KONWIHR

COMPETENCE NETWORK FOR TECHNICAL, SCIENTIFIC HIGH PERFORMANCE COMPUTING IN BAVARIA

PBOHEX

Pareto-based optimization of liquid-gas heat exchangers

Heat exchangers are vital in many industrial processes and find a widespread use in our daily life. Hence, their optimization is of great importance. However, for such devices the optimization objectives involved are numerous and often conflicting. The interaction among these different objectives gives rise to a set of compromised solutions, known as the Pareto optimal set.

The work proposed aims at the multi-objective optimization using the Pareto approach applied to liquid-gas heat exchangers that contain pin-fins and finned tubes. This optimization is to be achieved through the determination of optimal heat transfer enhancement elements regarding shape, height, material, array arrangements and all other parameters that lead to a high heat transfer rate with a low pressure drop at acceptable price and compactness.

The optimization approach adopted uses genetic algorithms (GA) to explore the design space and search for the final Pareto-frontier. The GA will be coupled to an automated CFD analysis for the determination of the objective functions of the initial populations and of the new individuals (offsprings) generated during the progress of the GA.

High performance computing (HPC) tools developed within preceding FORTWIHR and KONWIHR projects are essential for this investigation. They will give the opportunity to tackle real industrial applications where complex CFD models with the coexistence of different physical phenomena (e.g., fluid flow and heat transfer) are needed. The HPC tools will also be used to speed-up the optimization process by simultaneous simulations of the initial population individuals and by the parallel processing of the offspring evaluations.

For complex models that need tremendous analysis and CPU efforts, approximation techniques will be employed to approximate the objective functions of the non-apriori known individuals of each GA iteration. This significantly reduces the total number of investigations required. Comparison between the investigated approximation techniques will be done.

The final objective is to have a computational tool available that allows to optimize heat exchangers based on genetic algorithms and HPC CFD methods. That will be tested on a variety of different heat exchanger configurations.