LF Energy Summit Europe 2026

The EV charging industry is undergoing a massive transformation, mirroring the broader tech shift toward Software-Defined Vehicles (SDV) and Software-Defined Networking (SDN). Welcome to the era of Software-Defined Charging (SDC).
Historically, charging stations were built using rigid hardware blocks, requiring separate microcontrollers for every new functional domain: from complex protocols (ISO 15118) to load balancing and UI. This fragmented, bare-metal architecture slows innovation, drives up BOM costs, and makes feature updates nearly impossible.

SDC changes the paradigm by decoupling hardware from software. By consolidating high-level logic onto a centralized computing unit, manufacturers can drastically reduce complexity and completely transform the product lifecycle. Instead of deploying static hardware that quickly ages out, SDC enables a continuous software lifecycle via over-the-air updates, keeping field assets relevant for years.

In this talk, we will explore how the open-source EVerest project acts as the operating engine for SDC, enabling the hardware ecosystem to mix-and-match physical components, build flexible satellite systems, and utilize digital twin testing.

As buildings become active participants in the energy transition, understanding flexibility at the building and occupant level is becoming increasingly important for addressing grid congestion challenges. While many energy system models rely on simplified representations, detailed grid-integrated building models can provide deeper insight into how HVAC systems, EV charging, distributed energy resources, and occupant behavior interact with the grid in practice.

This talk presents CityLearn as an open-source platform for studying coordinated energy flexibility strategies across buildings and districts. Recently integrated into the LF Energy community, CityLearn enables researchers and practitioners to explore how building dynamics and occupant-centric control approaches can reveal flexibility potential often overlooked in simplified representations. The session will discuss advanced control strategies for HVAC systems, electrical storage, EV charging, and occupant comfort, together with the importance of connecting detailed building intelligence with broader power grid modeling frameworks to better capture realistic building-grid interactions.

GEISA, the Grid Edge Interoperability and Security Alliance, is an effort within LF Energy to enable application interoperability in edge devices like smart meters and distribution automation devices. To ensure implementations comply with the specification, a dedicated conformance application has been developed to validate that the specification is correctly implemented on target hardware. In this talk, Kévin will present the role of the conformance application as well as the benefits to use and extend it.

We ran a comparative AI audit of the microgrid standards corpus — IEEE 2030.5, IEC 61850, CIM, OpenADR 3, ISO 15118, IEEE 1547, SunSpec — using Claude, Mistral, and a locally hosted model with identical prompts and metrics. This lightning talk shows the convergent findings: which clauses already cover AI-agent operation on DER, where the gaps are, and the v0 interface contract (GAIFARE) we drafted in the ORES repo to fill them — capability declaration, bounded authority, decision auditability, data governance. Prompts, metrics, and gap map are public.

Home Energy Management Systems (HEMS) increasingly operate in non-stationary conditions: changing tariffs, evolving household behavior, new DER assets, and fragmented device ecosystems. Most current deployments still rely on static rules and periodically retrained models, which degrade under drift and are hard to maintain at scale.

This session presents a self-evolving HEMS architecture that closes the loop across three core tasks:
(1) classification for safety/reliability monitoring,
(2) regression for PV/load forecasting, and
(3) optimization that converts natural-language user goals into executable schedules.

An LLM-centered orchestration layer coordinates rule engines, ML models, user/context memory, and control APIs to support transparent interaction, adaptive decision-making, and continuous improvement. We share field-oriented case studies showing improved robustness in detection, better forecasting performance, and practical end-to-end scheduling execution.

The talk focuses on reusable architectural patterns and deployment lessons relevant to open, software-defined energy systems.

