Influence of the type of boundary conditions on the numerical properties of unit cell problems

Abstract

Aspects of the numerical solution of the systems of equations resulting from the computational homogenization of unit cell problems using the finite element method are discussed. Different kinematic boundary conditions and solution techniques are examined and compared, both, theoretically and numerically. It is found that the combination of boundary conditions and solver significantly influences the computational cost in terms of memory and cpu time. Examples for model and real world problems are presented.