Constraint optimisation method for the heat transfer in a wavy microchannel

• Main Author: Xu, Yan
• Other Authors: Ooi Kim Tiow
• Format: Final Year Project
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Science::Physics::Heat and thermodynamics; Engineering::Mechanical engineering
• Online Access: https://hdl.handle.net/10356/141099
• Institution: Nanyang Technological University

Wavy annular microchannels were achieved by using macro geometry fabrication methodologies. A solid cylinder with wavy profiles on its outer surface and an outer pipe geometry, which could be manufactured by conventionally machining methods, were superposed concentrically to yield annular microchannels with a gap of 0.3 . Geometrical protrusions in the microchannels were proven to be capable of enhancing heat transfer performance and lower energyconsumption [1].Experimental studies and numerical simulationswereconducted for single-phase heat transfer in the steady-state condition using distilled water as the coolant with a Reynolds number range of 1300-4600 [1]. It was proved in [1] that high heat transfer coefficient occurred at a cost of high pressure loss along the pipe. Mathematical correlations of average friction factor and Nusselt number were proposed by adding new terms to simulate the effects of wavy profiles. Average Nusselt number and friction factor were correlated with amplitude and wavelength of wavy profiles and Reynolds number. Genetic algorithms were developed to investigate the optimization problem with two objectives, heat transfer performance and pressure loss. Parameterizing the problem, amplitude (protrusion height), wavelength (protrusion pitch length), and Reynolds number were encoded as ‘genes’ describing a unique solution, which was valid for a specific microchannel setup. Regression correlations of Nusselt number and friction factor were implemented in the objective functions. Genetic algorithms selected solutions of good fitness in terms of good objective function values from candidate populations through repetitive applications of crossover, mutation, tournament selection operators, which are metaheuristic inspired by natural selection process. With a population size of 100, the solution population computed from the algorithms converged in a pattern in the coordinate plane of pressure drop and heat transfer coefficient after 200 iterations. Resulting populations from algorithms gave best-pareto-front solutions, which allowed further decision-making based on the values of objectives and feasibility of microchannel setups. To validate the results, computational fluid simulation was demonstrated.