This dissertation introduces desalination processes in general and multistage flash (MSF) and reverse osmosis (RO) in particular. It presents the fundamental and practical aspects of neural networks and provides an overview of their structures, topology, strengths, and limitations. This study includes the neural network applications to prediction problems of large-scale commercial MSF and RO desalination plants in conjunction with statistical techniques to identify the major independent variables to optimize the process performance.
In contrast to several recent studies, this work utilizes actual operating data (not simulated) from a large-scale commercial MSF desalination plant (48 million gallonsper day capacity, MGPD) and RO plant (15 MGPD) located in Kuwait and the Kingdom of Saudi Arabia, respectively. We apply Neural Works Professional II/Plus (NeuralWare, 1993) and SAS (SAS Institute Inc., 1996) software to accomplish this task.
This dissertation demonstrates how to apply modular and equation-solving approaches for steady-state and dynamic simulations of large-scale commercial MSF desalination plants using ASPEN PLUS (Advanced System for Process Engineering PLUS) and SPEEDUP (Simulation Program for Evaluation and Evolutionary Design of Unsteady Processes) marketed by Aspen Technology, Cambridge, MA.
This work illustrates the development of an optimal operating envelope for achieving a stable operation of a commercial MSF desalination plant using the SPEEDUP model. We then discuss model linearization around nominal operating conditions and arrive at pairing schemes for manipulated and controlled variables by interaction analysis. Finally, this dissertation describes our experience in applying a commercial software, DynaPLUS, for combined steady-state and dynamic simulations of a commercial MSF desalination plant.
This dissertation is unique and significant in that it reports the first comprehensive study of predictive modeling, simulation, and optimization of large-scale commercial desalination plants. It is the first detailed and comparative study of commercial desalination plants using both artificial intelligence and computer-aided design techniques. The resulting models are able to reproduce accurately the actual operating data and to predict the optimal operating conditions of commercial desalination plants. / Ph. D.
| Identifer | oai:union.ndltd.org:VTETD/oai:vtechworks.lib.vt.edu:10919/30462 |
| Date | 29 April 1998 |
| Creators | Al-Shayji, Khawla Abdul Mohsen |
| Contributors | Chemical Engineering, Liu, Y. A., Conger, William L., Velander, William H., VanLandingham, Hugh F., Sherali, Hanif D. |
| Publisher | Virginia Tech |
| Source Sets | Virginia Tech Theses and Dissertation |
| Detected Language | English |
| Type | Dissertation |
| Format | application/pdf |
| Rights | In Copyright, http://rightsstatements.org/vocab/InC/1.0/ |
| Relation | ETD.PDF |
Page generated in 0.0026 seconds