Modeling Stiffeners

One common use of beams is as stiffeners in shell structures. Ship hulls and aircraft fuselages, for example, are often built using this technique. An important property here is the offset between the shell midsurface and the centerline of the attached beam. To a large extent, it is the tension of the beam rather than its bending that provides the high stiffness. In order to model this effect, you can use two different approaches:

•

Model the beam centerline and shell midsurface at their correct locations. Make sure that there are edges in the shell geometry placed as imprints under the stiffeners. The two sets of edges are then connected using suitable couplings.

•

Draw only the “imprint” on the shell, and select these edges in the Beam interface. Connect the degrees of freedom in the two interfaces with a coupling that mathematically takes the offset into account.

In both cases, you use the Solid-Beam Connection multiphysics coupling, but with different settings. From the point of solution accuracy, the methods are equivalent. The latter method is more convenient, but it has the drawback that the beams will be visualized as being located in the plane of the shell.


	Both types of connections are shown in the example Connecting Shells and Beams: Application Library path Structural_Mechanics_Module/Beams_and_Shells/shell_beam_connection

Since the beam elements use special shape functions, a connection to another element type like a shell will not be conforming. Consider a beam attached to a shell element having the default discretization order (quadratic). There will then be two beam elements connected to each edge of a shell element. All displacements and rotations in the shell element are assumed to have a parabolic distribution. In the beam, however, the axial displacements are linear, while the transverse displacements are represented by cubic polynomials. An effect of this is that there will be local aberrations in the forces transferred between the shell and the beam, even though the solution is correct in an average sense. Because of this, you may see local stress fluctuations.