systematic approach for using density-based topology optimization to find design solutions for a heat exchanger distribution system with turbulent incompressible flow

<jats:p>Background:Topology optimization is a powerful tool that allows engineers to accelerate the optimization process of any given part, as opposed to parametric optimization. An optimization algorithm automates the iterative cycle of simulating, evaluating, and modifying a given part until reaching certain objective values. This expedites the engineering process and makes it possible to generate many different solutions for a given problem in a small amount of time. Evaluating these solutions can indicate favorable design choices, which can be implemented into a final design by the engineer. The strength of topology optimization lies in its ability to create new forms without restriction that fulfill certain imposed constraints, such as fabrication or operating limitations. On the other hand, many optimization settings exist that can vastly alter the result. It is not a replacement for the engineer. Instead of optimizing a part directly, the engineer now searches for the starting values, constraints and objective functions that result in the best solution. The seminal work on the method focuses on its application for structural mechanics in 1988, where its use produces the recognizable organic truss features. Since then, it rose in popularity with the advent of additive manufacturing, which facilitates the manufacturing of the complex generated structures. In recent time, it has been expanded for use in other applications, notably for fluid mechanics, the application in focus for this paper. Commercial and open-source programs are available that are capable of topology optimization, such as COMSOL, OPENFoam, and ColdStream. One of the goals of this thesis is to quantify the results from these programs generated from identical initial conditions and compare their strengths and weaknesses. Likewise, the effects of modifying initial values of the optimizer will be documented. The work done for this thesis is in collaboration with SAFRAN Tech and therefore the exact part geometry and conditions of the optimization problem are confidential. Instead, a theoretical optimization problem will be presented in this thesis, which represents the actual one: a heat exchanger collector with turbulent incompressible flow. The examined system is intended for integration into a gas turbine engine with a recuperated cycle. This type of engine aims to increase thermal efficiency by extracting the heat from the exhaust and injecting it into the flow before the combustion chamber. All the necessary additional ducting and heat exchangers must be designed with the smallest amount of losses possible to avoid a net loss in efficiency by adding the recuperated cycle to a given engine. Tasks:The primary task for this thesis is the quantification of optimization results from several CFD programs and their evaluation and comparison. The model used will be that of a heat exchanger collection system for a recuperated cycle gas turbine engine. The flow speed in such applications means that a turbulent flow regime is unavoidable. Topology optimization for turbulent flow is known to be problematic; thus, methods need to be developed to overcome this. COMSOL OpenFOAM and ColdStream are the programs that will be evaluated for their effectiveness. They will be tested concerning their performance on an HPC cluster and the quality of their results. In addition, an approach will be developed to obtain useful results using topology optimization. This will comprise the choice of starting values and optimization parameters.</jats:p>