Large-grained parallelism in equation-based flowsheeting using interval Newton/generalized bisection techniques
Schnepper, Carol Ann
This item is only available for download by members of the University of Illinois community. Students, faculty, and staff at the U of I may log in with your NetID and password to view the item. If you are trying to access an Illinois-restricted dissertation or thesis, you can request a copy through your library's Inter-Library Loan office or purchase a copy directly from ProQuest.
Permalink
https://hdl.handle.net/2142/19671
Description
Title
Large-grained parallelism in equation-based flowsheeting using interval Newton/generalized bisection techniques
Author(s)
Schnepper, Carol Ann
Issue Date
1992
Doctoral Committee Chair(s)
Stadtherr, Mark A.
Department of Study
Chemical and Biomolecular Engineering
Discipline
Chemical Engineering
Degree Granting Institution
University of Illinois at Urbana-Champaign
Degree Name
Ph.D.
Degree Level
Dissertation
Keyword(s)
Engineering, Chemical
Language
eng
Abstract
The overall objective of this work has been to apply high-performance, parallel computers to equation-based chemical process flowsheeting. The primary focus was on investigating parallel techniques for improving the reliability of equation-based flowsheeting, and the secondary goal was to examine the effects of parallel function evaluation on the performance of traditional, Newton-like solvers.
To accomplish the principal goal, an interval Newton/generalized bisection technique was studied. This approach to solving systems of nonlinear equations often involves high computational cost and therefore has been considered impractical for solving the large, sparse systems associated with equation-based flowsheeting. However, advances in parallel computer architectures and increases in computational power indicate that efficient implementations will become practical for solving flowsheeting problems. In the first stage of this study, SQUIB, a serial code using an interval Newton/generalized bisection algorithm for solving large, sparse flowsheeting systems, was developed. Then two large grained parallel algorithms, one synchronous and the other asynchronous, based on the serial algorithm were developed and implemented with the Uniform System programming model on a BBN TC2000 high-performance computer. All of the programs were used to solve several small flowsheeting problems. The serial code usually successfully located solutions in well under the number of computations predicted by the worst-case scenario, and it also isolated multiple roots. The asynchronous parallel code outperformed the synchronous code, and the results from running the asynchronous code on increasing numbers of processors indicate that maximum speedups are limited to about four or five. However, if additional levels of parallelism are exploited in future research, the resulting speedups should improve as the number of processors increases. These methods thus hold promise for solving flowsheeting problems when other, less computationally expensive, approaches have failed.
To achieve the secondary goal, the function evaluations required for the Newton-like solvers in a prototype equation-based flowsheeting package were computed in parallel on a Cray-2. The results indicate that this strategy could improve performance for very large problems or for problems involving several computationally intensive functions.
Use this login method if you
don't
have an
@illinois.edu
email address.
(Oops, I do have one)
IDEALS migrated to a new platform on June 23, 2022. If you created
your account prior to this date, you will have to reset your password
using the forgot-password link below.
