A detailed review of pulsating heat pipe correlations and recent advances using Artificial Neural Network for improved performance prediction

The unique thermal properties and flexible design of pulsating heat pipes (PHP) offer opportunities for relatively lightweight, low-cost, and reliable phase-change thermal solutions. Nevertheless, the perfor-mance of PHP is affected by multiple factors making mathematical predictions of their performance difficult. So, costly experiments in restricted test environments and time-consuming numerical analy-sis are typical methods for detecting internal thermo-hydrodynamics and data acquisition. Since theo-retical models are entirely data-driven that require substantial data for validation, shortfalls in available data affect their prediction accuracy. This trend reduces the accuracy of semi-empirical correlations ( SEC ) and other mathematical models obtained through Regression Correlation Analysis, the Buckingham The-orem, and Intelligent predictions. In this review, the major developments and shortcomings of the SEC between 2003 and 2022 were reported. The opportunities for improvement have been discussed exhaus-tively. Moreover, the recent advances in Artificial Neural Networks (ANN) for PHP performance prediction have been adequately reviewed. Since ANN models are based on black-box analyses, with few physical explanations of heat transfer phenomena, this review suggests a potential coupling between ANN models and SEC . By inferring real phenomena from the dimensionless numbers in SEC , faster, more accurate, and holistic PHP thermo-hydrodynamics can be attained.(c) 2023 Elsevier Ltd. All rights reserved.