Strategies for Using Advanced Computer Architectures in Chemical Process Engineering

先进的计算机体系结构，特别是那些涉及某种形式的并行和/或矢量处理的计算机体系结构，提供了极大地提高工程师实际建模、模拟、设计和优化复杂化学制造过程的能力，包括稳态和非稳态。这个项目的目标是开发解决这些问题的技术，充分利用这种增强的处理。这项工作的一个重点是开发在矢量/并行计算机上有效解决这些系统的技术。所考虑的技术是基于额直接法和多额直接法，以及具有不完全分解前置条件的迭代Krylov子空间方法。这个项目的第二个重点是使用并行性来保持过程模拟和优化计算的鲁棒性。将应用基于区间分析的技术。由于区间方法适用于并行计算，因此它们在这些问题上的应用现在是可行的。先进的计算机体系结构在化工过程工程中的应用具有很大的潜力。矢量/并行机器提供的快速周转可以提高工程生产率，这意味着更多的关键截止日期，更快的上市时间，以及对业务、经济和监管条件变化的更及时的响应。使用更复杂和更现实的过程模型的能力意味着更好地理解过程基础，从而减少过度设计的需要，并提供更好地实现环境和质量控制目标所需的见解。

9322682 Stadtherr Advanced computer architectures, in particular those involving some form of parallel and/or vector processing, provide the power to greatly enhance the ability of the engineer to realistically model, simulate, design, and optimize complex chemical manufacturing processes, steady- and unsteady-state. The goal of this project is to develop techniques for solving such problems, taking full advantage of this enhanced processing. One focus of this work is the development of techniques for efficiently solving these systems on vector/parallel computers. The techniques to be considered are based on the frontal and multifrontal direct methods, and on iterative Krylov subspace methods with incomplete factorization preconditioners. A second focus for this project is the use of parallelism to maintain robustness in process simulation and optimization computations. Techniques based on interval analysis will be applied. Since interval methods are amenable to parallel computing, their use on these problems is now feasible. The application of advanced computer architectures in chemical process engineering is one of great potential. The rapid turnaround that can be provided by vector/parallel machines improves engineering productivity, meaning more critical deadlines met, faster times to market, and more timely responses to changes in business, economic, and regulatory conditions. The ability to use more complex and realistic process models means a better understanding of process fundamentals, thus reducing the need for overdesign, and providing the insights needed to better attain environmental and quality control goals.