ECCOMAS 2024

Design of Parallel Flow Manifolds using Topology Optimization in Open-Source Software SU2

  • Vermani, Sanjay (Flemish Institute for Technological Research)
  • van der Mark, Luuk (Delft University of Technology)
  • De Servi, Carlo (Flemish Institute for Technological Research)
  • Anand, Nitish (Flemish Institute for Technological Research)

Please login to view abstract download link

Flow maldistribution within heat exchangers is a widely studied issue, especially for applications using parallel tubes or flow channels [1]. The primary reason behind such maldistribution is the pressure drop in the channels, which is directly governed by the manifolds. Consequently, the design of the heat exchanger manifold becomes critical to provide the required pressure drop and hence the optimum flow distribution. In this respect, the use of simplified header design methods available in literature becomes challenging due to the vast range of operating conditions and flow distribution requirements. Hence, to circumvent this selection process, this work presents a topology optimization framework to design a manifold using open-source CFD-Suite SU2. To achieve the objectives of this paper, the incompressible Navier Stokes solver in SU2 is extended for the fluid flow topology optimization problems using the well-known Brinkman penalization term [2]. In addition, new objective functions specific to topology optimization were implemented, such as maldistribution function, volume fraction, and power dissipation. The flow solver is then coupled to a gradient-based optimizer, using the objective and constraint sensitivities obtained from the adjoint solver of SU2. To verify the implementation, a one inlet and multiple outlet test case was selected. The manifold for this test case was designed using a popular design method as per Ref. [3] and is then compared with that obtained using topology optimization. In addition, to further understand the behavior of topology optimization, different combinations of objectives and constraints are studied for different flow conditions. The results show a significant improvement in the flow distribution in the channels, demonstrating the potential of this approach to explore new possibilities in the parallel flow header designs.