Neural Network-based Performance Prediction of Heat Pipe and Loop Heat Pipe Heat Exchangers: A State-of the- Art Review

As the demand for compact and high-efficiency thermal management systems intensifies, Heat Pipes (HPs) and Loop Heat Pipes (LHPs) have emerged as vital solutions due to their passive operation and superior heat transfer characteristics. However, accurately predicting their performance across varying operating conditions remains a formidable challenge. This review explores the transformative role of Artificial Neural Networks (ANNs) in modeling these complex systems, offering a data-driven alternative to conventional empirical and numerical methods. It critically evaluates diverse ANN architectures, training strategies, and hybrid approaches- including physics- informed models- that bridge the gap between physical laws and predictive analytics. The findings reveal that ANNs not only enhance prediction accuracy but also significantly reduce computational burden, making them ideal for real-time optimization in next-generation thermal systems. This study also outlines current challenges and paves the way for future research that combines domain knowledge with AI to revolutionize thermal engineering design and control. Major Findings: This study shows that Artificial Neural Networks (ANNs) effectively predict the nonlinear performance of heat pipes and loop heat pipes with higher accuracy than traditional methods. Integrating physics-informed and hybrid models enhances prediction reliability and physical consistency. ANN approaches also reduce experimental and computational costs, enabling real-time thermal system optimization.