Modeling and Pareto based multi-objective optimization of wavy fin-and-elliptical tube heat exchangers using CFD and NSGA-II algorithm

[Display omitted] •Multi objective optimization of wavy fin-and-elliptical tube heat exchangers has been performed.•The results obtained from CFD simulations have been used as inputs for the GMDH type artificial neural networks.•Four parameters have been chosen as design variables to optimize two ob...

Full description

Saved in:
Bibliographic Details
Published in:Applied thermal engineering Vol. 111; pp. 325 - 339
Main Authors: Darvish Damavandi, Mohammad, Forouzanmehr, Mostafa, Safikhani, Hamed
Format: Journal Article
Language:English
Published: Elsevier Ltd 25.01.2017
Subjects:
Elliptical tube
GMDH
Multi objective optimization
NSGA-II
Wavy fin
NSGA-II
Elliptical tube
GMDH
Multi objective optimization
Wavy fin
ISSN:1359-4311
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:[Display omitted] •Multi objective optimization of wavy fin-and-elliptical tube heat exchangers has been performed.•The results obtained from CFD simulations have been used as inputs for the GMDH type artificial neural networks.•Four parameters have been chosen as design variables to optimize two objective functions.•From among the optimal points, several points with unique features have been introduced and explained. In this paper, a multi-objective optimization (MOO) of wavy fin-and-elliptical tube heat exchangers has been performed by using Computational Fluid Dynamics (CFD), Artificial Neural Network (ANN) of Group Method of Data Handling (GMDH) type, and Non-Dominated Sorting Genetic Algorithm II (NSGA-II). This multi-objective optimization is aimed at achieving maximum heat transfer and minimum pressure drop. For this purpose, the considered objective functions, Colburn factor (j) and friction factor (f) are optimized with regards to the design variables (four variables). The CFD results are validated by means of experimental findings. Polynomials of the GMDH type neural network are formed based on the CFD results. These polynomials relate the objective functions to the design variables. Ultimately, the NSGA-II algorithm obtains the Pareto optimal points by using the input data from the neural network. From among the optimal points, several points with unique features are introduced and explained. The investigation of optimal points indicates that with a slight reduction in heat transfer, pressure drop can be reduced considerably. By combining and simultaneously using the CFD, neural network and NSGA-II optimization algorithm, very useful and valuable results are obtained; which otherwise couldn’t be achieved without the mutual use of these techniques.
ISSN:1359-4311
DOI:10.1016/j.applthermaleng.2016.09.120