Multi-objective optimization on fluid flow and heat transfer of 3D micro-fin surfaces

Abstract

Convective heat transfer enhancement (HTE) techniques push to increase the removal of excess heat from a solid area to the working fluid. In passive HTE techniques, the convective heat transfer is enhanced through geometric means by improving the flow and inserting attentively designed roughness — often termed fins, dimples, twisted tapes, rib tabulators, and wire coils — on the solid bodies throughout the flow passages. Enhanced surfaces, either macro or micro-textured, are commonly applied in commercial applications such as heat exchangers, cooling systems, chillers, refrigerants, power plants, air conditioners. Micro-fins (< 0.5 mm tall) have the advantage of reducing thermal resistance and the disadvantage of increased pressure drop when applied in heat-exchange applications. Turbulence and heat transfer are two critical factors in the chain of world energy consumption. Hence, the development of more efficient energy transfer techniques in engineering applications through control and optimization of the thermal turbulent boundary layer and turbulence continues to be a subject of intense interest. This study reviews a passive flow control technique with a focus on unique characteristics of heat transfer in devices called riblets. These surfaces are considered beneficial to many commercial applications as they reduce drag skin friction between a surface and fluid and, in some cases, can enhance heat transfer in the turbulent flow regime. Hence, riblets are potential candidates to improve the efficiency of HTE equipment as technology pushes to revolutionize the design and manufacturing of heat exchanger equipment. Studies of flow fields around the fins have appeared in the literature but often do not include conduction analysis in the fin, leaving a key optimization parameter unexplored. In addition, while 2 dimensional (2D) helical micro-fin tubes are the most widely used in heat exchanger applications, experimental results have demonstrated that many 3D micro-fin geometries outperform 2D fins. Hence, one of the goals of this study is to fill these gaps by finding out the effects of 2D and 3D micro-fin design variables on heat-transfer enhancement and pressure drop. Another goal of this study is to evaluate the potential of a data-mining model as a surrogate of computational fluid dynamic (CFD) models in 2D and 3D micro-fin tubes. Data-mining is a procedure of extracting and recognizing useful information and succeeding knowledge from databases using mathematical, statistical, artificial intelligence, and machine learning technique, and is still not often used in heat transfer studies. Two different least squares-support vector regression (LS-SVR) models are employed to estimate the Colburn j factor (j) and the Fanning friction factor (f) as functions of micro-fin geometric variables. Results of the parametric study showed that the best 2D micro-fin tube can enhance the thermo-hydraulic efficiency index (η) up to 1.18. Results of the LS-SVR model showed that the percentage of average absolute error (AAE) between simulated and estimated values of the Colburn j factor and the Fanning friction factor is 2.05% and 2.93% for 3D micro-fin tubes, respectively. The statistical results of this study proved that LS-SVR model can be an accurate surrogate for the CFD-based model. Because of the great potential for optimization of micro-fin surfaces and the lack of any optimization work encompassing 3D geometries in this field, this study proposes a robust methodology to solve the multi-objective optimization problem for this class of heat exchanger surfaces with an accurate, first-principles numerical model and thereby determine the flow physics that characterize the optimal geometries. This study provides a representation of the geometry of 3D micro-fins with enough degrees of freedom to encompass a wide range of physically realizable surface geometries. To reduce computational time, a fluid flow channel is presented with a novel computational domain. The channel contains two-sided periodic domains and micro-fins on the top and bottom walls having an opposite angle. An accurate and parallelizable numerical model is developed based on Reynolds average Naiver-Stokes equations (RANS) and a realistic k-ɛ model. The simulation procedure applies ANSYS Fluent 19.1 to solve convective heat transfer and turbulent flows over micro-fins and conductive heat transfer for solid conduction to maintain robustness for geometries that do not yield unity fin efficiencies. A boundary condition of constant temperature is defined at the channel walls. This study provides a robust methodology to integrate the numerical model into a multi-objective optimization algorithm named non-dominated sorting genetic algorithm (NSGA-II) to generate the best trade-off (Pareto front) solutions between enhancement of the Nusselt number and the Fanning friction factor of a micro-fin surface to a smooth surface at Reynold number (Re) of 49,000. Information from the Pareto front solutions will be used to explore the thermal and flow fields of the optimal solutions. The result of the Pareto front showed the significant effects of micro-fin height and pitch on objective values of optimal design geometries. This study presents a unique intersection of the drag reduction approach and heat transfer enhancement by finding unique operating geometries. For example, a particular optimal design geometry reduced drag by 8% but was also beneficial to heat transfer increasing by 21% compared to a smooth surface.