Finite Strip Method in a Parallel Computing Environmentby J. A. Puckett, Univ of Wyoming, Laramie, WY, USA,
R. J. Schmidt, Univ of Wyoming, Laramie, WY, USA,
Document Type: Proceeding Paper
Part of: Parallel and Distributed Processing in Structural Engineering
The finite strip method has been used to model many systems in structural mechanics. The orthogonality of appropriately selected shape functions allows a continuous system to be effectively transformed into a set of smaller discrete subsystems that can be solved independently. The orthogonality of the shape functions and their derivatives simplifies the energy minimization procedures. The summation is removed from the integration and thus uncouples the resulting algebraic equations. This produces an intrinsically parallel algorithm that can be implemented efficiently in either a shared or a distributed memory environment. The algorithm for the assembly and solution of a single subsystem is quite similar to a traditional finite element analysis. The solution is repeated for each summation term in the shape function. The state variables are in turn used to recover the generalized stresses. This assembly and solution process can be executed concurrently for the various summation terms considered. The implementation of the orthogonal shape functions requires de-generalizing the standard finite element procedure in one direction.
Services: Buy this book/Buy this article
Return to search