Simulation using finite element analysis (FEA) is used to predict physical effects, such as stress, fatigue, vibration, or heat transfer, without the expense of creating physical prototypes. Design models are subdivided into finite elements called a mesh. A solver uses mathematical equations to predict the behavior of each mesh element when subjected to static and dynamic loading conditions, then adds up the individual results to predict the behavior of the entire object. Using traditional FEA for complex problems can be extremely time consuming with many factors affecting the accuracy of the results.
A big problem with traditional FEA: The mesh
Simulation depends on the mesh that you use, so having a high-quality mesh representation of the geometry is essential for your simulation to be accurate. One of the biggest challenges of traditional FEA is that because the mesh is a faceted representation of a usually smooth model, it is inherently inaccurate, so results based on the mesh reflect that inaccuracy. A common technique to better represent the model is to create a finer faceted mesh that better captures shape and features. However, this approach comes at a high computational cost due to the higher element count.
The more complex the model, the harder it is to get a good mesh representation. For complex simulations, the process of building simulation models from CAD can account for 60%-90% of the design-through-analysis pipeline, and some companies budget millions of dollars each year just for geometry clean-up and mesh generation.