Improving efficiency and optimizing heat transfer in a novel tesla valve through multi-layer perceptron models
Peng Cheng, Jianjun Xu, Jitendra Kumar, Hamad Almujibah, Husan Ali, Tamim Alkhalifah, Salem Alkhalaf, Fahad Alturise, Raymond Ghandour
Abstract
Over the past few years, the distinctive design and versatile applications of Tesla valves have captured considerable interest across diverse industries. In contrast to conventional check valves, Tesla valves employ interconnected channels, establishing a highly efficient and reliable fluid flow control mechanism. This research delves into an investigation of the optimum geometric parameters that significantly influence the performance of a novel Tesla valve. The study focuses on three key geometric characteristics: the valve angle (α), the distance between consecutive stages (D), and the distance between the divider wall in the second stage of each step group and the wall of the straight channel (H). The authors carried out a numerical study using computational fluid dynamics to acquire the results. Four multi-layer perceptron models, each with a structure of 3-2-2-1, were applied to predict the selected responses of Nusselt numbers in the forward (Nuf) and reverse (Nur) directions, as well as pressure drops in the forward (ΔPf) and reverse (ΔPr) directions. The findings revealed that among all the variables examined, the parameter H exerted the most substantial influence on all measured responses. It was concluded that by incorporating specific values of α = 34.065°, D = 9 mm, and H = 5.624 mm during the manufacturing process of the valve and altering the flow direction from forward to reverse, a remarkable improvement of approximately 271.7% in pressure diodicity was achieved.