The computational domain is divided into finite elements. The elements connect at nodes. Either one-dimensional (1D), two-dimensional (2D), three-dimensional (3D) or axi-symmetrical (2D) domains can be used. All fields (velocities, stresses, etc.) are assumed to be continuous over the domain! This differs from common finite element programs where quantities like stresses, etc. can show jumps over element borders. This continuity gives a more diffuse but at the same time a more stable numerical scheme. Furthermore it has the advantage that all fields can be stored in the nodes (and not in integration points as in common finite element programs); this leads to a relatively easy implementation of remeshing and refining techniques.

Only first order in time equations are solved. Time derivatives are approximated with Euler backward time discretization.

tochnog 2001-09-02