A continuum three-dimensional, fully coupled, dynamic, non-linear finite element formulation for magnetostrictive materials

Magnetostriction is a phenomenon observed in all ferromagnetic materials. It couples elastic, electric, magnetic and in some situations also thermal fields and is of great industrial interest for use in sensors, actuators, adaptive or functional structures, robotics, transducers and MEMS. In this work, the governing equations of the three-field problem (i.e., the interactions of elastic, electric and magnetic effects) are formulated in three dimensions, accounting for non-linear (through magnetic body forces represented by the Maxwell tensor) and dynamic effects, and with constitutive equations resembling those of piezoelectricity. Through manipulation of Maxwell equations it is possible to find suitable expressions for developing the numerical weak, Galerkin and matrix forms in a natural way, including seven residuals (one for each nodal degree of freedom) and non-symmetric tangent, 'capacity' and mass consistent matrices. Simple backward Euler and central difference schemes can be used for the time domain integration. The only assumption made in this work for simplification is that the time variation of electric induction is negligible. This is justified by the relatively low frequencies ( GHz) under which magnetostrictive materials usually work. The principal feature of the equations is the use of a magnetic potential (without much physical meaning) that allows a complete 'displacement' finite element formulation: all elastic, electric and magnetic nodal unknowns are zero derivatives. This allows the algorithm to be treated in a standard way, and important effects such as eddy currents can be obtained naturally. The formulation is implemented in the research finite element code FEAP. Although seven degrees of freedom per node is computer expensive to solve (especially for 3D problems), the current trend in the performance of computers, even personal ones, makes it worthwhile to build complete finite elements following the well-established (in mechanics) Bubnov–Galerkin procedures. The paper closes with a simple one-dimensional example for purposes of numerical validation.