FEA Mesh Generation and Element Size Accuracy

Mesh generation is the foundational process of dividing a continuous Computer Aided Design (CAD) geometry into smaller, discrete components called elements so that Finite Element Analysis (FEA) solvers can calculate physical behaviors such as stress, heat transfer, or fluid flow. The sizing of these elements plays a decisive role in the precision of the simulation; smaller elements typically produce significantly higher numerical accuracy and better capture localized stress concentrations, but they also drastically increase computational time and memory usage. Understanding this balance, along with performing mesh convergence studies, is essential for obtaining reliable engineering data without unnecessarily exhausting computing resources.

In modern CAD environments, 3D models are defined as continuous, mathematically ideal surfaces and solids. However, governing differential equations for physical phenomena cannot be solved analytically for complex geometries. Mesh generation bridges this gap through discretization, replacing the continuous domain with a network of simple geometric shapes—such as one-dimensional beam elements, two-dimensional triangular or quadrilateral shells, and three-dimensional tetrahedral or hexahedral solids. Each element contains nodes where the governing equations are solved, while mathematical shape functions approximate the physical field between these nodes.

Element size directly impacts the stiffness and numerical sensitivity of the FEA model. When large, coarse elements are applied, the mathematical formulation tends to artificially stiffen the structure. Coarse meshes struggle to accurately track steep physical gradients, such as rapid changes in stress near sharp corners, holes, fillets, or applied point loads. As a result, an overly coarse mesh often leads to a severe underestimation of peak stresses and inaccurate deformation predictions, giving engineers a false sense of structural security.

Conversely, refining the mesh by reducing element size allows the model to capture geometric nuances and sharp field gradients with high fidelity. As element sizes approach zero, the numerical FEA solution approaches the exact analytical solution of the underlying equations. However, this increase in precision comes at a steep computational cost. Halving the global element size of a 3D solid model can increase the total degree of freedom count by nearly an order of magnitude, drastically driving up solve times and file sizes.

To achieve reliable accuracy efficiently, engineers rely on mesh convergence studies and localized refinement rather than globally reducing element sizes. A mesh convergence study involves running successive simulations with decreasing element sizes until key output metrics, such as maximum Von Mises stress or displacement, stabilize within an acceptable percentage threshold. Furthermore, localized mesh controls enable designers to specify dense, fine elements exclusively in critical regions with high stress concentrations, while maintaining coarser elements throughout non-critical, uniform sections of the CAD model. This targeted approach ensures maximum simulation accuracy while keeping computational demands manageable.