The treatment of the locking phenomenon for a general class of variational inequalities (Q596185)

The numerical locking phenomenon for a general class of variational inequalities is considered. This means that the variational inequality depends on a small parameter and the numerical locking corresponds to the absence of the uniform convergence with respect to it. The considered class of variational inequalities is a generalization of a variational equation introduced by \textit{D. Chenais} and \textit{J.-C. Paumier} [Numer. Math. 67, 427--440 (1994; Zbl 0798.73054)]. The authors obtain sufficient conditions for the uniform convergence of the standard Galerkin approximation method. This generalizes the conditions obtained by Chenais and Paumier in the conforming case and by \textit{D. Capatina-Papaghiuc} and \textit{J.-M. Thomas} [ibid. 81, No. 2, 163--186 (1998; Zbl 0918.65070)] in the nonconforming case. The general result obtained is applied to a Signorini problem with stiff transmission. Numerical experiments are presented.