/): Exploring Foundation Models: A use-case driven approach Luc Nies – Alliander As the largest Dutch DSO, Alliander is investing in novel AI techniques be better equipped in dealing with the increased complexity of modern distribution grids. Foundation Model for the energy grid a specifically promising as it has the potential to revolutionize the energy sector by leveraging advanced AI capabilities on extensive datasets. In this presentation we explain our strategy on how to build such a model, one use-case at a time, and show some early results how we apply it to real world problems.
- [/data/ibm](https://gridfm.org/data-ibm/): Can we share models trained on our grid data? Kathrin Grosse – IBM Research The relationship between data volume and accuracy is well-established: increased data availability typically comes with higher performance. However, when data is limited, the practical implementation of data sharing introduces significant complexities about who obtains access to the data. Not publishing data, but a trained model instead, seems to be a way to prevent releasing the data. In this talk, we focus on the different possibilities to extract data from a trained model: for example, confidence values are used in membership inference to conclude whether data was […]
- [/powermarkets/rebase](https://gridfm.org/powermarkets-rebase/): Custom weather and time series visualisations for power markets Sebastian Haglund – rebase.energy Custom weather and time series visualisations for power markets rebase.energy is a digital energy startup based in Stockholm that provides forecasting and visualisation services to energy companies. The Rebase Map (map.rebase.energy) is an open forecasting and visualisation tool that provides situational awareness to power operators and traders. The main novelty of the tool is the ability to combine weather and energy data in a single view.
- [/powermarkets/dtu](https://gridfm.org/powermarkets-dtu/): Learning to Bid and Schedule in One-Price Power Markets Farzaneh Pourahmadi – Technical University of Denmark We propose a learning-based strategy for bidding and scheduling in day-ahead electricity markets with one-price imbalance settlement, focusing on hybrid renewable systems combining wind generation and hydrogen production. In such markets, conventional approaches often result in high-risk, all-or-nothing bidding behavior due to the uncertainty in imbalance prices—a phenomenon we term “betting.” To address this, we develop a data-driven framework that learns linear decision policies from contextual features to jointly optimize electricity bidding and hydrogen scheduling. By incorporating explicit risk constraints, our method transitions from […]
- [/powermarkets/axpo](https://gridfm.org/powermarkets-axpo/): Data science for short-term energy trading Azzurra Casali – Axpo Short-term energy trading faces unique challenges, including the intermittency of renewable energy sources, market volatility, and the need for rapid, data-driven decision-making. In this discussion, we will explore how to address these issues using data science by integrating time series forecasting, machine learning, and optimization techniques into trading strategies. Real-world use cases from diverse markets and geographies will illustrate the practical implementation of these approaches, guiding attendees to gain valuable insights into the intersection of data science and energy trading.
- [/industry/TenneT](https://gridfm.org/industry-tennet/): Foundation Models and Topology Optimization Jan Viebhan – TenneT TSO In this talk, I will explore the potential of foundation models for topology optimization in power grid congestion management.
- [/industry/swissre](https://gridfm.org/industry-swissre/): Optimization at Swissgrid: Current Practices, Challenges, and the Promise of Foundation Models Stavros Karagiannopoulos – Swiss Grid not yet available
- [/industry/hitachi](https://gridfm.org/industry-hitachi/): From Reactive to Proactive: Digital Innovations for Power Grid Intelligence Milos Subasic – Hitachi Energy As energy systems grow more complex, traditional deterministic automation is no longer sufficient. This presentation highlights how Hitachi Energy is advancing power grid operations through AI-driven innovation. Key focus areas include: machine learning-based forecasting that enhances grid stability and asset planning without statistical training; intelligent alarm management that reduces operator overload through event clustering and contextual visualization; and GPU-accelerated grid analytics developed in collaboration with NVIDIA, enabling near-real-time simulations for state estimation and contingency analysis. Together, these innovations empower operators to transition from reactive to […]
- [/gridfm/iea](https://gridfm.org/gridfm-iea/): Insights from the IEA’s Energy and AI report Eren Cam – International Energy Agency The development and uptake of artificial intelligence (AI) has accelerated in recent years – elevating the question of what widespread deployment of the technology will mean for the energy sector. There is no AI without energy – specifically electricity for data centres. At the same time, AI could transform how the energy industry operates if it is adopted at scale. The latest report from the International Energy Agency (IEA) provides analysis based on new global and regional modelling and datasets, as well as extensive consultation with […]
- [/gridfm/hq](https://gridfm.org/gridfm-hq/): Steps towards generative AI for power systems at Hydro-Québec Francois Miralles – Hydro-Québec Hydro-Québec’s network is undergoing significant changes due to the addition of generation and transmission capacity to meet future energy needs of Québec’s province. This leads to ever larger amount of network simulations as well as simulation results analysis needs. This presentation will give an overview of some generative artificial intelligence applications we envision to address the challenges faced by the company. We will also review the ongoing development activities at Hydro-Québec aimed at participating in the GridFM initiative, like data curation and downstream task specifications. This presentation […]
- [/industry/hertz](https://gridfm.org/industry-hertz/): A Gpu-native Approach on Tackling Grid Topology Optimization Christian Merz – 50 Hertz In the short presentation I will present how Elia Groups tackles topology grid optimization using a gpu-native approach. In addition, I will provide an overview on our ongoing research, a short outlook on our vision OptOmni and how we could envision an integration of foundational models into our optimizations.
- [Aachen](https://gridfm.org/aachen/): After three successful workshops on Foundation Models for the Electric Grid at Yorktown Heights in New York, Imperial College in London, and Argonne National Laboratory in Illinois, the GridFM community will get together at its 4th workshop at Aachen, Germany on September 8th and 9th. Please register here before August 31st,2025 For remote attendance please join here. Remember that all times are CEST. Agenda Times Day 1 – Sept. 8th Day 2 – Sept. 9th 8:00 – 8:20 am Coffee (sponsored by IBM) 8:20 – 8:30 am WelcomeAntonello Monti (RWTH Aachen) Day 1 SummaryHendrik Hamann (SBU, BNL) 8:30 -10:40 am […]
- [GridFM - introduction page](https://gridfm.org/): Foundation models (FMs), pre-trained on large datasets and readily adaptable to a broad set of applications, are revolutionizing the field of artificial intelligence (AI). Powerful FMs for language and weather have recently emerged, proving that such models can be developed for complex systems. The GridFM project pioneers the concept of FMs for the electric power grid to be trained on grid data – as opposed to text data – with the overarching goal to develop the underlying technology to cope with the increasing complexity and uncertainties of a faster growing grid (e.g., due to hyperscalar data centers, crypto mining etc.). […]


[comment]: # (Generated by Hostinger Tools Plugin)