Machine-Learning Assisted Flapping Agitator Design Methodology Towards Enhanced Thermal-Hydraulic Performance

University Of Missouri-Columbia, Columbia MO

Investigators

Abstract

Thermal transport in air-cooled heat exchangers is critical for effective cooling of power plants, data centers, and electronic devices. However, the performance of an air-cooled heat exchanger is often restricted by insufficient mixing of hot and cold air, which is needed to reduce the required fan power. Over the past few decades, much effort has been devoted to investigating enhanced heat transfer by introducing additional air turbulence. Among all the methods, flow-induced vibration of a flexible thin-film agitator has drawn much attention since it requires no external power, and the large flapping amplitude of the agitator strengthens the intensity of the turbulence. This project will conduct comprehensive experimental investigations to fully establish the flapping dynamics of agitators and resulting heat-transfer characteristics. It will also develop a machine-learning assisted design methodology to enhance performance. The project will integrate machine learning into fluid dynamics research and education, instill the excitement of engineering in high-school students through summer camps, and attract prospective students from under-represented groups towards STEM fields in college. The objective of this project is to develop a machine-learning assisted methodology for fast-optimization of the flexible agitator design in various channel flow conditions towards the maximum synthetic thermal-hydraulic performance. The research approach includes i) Acquire the training data through systematic experimental characterization of self-agitator dynamics on the heat transfer enhancement; ii) Characterize the vortex dynamics due to the added self-agitator in the channel flow with a state-of-the-art stereo time-resolved particle image velocimetry system; iii) Establish a machine‐learning assisted design methodology by using experimental results as training and guiding data to bridge between the given flow and structure conditions and the resultant self-agitator vibration mode, vibration amplitude, vibration frequency, vortex shedding frequency and evolution on the final thermal-hydraulic performances. This work will serve as a comprehensive step towards achieving a fundamental understanding of vorticial flow dynamics under the effect of fluid-structure interaction. Such a knowledge base will pave the way for designing novel air-side heat exchangers that can achieve high-efficiency heat transfer without unduly increasing pumping power. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.