3D Shape Optimization of Consecutive Manifolds using Free Form Deformation and Adjoint-Based Methods

Abstract

Current design methods for oblique manifolds were first presented in the paper of London et al. in 1968. This method presents a single equation to shape the dividing manifold of several manifold configurations. However, current literature shows a potential to improve the current design method by implementing advanced techniques. This paper aims to bridge this gap by presenting an innovative design methodology incorporating these advanced techniques. The proposed design method integrates free-form deformation and adjoint-based methods within a 1st-order shape optimization framework. The objective function focuses on the minimization of mass flow mal-distribution, evaluated by an incompressible Reynolds-Averaged Navier-Stokes solver. Through three initial tests, this paper demonstrates a consistent improvement in the objective for all obtained designs. Noteworthy is the finding that small changes to the initial design lead to significant enhancements, evidenced by an 82.5\% decrease in the mass flow mal-distribution. Off-design testing further substantiates the effectiveness of the design method, showcasing superior performance under varying initial conditions for the optimized designs. Further testing not only exhibits improved objectives but also reveals common features among results for diverse initial designs. This paper shows a novel and effective design methodology for consecutive manifold systems, laying a robust foundation for an improved design method.