Digital Twin for Operational Optimization of Electrolysis and Fuel Cell Systems

Electrolysis and fuel cell systems, particularly those based on the SOEC/SOFC solid oxide technology developed at IKTS, can make a significant contribution to national and international emissions targets. In order to achieve this, however, infrastructure and process chains require significant adaptation. In this context, modeling can be used to gain deep insights into the underlying mechanisms in a way that is time- and cost-efficient.

One key factor in terms of the energy transition towards a sustainable energy and raw materials system is the transient behavior of energy sources and sinks within the grid. For this reason, researchers at Fraunhofer IKTS developed a methodology that enables process models to be designed and parameterized on the basis of even a very limited data set derived from technical documentation and experimental data.

Area of application

A digital twin of a high-temperature electrolysis demonstration plant was created and validated in an open-source environment. It includes all fundamental components of the system (stack module, heat exchanger, electrical heaters, afterburner). Thermal capacities were derived from technical documentation and materials properties, while additional losses were identified and parameterized based on a comparison with plant measurement data. The model was validated by means of a dedicated measurement campaign. Using a training data set of 150 hours, a mean deviation of the stack temperature of less than 0.6 K was achieved.

Model structure and data basis

The digital twin is based on OpenModelica with the Modelica Standard Libraries and specially developed, component-specific submodels. Each submodel represents one component of the overall system and, in addition to its specific function, includes both a thermal capacity and thermal losses (convection, radiation). Thermal capacities were calculated from CAD models of the components and the associated material properties (density, specific heat capacity). Additional thermal losses were adjusted based on model calibration using steady-state and transient plant measurement data (temperatures, power consumption, mass flow rates). Model validation was carried out by comparing heat-up curves, load step responses, and steady-state operating points, with iterative fine-tuning to minimize deviations.

Providing a valid, dynamic representation of the real system with minimal parameterization effort, the OpenModelica-based digital twin contributed significantly to experimental planning and operational optimization.

Services offered
 

• Modeling at component and system level
• Support for controller development and commissioning
• System analysis for optimization and training purposes