Shape and Topology Optimization in Computational Fluid Dynamics Including Heat Transfer Using Gaussian Processes and Adjoint Methods

摘要

Methodologies for shape and topology optimization in Computational Fluid Dynamics (CFD) including heat transfer are presented and applied to the design-optimization of industrial parts. The CFD and heat transfer optimization methodologies are implemented within AcuSolve® software. To optimize geometrical shapes with predefined topology, the parameterization of the shape is achieved using a moderate number of design variables and the deformation and optimization of the shape is based on a Gaussian process model. When, additionally, the topology of the shape is sought, the domain is parameterized with a field of porosity design variables and the optimization is performed using the continuous adjoint approach to the Galerkin Least Squares solver on which AcuSolve® solver is based. The shape and topology optimization methods are implemented in several industrial applications such as the design of a heat exchanger for maximum heat transfer and minimum total pressure losses, the fluid-thermal optimization of a bladeless fan, the design-optimization of a battery cooling model and the topology optimization of a manifold, an elbow duct and a serpentine flow.