INVESTIGATION AND OPTIMISATION OF THERMAL PERFORMANCE IN WAVY MICROCHANNELS

Abstract

In this dissertation, steady, laminar liquid-water flow in periodic microchannels of different wall curvatures (sinusoidal, and generalised waveform shape) was investigated numerically. Surrogate models were constructed to describe the correlation between the geometric variables and the objective functions characterising the heat transfer performance and pressure loss penalty. Subsequently, two different optimisation techniques were executed to identify the optimal geometric configuration in each case at different Reynolds number. The sequential quadratic programming (SQP) was used for single-objective optimisation, while non-dominated sorting genetic algorithm II (NSGA-II), a variant of evolutionary algorithms, was adopted for the multi-objective optimisation. Pareto frontier curves were also plotted to provide a clear overview on the evolution of two opposing objectives, that of pressure loss and heat transfer enhancement, as the geometric parameters were varied accordingly. Eulerian-based (vector field maps and helicity isosurface contours) and Lagrangian-based (Poincaré maps) analyses were performed to analyse the fluid flow behaviour.