Parallel processing of central difference transient analysis for three-dimensional nonlinear framed structures

The central difference algorithm is a simple but effective time-integration algorithm suitable for parallel processing. This paper discusses the development and implementation of the method for the fully nonlinear, transient dynamic analysis of three-dimensional framed structures. A data structure is presented in which only one nearest-neighbor communication is required per time step, and potentially little overhead is incurred for the parallel implementation. Because this explicit method is only conditionally stable, the large disparity of natural vibrational frequencies of framed structures indicates that central difference analysis is more useful for applications having short-duration loading. An example is presented indicating the speed-up and efficiency of the algorithm when implemented in a distributed-memory, coarse-grained parallel processing environment.