Petrov-Galerkin formulations of the Timoshenko beam problem (Q1103440): Difference between revisions

The paper presents the Petrov-Galerkin formulation as an alternative to the Reissner-Mindlin formulation in order to obtain an approximation of Timoshenko's beam solution by finite elements for straight bars under bending moments and shear stresses. The well known Petrov-Galerkin formulation represents in fact an optimization of Galerkin's classical residual formulation - weighted residual method in FEM formulation - using a much optimal way of weighted functions. The authors have introduced the notion of ``vector interpolation'' whose elements are interpolation functions corresponding to the degrees of freedom - translations and rotations - for a straight bar under bending moments and shear stresses of finite elements constructed by means of optimal weighting functions. As a result, the extension of the ``energy best approximation criterion'' is obtained for all available domains of finite elements which results in the optimal rate of convergence in the energy norm in Hilbert space. The Petrov-Galerkin formulation appears as a better performance compared with other solutions by finite elements because it eliminates the distorsion effect of the rate between the transversal section high of bar and its length that permits to obtain nodally exact values of the problem without taking count of beam load variations. In order to extend the Petrov-Galerkin formulation for obtaining interpolation functions to bidimensional problems, the authors suggest discontinuous weighting functions. With these functions it is possible to obtain nodally exact solution for the case of constant forces corresponding to structures discretized by finite elements. The paper is very interesting from the theoretical point of view, but we appreciate that the conclusions would be much conclusive if they should be sustained by comparative numerical examples.