Numerical investigation of micro solid oxide fuel cell performance in combination with artificial intelligence...

Area of Science:

• Energy Conversion and Storage
• Computational Materials Science
• Electrochemistry

Background:

• Proton-conducting solid oxide fuel cells (H-SOFCs) offer efficient energy conversion.
• Direct internal reforming (DIR) of methane is a key process for H-SOFC operation.
• Accurate performance prediction is crucial for H-SOFC optimization.

Purpose of the Study:

• To develop a multiphysics numerical model for an anode-supported H-SOFC with DIR of methane.
• To utilize artificial intelligence (AI) models for predicting H-SOFC performance.
• To investigate the impact of the air-to-fuel (A/F) ratio on H-SOFC performance.

Main Methods:

• A multiphysics numerical model solving coupled nonlinear equations (continuity, momentum, mass transfer, chemical/electrochemical reactions, energy).
• Application of K-nearest neighbour (KNN) and artificial neural network (ANN) AI models to numerical simulation results.
• Parametric study varying the air-to-fuel (A/F) ratio.

Main Results:

• Increasing the A/F ratio was found to decrease current density and overall cell power.
• Setting the A/F ratio to 0.5 resulted in a 7% increase in current density and a 2% increase in power density compared to A/F=1.
• The ANN model demonstrated high accuracy, with an error rate <1% and R-score ~99%, showing excellent agreement with numerical results.

Conclusions:

• The developed multiphysics model accurately captures H-SOFC behavior.
• Optimizing the A/F ratio is critical for enhancing H-SOFC performance.
• AI models, particularly ANN, are effective tools for predicting and optimizing H-SOFC performance.