Solid oxide electrolysis cell 3D simulation using artificial neural network for cathodic process description

An artificial neural network (ANN) is used for modeling electrochemical process in a porous cathode of SOEC. The neural network has the following input parameters: the overvoltage, the hydrogen and steam composition at electrode/electrolyte interface. Data for training and validating the ANN simulator is extracted from a validated model. Once the model is identified, the ANN can be successfully used for simulating electrochemical behavior of a SOEC electrode. The analytical expression of the network has been implemented in a three-dimensional multiphysics model of SOEC serial repeat unit (SRU). The expression takes into account micro-scale effects in the macro-scale model with a minimum cost of computation time. Gas flow velocity, species concentrations, current density and temperature distributions through the SRU have been calculated. It has been shown that the ANN could be used in the macro-scale model giving coherent results.

► 3D simulation of high-temperature electrolysis cell using a commercial software. ► The artificial neural network can be successfully used for simulating electrochemical behavior of a SOEC electrode. ► The network analytical expression can be used as a boundary condition in a 3D multiphysics macro-scale model in order to reduce the time computing and freedom degrees.