Entropy analysis of mixed-mode flow in lid-driven corner-cut cavities filled with LTNE porous media using a hybrid CBS-FEM and machine learning approach

This study investigates entropy generation and mixed-mode heat transfer characteristics in corner-cut enclosures filled with a dual-temperature (LTNE) porous medium and subjected to heterogeneous inner heating. A novel hybrid framework, integrating the Characteristic-Based Split Finite Element Method (CBS-FEM) with machine learning techniques, is developed to analyze complex thermal behaviors and enhance heat transfer efficiency. The CBS-FEM accurately captures coupled convection–conduction mechanisms and nonlinear temperature fields, while machine learning models are employed to predict entropy generation trends and improve computational accuracy. The effects of geometrical modifications, heating intensity variations, and dual-temperature medium properties on entropy production are systematically analyzed. Results reveal that velocity components exhibit oscillatory patterns under heterogeneous periodic heating, while uniform and linear heating cases show a steady monotonic behavior. The flow field becomes non-uniform under oscillatory temperature conditions, whereas symmetrical patterns emerge for uniform and linear heating modes. These findings provide valuable insights into thermal management and entropy minimization strategies, with potential applications in building physics, such as optimizing indoor thermal comfort, energy-efficient heating, ventilation, and cooling (HVAC) systems, and sustainable building envelope design. Overall, the study contributes to the development of energy-efficient thermal systems in advanced engineering applications.