Fuel cell electric vehicles are described on cell, stack and system levels. In driving operation, multi-physics coupling across subsystems (reactant supply, humidification, thermal management, etc.) reshapes cell- and stack-level boundary conditions, impacting performance and degradation mechanisms. Isolated single-topic approaches on one specific level may have limited transferability, as cross-level interdependencies under changing operating conditions can negate improvements or shift limiting factors. This underscores the development of validation environments (VEs) that represent cross-level interactions and evolve as experimental evidence redirects research questions.
Models such as the V-Model provide phase-oriented logic for developing VEs when validation scope, boundary conditions and acceptance criteria can be specified upfront and remain stable. However, in PEMFC VE development, experimental conclusions frequently reshape hypotheses, operating conditions and research topics across successive cycles. Consequently, existing approaches often provide limited methodological support for a traceable and repeatable evolution of VEs where iterative reconfiguration is essential.
To address this need, we developed the Development Model of the Validation Environment (eMVU, German for Entwicklungsmodell der Validierungsumgebung) for PEMFC technology to enable a structured, model-based and iterative evolution of VEs. Embedded in the system triple of product engineering, the eMVU guides the iterative transformation of objectives into validation configurations (VCs) through model-based derivation of boundary conditions, test requirements and extension measures. It structures each development cycle into the five phases design, specification, implementation and commissioning, experiments and results processing, as well as derivation of measures with feedback of the resulting insights into the objectives of the subsequent cycle.
The framework is demonstrated by realizing a fully functional baseline VE and deriving two additional VCs across eMVU cycles. Their implementation and operation provide experimental evidence that both the eMVU and the resulting VE enable traceable, repeatable and targeted assessment of cross-level interdependencies with measurable impact on cell and stack behavior.