Naslov
Multi-Objective Optimisation of a Tube Bundle Heat Exchanger

Autori
Škurić, Vanja

Vrsta, podvrsta i kategorija rada
Ocjenski radovi, diplomski rad, diplomski

Fakultet
Fakultet strojarstva i brodogradnje

Mjesto
Zagreb

Datum
18.07

Godina
2014

Stranica
105

Mentor
Jasak, Hrvoje

Neposredni voditelj
Rusche, Henrik

Ključne riječi
optimisation ; OpenFOAM ; heat exchanger ; multi-objective ; single-objective ; MOGA ; SOGA ; evolutionary algorithm

Sažetak
Optimisation is a discipline of numerical mathematics which aims to improve the operation of a system or process in order to be as good as possible. Optimisation algorithms work to minimize (or maximize) an objective function, typically calculated by the user simulation code, subject to constraints on design variables and responses. Optimisation using CFD is discussed, giving insight to different optimisation approaches and their capabilities. Mathematical models describing fluid flow and heat transfer are presented. Tube bundle heat exchanger optimisation case is examined. The problem consists of finding the best positions of the tubes to maximize heat exchange (i.e. temperature increase of the fluid) while minimizing pressure loss. The two corresponding parameters being optimised are temperature increase and pressure drop of fluid between inflow and outflow. Coupling of different tools needed for the multi- objective optimisation of heat exchanger is presented, as well as the appropriate workflow. The Dakota software package, is used for optimisation ; Salome, is used for automatic geometry generation and OpenFOAM is used for meshing and CFD simulation. All of these software packages are open source. Different Dakota optimisation algorithms are described. For the heat exchanger optimisation Multi-objective Genetic Algorithm is used. Different parameters used in heat exchanger optimisation runs are explained. Results of optimisation runs using different parameters are presented. The optimisations were executed with different population sizes, different numbers of generations, different mutation and crossover rates. The results are presented in a Pareto front form. Comparison between acquired Pareto fronts is given. Multi-objective optimisation based on weighting factors method is discussed. The idea is to use the approach to generate Pareto fronts with equally spread points. Workflow and optimisation parameters are examined. Three optimisations were con- ducted and the analysis of the results is presented.

Izvorni jezik
Engleski

Znanstvena područja
Strojarstvo

POVEZANOST RADA