On 28 April 2025, continental Spain and Portugal experienced a total blackout, while the rest of the European grid was largely unaffected. This most severe European blackout in over 20 years is analysed using a time-domain model of the Iberian system derived from an anonymised network snapshot. In its final report on the incident, ENTSO-E’s Expert Panel identifies LF Energy’s Dynaωo, an open-source, industrial-grade suite of simulation tools, as the basis for its independent voltage stability study. Simulations closely reproduce recorded measurements and clarify the roles of cross-border exchanges, TSO-DSO interactions, distributed renewable energy sources and limited reactive power margins in the voltage rise that triggered the cascade. The study highlights the value of relying on an open-source industrial tool and shows how Dynaωo has evolved into a robust, high-performance solution. The presentation will detail the modelling and simulation approach, key challenges, and the main insights gained on the 28 April 2025 events, demonstrating how Dynaωo delivers the level of transparency, reproducibility and scientific rigour now expected for the analysis of major power system events.

DyCoV (Dynamic Compliance Verification) is an Open-Source tool built by RTE for bringing transparency to the process of verifying grid-code compliance of the dynamic behavior of new generation facilities. It contemplates both *electric performance* (i.e., "does the behavior pass the requirements?") and *generic model validation* (i.e., "does the model match the actual behavior?").

The tool builds upon Dynawo, RTE's Open-Source dynamic simulator, which in turn uses Open Modelica for its RMS models, including WECC and IEC generic models. This provides a fully transparent process for all stakeholders, thus avoiding the pitfalls of proprietary models and black-box tools. In addition, DyCoV removes the drudgery of measuring, comparing, and extracting all required KPIs from curves: rise/settling times, FRT responses, point-wise diffs, etc.

The tool is built with Python and structured as a series of independent tests, each producing its own report in PDF. The current tests correspond to RTE's connection grid code (DTR document), but DyCoV design uses templating (Jinja) and configuration files to make it easily adaptable to other grid codes, avoiding code changes as much as possible.

Modern AI for power flow and OPF is on the rise, but it will still require large amounts of trustworthy training data and strong conventional solvers for validation (“solver‑certified AI”). An important application are adjacent‑state studies needed for N‑1 screening and decision planning. This session argues that classic algorithms will not vanish, but must be redesigned for batch computation: solving many related grid states faster than repeated Newton-style runs. We reshape ideas from perturbation theory into a load-flow solver pattern that reuses structure across scenarios and maps well to parallelizable hardware. We introduce the key mathematics behind the approaches including the holomorphic embedding load flow method (HELM) and related methods. We then show how to turn them into an engineering workflow that can outperform repeated Newton-Raphson solves in high-throughput studies. We close with an open roadmap: an industrial‑grade implementation under development at 50Hertz/Elia Group and a clear invitation to collaborate on open interfaces, benchmarks, and integration with LF Energy tooling.

Power Grid Model is a high-performance, open-source calculation library built for advanced distribution system analysis.

This joint presentation by Alliander and SOPTIM tells a story of open source collaboration and innovation between a Grid Operator and Vendor within the LF Energy Power Grid Model (PGM) project.

Alliander—maintainer of the Power Grid Model project—and SOPTIM—a key contributor—have met each other at a LF Energy summit, which has led to a productive partnership, including shared development efforts, knowledge exchange, and the introduction of new features that enhance the capabilities of the model.

Overall, the presentation provides a tangible example of how open collaboration within LF Energy can drive innovation, strengthen ecosystems, and accelerate the development of open-source solutions for the energy sector, while also diving into the technical part of new features that support the various use cases at Alliander and SOPTIM.

Grid operations are becoming fundamentally more complex. The classical DC load-flow worked well in the past, but its simplifying assumptions are increasingly violated in modern grids with high renewables, reactive power flows, and tighter operational margins.

DC+ is a voltage-sensitive linearization of the AC load flow that retains voltage magnitude, angles, and reactive power while remaining orders of magnitude faster than full AC solvers.
Conceptually, it corresponds to a single Newton step around the AC base case, capturing the direction and severity of violations far more accurately than classical DC, while scaling to millions of N‑1 evaluations per second on GPUs.

DC+ enables a new class of scalable, open optimization workflows:
Potential use cases of DC+ are:
- Fast contingency screening and ranking?
- Topology optimization with voltage awareness?
- Security constrained voltage Optimization?