Improving thermal and hydraulic performances through artificial neural networks: An optimization approach for Tesla valve geometrical parameters

Abstract

Customizing the configuration of Tesla valves should be adjusted to match the specific goals of the desired purpose. Channels with Tesla design have various applications in hydraulic and thermal fields. If the purpose is hydraulic application, this design of channels is used in the construction of one-way valves. The main focus may not be on thermal efficiency, instead, the emphasis lies on attaining a strong diode characteristic. However, in situations involving thermal applications, and considering that the thermal performance of these channels is suitable, they are used in the construction of heat sinks and heat exchangers. Microchannel heat sinks with Tesla design are used in cooling of CPUs, batteries, etc. In these situations, the goal is to minimize the pressure drop, while maximizing convective heat transfer. This research introduces an innovative structural design for the Tesla valve and presents two artificial neural network (ANN) models for predicting and optimizing its thermal and hydraulic capabilities. The design parameters being studied are the inner radius of the fins (R), the length of the folded segment of the divider (L), and the angle of the folded segment of the divider (θ). The researchers are examining how these variables affect the Nusselt number (Nu) and pressure drop (ΔP) in both forward and reverse flow directions. Based on the findings, by utilizing variables of optimal design 4 (θ = 130°, L = 3 mm, and R = 0.258 mm) and optimal design 5 (θ = 130°, L = 3 mm, and R = 0.586 mm) during the valve manufacturing process and altering the flow direction from forward to reverse, the thermal performance and diodicity increased by about 26.6% and 78.2%, respectively.

Keywords

• Artificial neural network
• Multi-layer perceptron
• Optimization approach
• Geometrical parameters
• Tesla valve
• Pressure drop