Three-dimensional shape optimization of internal fluid flow systems using arbitrary shape deformation coupled with computational fluid dynamics

A new method is developed for three-dimensional shape optimization of internal fluid flow systems. It solves two of the major obstacles associated with three-dimensional shape optimization. The first is the current lack of a general methodology for the parameterization of three-dimensional shapes. The second is the absence of a method to manipulate the shape of the analysis model while maintaining its ability to produce accurate results. This method, call arbitrary shape deformation (ASD) is similar to lattice deformation methods currently used in the field computer graphics.; The key component of this method is the creation of a parametrically defined volume of arbitrary shape. An object is embedded in the volume by calculating its parametric coordinates in the volume. The parametric volume is deformed by moving its control points and the embedded object is similarly deformed by using its parametric coordinates to calculate its new global coordinates. The control points of the parametric volume form the set of parameters used to change the object's shape. The deformation is volumetric so that the entire analysis model is deformed, and G1 continuity is maintained throughout the model.; Several examples are solved where ASD is coupled to a computational fluid dynamics (CFD) analysis code and gradient-based optimization algorithms. A Y and tee fittings are optimized to minimize the static pressure drop for laminar flow conditions. The shape of the inlet and outlets are left unchanged. Only the region encompassing the intersection of the inlet and outlet branches is modified. Non-intuitive shapes were obtained with significant improvements in the pressure drop. The Y pressure drop was lowered by 40% and the tee pressure drop was lowered by 52%. In addition, the final pressure drop for the tee was essentially the same as the Y.