A critical decision when running FEA is how you will represent the model being simulated. IGA is more accurate and robust than traditional FEA because it leverages exact geometry and superior basis functions.
In product development, computer-aided design (CAD) software is used to create models suitable for computer-aided manufacturing (CAM) and to create photorealistic renderings to communicate the design before it is manufactured. CAD models accurately represent the important geometric features and the smoothness of the manufactured object and can be changed to meet evolving design and engineering requirements.
Loss of accuracy through model simplification
Running FEA as part of the design-through-analysis process based on CAD is commonly referred to as computer-aided engineering (CAE). Traditional FEA programs require a snapshot of the CAD, often simplifying the model by removing details like holes and fillets, and freezing the model resolution at a certain tolerance by converting it to a mesh. This new model, now only an approximation of the CAD, becomes the simulation model. Once converted to a mesh, an up-close view of the model reveals it is no longer a smooth part, ready for manufacturing. Now it is a new faceted model, an approximation of the original CAD model, made up of lots of small elements of different shapes like triangles, quadrilaterals, tetrahedra, or hexahedra.
Tedious and time-intensive preparation
FEA works best if the mesh in the CAE model is of high quality, meaning that the elements are all of uniform size and shape. For example, if the mesh was composed of quadrilateral elements, it would be desirable for each quadrilateral to be as close to square-shaped as possible. For simple simulations, this can be trivial, but for complex simulations like automotive crash, this takes weeks, or even months, of manual effort to generate a single mesh.
Hard to get results back to CAD
Once an FEA simulation is performed, the results might suggest a change in the design is needed. However, since the CAE model is disconnected from the original CAD model, transferring FEA results back onto the CAD model may take just as long, if not longer, than creating the original CAE model. In practice, the significant delays in (1) building the original CAE model and (2) transferring data back onto the CAD often mean that the original CAD model has since changed and is no longer in sync with the FEA results, making it difficult to incorporate the required changes in the design. And the more complex the situation, in terms of the CAD or the physics to be simulated, the worse the situation becomes. These challenges often relegate FEA to the end of the detailed design phase of product design rather than informing the entire design process.
In contrast to the laborious and error prone process of translating CAD into CAE models, isogeometric analysis (IGA) performs the FEA simulation directly on smooth CAD geometry. Practically, there are three ways to introduce the benefits of IGA into a design-through-analysis CAE pipeline:
- Transforming FEA meshes into analysis-suitable CAD geometry. This approach requires the traditional mesh generation step of first approximating the CAD with a faceted mesh. However, this mesh is then converted automatically to a smooth watertight analysis-suitable CAD geometry, yielding a closer approximation to the original CAD, which can be used as a viable CAD substitute for future simulations and design updates, and providing more accurate and faster simulation results due to the smooth basis. Bezier elements can be extracted from the analysis-suitable CAD object and used directly in solvers that have been enhanced to handle smooth geometry, like LS-DYNA or Coreform’s in-development Analyze solver. While this isn’t true “isogeometric” analysis - i.e., the native CAD geometry isn’t actually used and mesh conversion is still required, being able to run simulations on a smooth basis still provides many compelling benefits.
- Analysis directly on traditional CAD models. CAD models are generally represented by what is called Boundary Representation (BREP) geometry. These BREP models can be used directly as input to IGA. The time saved from not meshing is significant, and the simulation is done directly on the CAD model. BREPs often contain many trimmed NURBS patches, which may lower the accuracy of the results. However, the integration with CAD makes this workflow attractive. Fundamental research has been completed in this area, but it is not yet commercially available.
- Analysis directly on analysis-suitable CAD. Some CAD technologies, like Autodesk’s T-splines, create watertight geometry that is well-suited for IGA simulation. Such geometry, which is ideal for 3D printing or aesthetic design, can be used directly in IGA solvers.
Benefits of IGA:
Below are some key benefits of IGA. Learn more by reading through our compiliation of published IGA research results.
- More accurate simulation, since analysis is run on the actual smooth geometry, not an approximation, and higher-order smooth basis functions are employed.
- Time saved from preparing the mesh.
- Time saved in actually running the simulation itself, especially for hard problems.
- Integration with CAD to enable fully integrated design iteration.
- More realistic post-processed results on the CAD geometry, which help convey the analysis results better to non-experts.