Modular Energy Hub Modeling Framework

Problem and Research Question

Modern energy systems do not deliver a single commodity. A district or industrial site may simultaneously move electricity, heat, cooling, and gas through converters, storage, and networks, and the optimal way to operate such a multi-energy system depends on how those carriers interact. The central question: how can the optimal operation of a multi-energy system be formulated and solved as a unified optimization problem, using a modular modeling framework that separates the system topology from the optimization solver?

Hand-built models tie a specific system configuration to a specific solver formulation, so every new topology requires rewriting the optimization code. The gap this work addressed was a reusable abstraction: describe the topology once, then let a generic solver interface optimize it.

Method

This project produced an open-source optimization framework for modular energy hub modeling, developed during Google Summer of Code 2016 at Empa. The framework combines:

• Energy hub abstraction — represents multi-energy systems as hubs that convert, store, and distribute multiple energy carriers (electricity, heat, cooling, gas) through a unified mathematical formulation.
• Modular architecture — separates the system topology description (which converters and storages are connected, their efficiencies and capacities) from the optimization problem formulation, so different system configurations can be modeled without rewriting solver code.
• Python and Pyomo — implemented in Python using the Pyomo optimization modeling language, enabling access to multiple solvers (linear programming, mixed-integer programming, nonlinear programming) through a consistent interface.
• Open-source release — released as open-source software under the GSoC program, with the project selected as an exemplary GSoC project by the program organizers.

The modular energy hub approach is foundational to multi-energy system optimization: rather than building a custom model for each system configuration, the framework provides a reusable abstraction that can represent district heating networks, building energy systems, industrial cogeneration plants, and community microgrids within a single modeling paradigm.

Personal Contribution

Frederik Banis was the sole developer of the framework during Google Summer of Code 2016 at Empa. He designed and implemented the modular energy hub abstraction, the Pyomo-based optimization interface, and the example system models. The project was supervised by Empa researchers and was selected as an exemplary GSoC project.

Validation and Key Results

• The framework successfully modeled and optimized representative multi-energy system configurations, demonstrating that the modular abstraction can represent different system topologies without code changes.
• The open-source release was listed as an exemplary project by Google Summer of Code, indicating recognition of the framework’s quality and documentation.
• The modular energy hub approach aligns with the broader energy hub literature (Geidl et al., ETH Zurich), extending the methodology into an accessible, reusable software implementation.

Related Outputs

• Open-source software: Google Summer of Code 2016, Empa. GSoC Archive
• Publication catalogue entry: data/publications.yaml, id 10.
• Related project: the framework is the basis of the Modular Energy Hub Framework project page.

Collaborators and Institutions

• Empa, Swiss Federal Laboratories for Materials Science and Technology — supervision and hosting of the GSoC project.

Status and Next Steps

Status: Historical (2016). The framework is no longer actively maintained but remains available as open-source reference software.

Transfer to current research: The modular energy hub framework represents an early application of the optimization-under-uncertainty methodology that later evolved into model predictive control for microgrids and, ultimately, adaptive decision-making for biomedical sensing. The core insight — separate the system model from the decision algorithm — is a recurring theme across the research program: from energy hub optimization to MPC for microgrids to adaptive experiment selection for closed-loop discovery.

See the related project page for the broader research context, the Methods page for the optimization and decision-making methodology, and the Research page for the research program overview.