Detailed Description of the Work Programme

The work programme for the Oproda project is structured to tackle the complexities and challenges associated with the modeling and optimization of Solid Oxide Cell (SOC) systems. It encompasses a comprehensive approach, combining theoretical modeling, experimental validation, and the application of cutting-edge data-driven methodologies. The programme is divided into several key areas:

Explainable and Probabilistic Data-Driven Modelling of SOC Systems This segment focuses on overcoming the “curse of dimensionality” that plagues high-dimensional data modeling. It aims to develop probabilistic models that can provide insights into the operation and degradation mechanisms of SOC systems. The methodologies employed include Gaussian processes, variational Bayes methods, and equation discovery methods. These approaches will allow for the embedding of prior expert knowledge into models, enhancing their explanatory power and reliability.

Addressing Limited or Partial Training Datasets One of the significant hurdles in modeling SOC systems is the often limited or partial nature of available datasets. The work programme proposes to overcome this by integrating prior knowledge into the modeling process and employing semi-supervised learning techniques. This approach will help in dealing with scenarios where certain operational modes or faults have not been extensively observed or are completely new.

Experimental Validation To ensure the models developed are grounded in reality, extensive experimental validation is planned. The project benefits from access to multiple test beds capable of simulating a wide range of operating conditions and faults. This infrastructure allows for the generation of comprehensive datasets, which are crucial for the training and validation of the developed models. The experiments are designed to cover nominal operating conditions, as well as accelerated degradation tests for both SOFC (Solid Oxide Fuel Cells) and SOEC (Solid Oxide Electrolysis Cells).

Work Packages

The work programme is organized into five work packages (WPs), each focusing on a distinct aspect of the project:

WP1 - Requirement Analysis: Identifies and reviews operating conditions, faults, and experimental requirements. It aims to understand the dominant degradation mechanisms and the mathematical models currently used to describe them.

WP2 - Probabilistic Data-Driven Models: Develops the methodologies for creating the models, focusing on variational Bayes methods, Gaussian process modeling, and equation discovery.

WP3 - Long-Term and Accelerated Degradation Tests: Generates necessary datasets through experimental work, covering a range of conditions and fault modes.

WP4 - Validation of Developed Models: Validates the accuracy and reliability of the models under various operating conditions and fault scenarios.

WP5 - Dissemination and Outreach: Focuses on sharing the results and insights gained from the project through scientific publications, workshops, and web portals.