RBF Morph: a mesh morphing add-on for ANSYS Fluent
RBF Morph is a morpher that combines a control of the geometrical parameters with an extremely fast mesh deformation, fully integrated in the CFD solving process. The final goal of RBF Morph is to perform parametric studies of component shapes and positions typical of the fluid-dynamic design like multi-configuration studies, sensitivity studies, design developments and shape optimization. This product is now available for FLUENT users as an add-on.
Executing Simulation Driven Product Development typically includes performing parametric studies (i.e., multi-configuration studies, sensitivity studies, Design Of Experiments) where a component’s shape or position is updated to assess the impact on fluid-dynamic characteristics. A common approach is to update the initial geometry, re-mesh the entire domain, and then re-run the flow analysis. A viable alternative in many cases is to modify the mesh and re-run the flow analysis without going back to the geometry step until the optimum configuration is determined. Distributed by an ANSYS, Inc. partner, RBF Morph allows an ANSYS FLUENT user to perform shape modifications through mesh updating. Using this tool, there is no need to update the geometry until after the final design is selected. The entire set-up can be done inside ANSYS FLUENT using a comprehensive GUI which allows a user to define the morphing problem. TUI commands are also available to drive the morpher by means of simple scripts. No additional mesh I/O is required as the morpher acts directly during the parallel solving stage.
RBF Morph development has been driven by a space research fluid dynamic analysis performed for the european launchers top team considering that the most important requirements for high fidelity mesh morphing are:
Parallel morphing of the grid
Ability to morph very large models (hundreds of millions of cells) in a few minutes
Support for every kind of mesh element type (tetrahedral, hexahedral, polyhedral, prismatic, hexcore, non-conformal interfaces, etc.)
RBF Morph satisfies these requirements using state-of-the-art RBF (Radial Basis Function) techniques. RBFs are capable of interpolating a prescribed mathematical function, which is defined at discrete points within a domain, yielding exact values at these points. The behavior of the function between points depends upon the type of the mathematical function prescribed.
Mesh morphing with RBF Morph is executed in three steps:
Step 1: setup and definition of the problem
Step 2: solution of the RBF system
Step 3: morphing of the surface and volume mesh
Mesh modifiers are prescribed in Step 1, each with its own user specified magnitude (a scalar value sets the intensity of the modifier). In Step 2, the effect of each modifier upon the mesh can be verified by previewing its action without actually moving the nodes. An undo capability can be exploited, allowing a user to verify resulting mesh quality and update the setup if problems are found, such as negative cell volumes. The morphing operation in Step 3 can be performed in parallel, which is highly valuable when morphing large problems, and it scales practically linearly.
A single solution can be amplified by a scalar multiplier in a non linear fashion that allows rotations to be properly accounted for. Several solutions can be combined together by summing their effects having considered the original shape as the starting point. The result is a shape parametrization of the fluid domain. The superposition of several solutions can be previewed or imposed using the GUI or automated in batch mode using a journal file thanks to the general morphing command:
(rbf-morph ‘(("sol-1" amp-1) ("sol-2" amp-2)...("sol-n" amp-n)))
that executes a morphing of the mesh using an arbitrary number of RBF solutions. The new command makes the CFD model parametric with respect to shape.
Innovations & Advantages
RBF Morph is a preferred choice because it has been designed for cutting-edge applications to satisfy the most demanding users and to provide unique functionalities not available on the market. The result is that there is more than one good reason to prefer it:
it's fast and parallel
it allows to manage very large models
it's fully integrated in the CFD process
it's easy and quick to setup
it gives exact control of nodes and features
it allows for full parameterization of the mesh
it manages any possible mesh element type
no particular I/O and disk usage is required in addition to the standard calculation (the mesh is morphed while solving)
it allows to submit large DOE studies in a few seconds
if zero is assigned to a parameter, it gives is exactly the same solution of the baseline
it fully supports the regeneration of the CAD.
Some of the features that makes it one of the fastest tool for mesh morphing available on the market today are:
Product Integration: full integration with ANSYS Fluent.
User Interface: dedicated GUI and TUI (scriptable).
Process Integration: morphing directly inside the solving stage without modifying the geometry, regenerating the mesh and setup again the case.
Mesh Topology: modification of the original surface and volume mesh producing a nodal smoothing without changing the mesh topology.
Surface Morphing: surface mesh can be modified by free surface deformation, rigid movement or scaling.
Volume Smoothing: high quality smoothing of the volume mesh with relatively large movements possible in a single step deformation.
Versatility: nodal smoothing is achieved by means of a mesh-less approach that is independent from the mesh structure, handling every kind of mesh element type (tetrahedral, hexahedral, polyhedral, prismatic, hexcore, non-conformal interfaces, etc.).
Reusability: the RBF solution can be applied to any different mesh representing the same geometry.
Consistency: mesh characteristics are preserved so mesh consistency is ensured (element size, type and distribution, prism layers, etc.).
Parallelism: parallel calculation for large models (many millions of cells).
Efficiency: flow solutions are fully readable through all the morphed mesh, reducing the number of iterations to converge.
Precision: exact movement is ensured for the moving nodes locations as well as exact feature preservation.
Parameterization: multi-parameter and multi-step problems.
CAD: transfer modifications back to CAD by applying to STEP files similar modifications done on the mesh, in order to fully support the re-design of the morphed surfaces.
Current and Potential Domains of Application
Within the space field this innovation was applied to aerodynamic CFD simulation and computing but it may have many applications in the automotive and aeronautic field.