AI-assisted numerical optimization of latent heat thermal batteries in hexagonal arrays with petal-shaped metal-foam layers

Shafi, Jana; Younis, Obai; Saif, Abdullah Mohsen; Chauhan, Bhupendra Singh; Fteiti, Mehdi; Ghalambaz, Mohammad

doi:10.22055/jacm.2026.49715.6186

AI-assisted numerical optimization of latent heat thermal batteries in hexagonal arrays with petal-shaped metal-foam layers

Articles in Press

Document Type : Research Paper

Authors

Jana Shafi ¹

Obai Younis ²

Abdullah Mohsen Saif ¹

Bhupendra Singh Chauhan ³

Mehdi Fteiti ⁴

Mohammad Ghalambaz ⁵

¹ Department of Computer Engineering and Information, College of Engineering in Wadi Alddawasir, Prince Sattam Bin Abdulaziz University, Saudi Arabia

² Department of Mechanical Engineering, College of Engineering in Wadi Alddawasir, Prince Sattam Bin Abdulaziz University, Saudi Arabia

³ Department of Mechanical Engineering, GLA University, Mathura, UP, India

⁴ Physics department, Faculty of Science of Bizerte, Carthage university Tunisia

⁵ Department of Mechanical Engineering, Dezful Branch, Islamic Azad University, Dezful, Iran.

10.22055/jacm.2026.49715.6186

Abstract

A modular latent heat thermal energy storage (LHTES) system is investigated to overcome the slow thermal response of phase-change materials (PCMs) while retaining a large PCM fraction and limiting obstruction of natural convection. The system comprises densely packed hexagonal modules, each containing a central heat pipe and a partial petal-shaped open-cell metal-foam layer. The geometric study shows that the petal number, amplitude, orientation, and center offsets jointly govern the conductive pathways and buoyancy-driven circulation. Relative to the circular baseline, which requires 126.4 min to reach a melt volume fraction (MVF) of 0.9 and 212.4 min to reach MVF = 0.1 during solidification, the best charging configuration reaches MVF = 0.9 in 69.4 min, reducing the melting time by 45.1%. The best discharging configuration reaches MVF = 0.1 in 133.7 min, reducing the solidification time by 37.0%. To enable rapid evaluation of coupled geometric effects, a dataset comprising 223,027 time-dependent records was generated using many numerical simulations and used to train a multilayer-perceptron surrogate. Independent test results yield a mean absolute error of 0.0017903 and a mean squared error of 5.42429 × 10⁻⁶. The thermal-design comparison supports renewable-heat storage, an application relevant to Sustainable Development Goal (SDG) 7 (Affordable and Clean Energy) and SDG 13 (Climate Action).

Keywords

latent heat thermal energy storage

phase-change material

metal foam

SDG 7

SDG 13

Subjects