[en] The discretization errors of finite element approximations are examined in an abstract frame which is totally indepedent of the way of obtention of the solution. It is found that the error may be decomposed in an equilibrium error and a compatibility error, which may be combined as orthogonal vectors. Both errors admit upper and lower bounds which may be computed in a post-treatment scheme.
[Université de Liège - ULg > Département d'aérospatiale et mécanique > Méthodes de fabrication >]
