At 2024 LF Energy Summit in Brussels, Christoph Schimeczek from the Deutsches Zentrum für Luft- und Raumfahrt (German Aerospace Center) presented FAME (Framework for Agent-Based Modelling of Energy Systems), a powerful open source tool designed to streamline and simplify the process of modeling complex energy systems. Key insights from the presentation follow (full video is at the end):
The Challenge in Energy System Modelling
Schimeczek began by describing the common pain points faced by researchers and developers working on energy system models. Although the focus is typically on agent logic and the dynamics of energy systems, much of the work often shifts to auxiliary coding tasks. These tasks include managing data inputs and outputs, configuring models, scheduling agent interactions, and optimizing performance through parallel execution. These challenges distract from the core modeling work and add considerable complexity to the development process.
Enter FAME
FAME was developed to eliminate these distractions, enabling developers to focus solely on the logic of their agents while automating and streamlining the rest. Schimeczek highlighted several key features of FAME:
- Data Management and Inter-Agent Communication: FAME organizes how data is exchanged between agents, schedules their interactions, and manages both input and output, reducing the need for custom scripts to handle these processes.
- Parallel Execution and Scalability: The framework supports parallel execution, allowing simulations to run faster, whether on a local laptop or a high-performance computing cloud. This scalability makes it accessible for both small-scale and large-scale applications.
- Full Lifecycle Support: FAME handles everything from model development to application, making the entire process—from configuring models to running simulations—efficient and straightforward.
- Open Source and Flexible: Licensed under Apache 2.0, FAME is fully open source, with flexibility in the types of inputs and outputs used. Developers can configure the framework without needing to modify the code itself.
Workflow and Key Features
The typical workflow for FAME users starts with creating a schema file that defines the model. Based on this schema, data is validated, binary input files are generated, and the simulation is executed. FAME automatically tracks metadata at every stage, ensuring traceability for all aspects of the simulation, from input data to configuration versions. This comprehensive tracking enables better reproducibility and transparency in research.
Real-World Application: AMIRIS
One of the standout examples of FAME’s capabilities is its integration with AMIRIS, an open source model for simulating electricity markets. AMIRIS can model business behaviors, power generation, and flexibility options with high temporal and spatial resolution. Schimeczek demonstrated how the framework allowed a full year’s worth of data simulation to be completed in just 10 seconds using a single computation core, underscoring FAME’s efficiency.
