Experimentally validated hybrid CFD-machine learning framework for predicting hydrogen evolution reaction performance of Ba0.5Ca0.5CoO3 perovskite electrocatalyst

Elham Fattahi, Dina I. Bakranova, Elham Mahmoudi, Amin Sokhansanj, Nurlan B. Bakranov, Аli Choruh, Nagihan Çaylak Delibaş, Ergün Nart, Elnaz Asghari, Aligholi Niaei

Results in Engineering, 2026

Full text Open access: Gold Metadata (CSL-JSON)
DOI 10.1016/j.rineng.2026.113370
Data confidence 85 high
How this score is calculated

It is built from checkable properties of the record itself: whether the article has a DOI, how many independent sources confirmed its data, whether the required fields are filled in, and how recent the latest confirmation is. Disagreement between sources and a missing identifier lower it. A value of 80 or above means the record was published without further checking; 60 to 79 means it is published at reduced confidence, with a manual check advisable. A record verified by an editor by hand is never pushed below 90 by a recalculation. This scores our data about the article, not the article itself.

Abstract

Accurate prediction of hydrogen evolution reaction (HER) performance under varying operating conditions conventionally requires extensive electrochemical measurements or computationally intensive simulations. To address this challenge, an experimentally validated framework integrating computational fluid dynamics (CFD) and machine learning (ML) was developed for mechanistic analysis and HER performance prediction of Ba 0.5 Ca 0.5 CoO 3 perovskite electrocatalyst. A 3D multiphysics CFD model was established by coupling fluid flow, electrochemical kinetics, and ionic transport. The simulated polarization curves showed good agreement with experimental measurements across a wide range of temperatures (25-65°C) and electrolyte concentrations (1-5 M KOH). The validated CFD model was subsequently used to generate a comprehensive dataset for training three ML models, namely Gaussian Process Regression (GPR), Artificial Neural Network (ANN), and Ensemble Learning. All developed models demonstrated excellent predictive capability, achieving R 2 > 0.985, with GPR slightly outperforming the other models by showing a narrower error distribution. The GPR model was further employed to predict HER current density at previously unseen operating conditions. It exhibited strong agreement with CFD predictions under the sampled operating conditions (R 2 > 0.998), demonstrating robust interpolation capability and its potential as a computationally efficient surrogate for CFD simulations. Furthermore, permutation feature importance (PFI) analysis identified applied potential as the most influential operating parameter for predicting HER current density. The proposed hybrid CFD-ML framework enables rapid and computationally efficient prediction of the experimentally anchored CFD response within the investigated operating domain.