Advances in Engineering Software - Special issue; special issue on large-scale analysis and design on high-performance computers and workstations
Reducing the bandwidth of sparse symmetric matrices
ACM '69 Proceedings of the 1969 24th national conference
| Hi-index | 0.00 |
Various composite structural configurations are increasingly manufactured using liquid composite molding processes. These processes provide unitized structures with repeatability and excellent dimensional tolerances. The new generation of materials and processes, however, pose significant challenges in affordable development and process optimization. Process modeling and simulations, based on physical models, play a significant role in understanding the process behavior and allow for optimizing the process variants in a virtual environment. Large scale composite structures require scalable process modeling and simulation software. An overview of various key issues for such scalable software developments are presented. Several items, including our experiences with the numerical techniques and algorithms for the solution of the representative physical models, are discussed. We also investigate and briefly discuss the various parallel programming models and software development complexities to achieve good performance. In this context, iterative solution techniques for large scale systems, based on a domain decomposition of the problem domain and non-invasive techniques to boost performance of scalable software, are discussed. Preliminary results of the performance of simulations in liquid composite molding are presented. The discussions presented, though in direct reference to scalable process modeling and simulations in liquid composite molding, are directly applicable to any unstructured finite element computations.