Artificial neural network (ANN)-based multi-objective optimization of the vapor chamber with liquid supply layer for high heat flux applications

• Published in: International communications in heat and mass transfer 2024-12, Vol.159, p.108302, Article 108302
• Main Authors: Bang, Soosik, Kim, Seungwoo, Ki, Seokkan, Seo, Junyong, Kim, Jaechoon, Lee, Bong Jae, Nam, Youngsuk
• Format: Article
• Language: English
• Subjects: Artificial neural network; Heat spreaders; Optimization; Vapor chamber
• ISSN: 0735-1933
• DOI: 10.1016/j.icheatmasstransfer.2024.108302

We developed a multi-objective optimization process using artificial neural networks (ANN) to estimate and enhance the thermal-hydraulic performance of vapor chambers (VCs). A numerical model was employed to evaluate the impact of various components on VC performance, and the resulting data were used to train the ANN model. This approach led to an optimized VC design that significantly reduced thermal resistance while substantially improving critical heat flux (CHF). This improvement was primarily due to the distinct roles of the liquid supply layer (LSL) and the evaporator wick. The optimized VC with liquid supply layer (VC-LSL) exhibited a thermal resistance 1/3 lower and a junction temperature 100 °C lower than those of the optimized VC without liquid supply layer (VC-NL) at heat flux of 500 W/cm2. This work demonstrates significant potential for maximizing heat transfer performance by establishing an optimal VC design adaptable to a wide range of heat fluxes. •We introduced the ANN-MOGA process for estimate the VC performance.•A 3D numerical model was developed to predict the VC's thermal performance.•The ANN model has been created that is capable of directly predicting the thermal resistance of the VC and the CHF.•Using the ANN-MOGA process, we derived optimized design variables that reduce thermal resistance for a range of CHFs.