Midas FEA nonlinear solution algorithms are the load, boundary, equation-solver, and iteration methods used to solve linearized and nonlinear finite element systems in midas FEA.
How It Works
The manual's general algorithm section covers nodal constraints, skewed support, constraint equations, nodal loads, pressure loads, body forces, prescribed displacements, direct equation solvers, iterative equation solvers, and nonlinear iteration strategies.
Direct equation solvers include skyline and multifrontal methods. Iterative solvers include conjugate gradient and GMRES with preconditioning options such as ILUT and Jacobi. The manual notes practical solver choices: multifrontal solution is favored for buckling, Lanczos, constraint-equation cases, and some dynamic contexts; GMRES is used when an iterative method must handle unsymmetric stiffness.
For nonlinear iteration, the manual describes initial stiffness, Newton-Raphson, modified Newton-Raphson, and arc-length methods. Convergence can be checked by force norm, displacement norm, or energy norm. Softening problems require stricter attention, and comparing more than one convergence criterion is recommended when the appropriate measure is uncertain.
Solver Development Notes
Keep equation solver selection tied to matrix symmetry, definiteness, constraint handling, and procedure type.
Commit state only after a converged increment or step, especially for path-dependent material and contact states.
Test Newton, modified Newton, and arc-length paths with both monotonic hardening and softening/snap-through cases.
Make convergence norms part of the test harness output, not only internal diagnostics.
김경종