Adjoint solver technology looks as one of the most promising tool for CFD development. And the current adjoint implementation embedded in Fluent 15 is today one of the most performing and reliable one.
Adjoint sensitivity data allows to predict the effect of the position of each node of the calculation mesh on an observed quantity (that could be a pressure drop a drag force or a complex function defined by the user). This means that the user can understand what areas have to be updated to obtain the maximum benefit.
Mesh morphing tool RBF Morph allows to accurately redefine the shape of the mesh and it has been coupled to the adjoint technology to fully exploit the potential of obtained sensitivity. Two main features are now available: adjoint sculpting and adjoint preview.
Adjoint sculpting allows to use sensitivity data obtained at the end of an adjoint run. RBF Morph can interact with the morphing box of the adjoint module so that shape information can be captured and used in a standard RBF Morph set-up. Several shapes (different portion of the model) can be captured in the same session and imposed enabling further constraints i.e. preserving details that cannot be modified.
Adjoint preview allows to compute adjoint derivatives with a respect to shape parameters imposed by the user. Standard RBF Morph shape modifications can be used and/or shapes defined using adjoint sculpting. The great potential is that the user can compute sensitivity with respect to an arbitrary number of design variables without any extra computational efforts. Sensitivity can be evaluated with respect to the baseline or with respect to a different configuration produced by a given status of shape parameters (a recalculation of CFD and adjoint solutions is obviously required). This allows to enable local optimisation workflow gradient based (as for instance the Steepest Descent Method) or to enrich DOE tables using sensitivity data.
Whatever is the workflow a key feature enabled using RBF Morph is the capability to update the CAD model of the original design into the new optimized one. A NURBS transformation is required for this purpose and NURBS morphing is then used to generate the new geometry.
Advanced coupling between RBF Morph and the adjoint solver is demonstrated with practical examples. Two examples on simple models that are part of RBF Morph Tutorials line (internal flow and external flow). Two examples on complex industrial application: the airbox of 3 cylinders engine, and the body of a sedan car.