Real-time capable transient model of liquid water dynamics in proton exchange membrane Fuel Cells
date: 06.07.2022
category: Sporočila za javnost
Researchers of Laboratory of internal combustion engines and electromobility (LICeM) have in collaboration with Faculty of Electrical Engineering, Mechanical Engineering and Naval Architecture in Split (FESB), University of Split developed and validated the Real-time capable mechanistically based transient model of liquid water dynamics in proton exchange membrane Fuel Cells. Due to these characteristics, the model can be utilized in advanced control methodologies and hardware-in-the-loop as well as digital twin applications. The results were published in a renowned journal - Journal of Power Sources (IF: 9.127).
Optimal control of liquid water dynamics plays an instrumental role in achieving optimised performance and prolonged lifetime of PEM Fuel Cells (PEMFC). Tackling these challenges calls for precise on-line monitoring and control tools such as coupled virtual observers taking into consideration also liquid water dynamics. The latter proves to be especially challenging to model due to varying retention and removal rates of liquid and gaseous water depending on the operating conditions thus representing a longstanding knowledge gap on the system level. To fill this gap, researchers have derived a 1D+1D system level mechanistically based PEMFC model composing two-phase flow water sub-model, which enables consistent system level treatment of liquid water dynamics in all seven most influential regions of the PEMFC (see Figure below), namely membrane, anode, and cathode channels, GDLs, and catalyst layers, while exhibiting real-time readiness with real-time factor of 0.0449 at the time step of 1ms.
Figure 1: Schematic representation of the seven regions modelling approach.
The model is extensively tested on single-cell data, which consists of five sets of experiments with different operating regimes and durations. Overall results exhibit good agreement with experimental data in all of the performed tests with R² factors larger than 0.95. It can thus be concluded that the developed framework, which consistently treats liquid water dynamics in all domains of the FC and thus realistically models retention and removal rates of liquid and gaseous water, successfully fills the knowledge gap in the area of real-time capable transient models of liquid water dynamics applied in FC models. It, therefore, makes possible development of advanced control methodologies and hardware-in-the-loop as well as digital twin applications.
Figure 2: Results of the calibrated model on the experimental data obtained under: (a) slow transients with ramping up-and-down power demand with R2 = 0.9737 and (b) combination of fast and slow transients R2 = 0.9552.
Link to the paper: https://doi.org/10.1016/j.jpowsour.2022.231598
