Predictive modeling simulation and control of large-scale commercial des alination plants

The dependency on desalination plants to produce water from the sea is increasingthroughout the world. This has created considerable interest in investigating new ways toimprove the design, control, and operation of desalination plants.This research involves the development and applications of quantitative models, dynamicsimulation and multivariable control for predicting and optimizing the operationalperformance of large-scale commercial desalination plants.Specifically, this research includes three parts. In part one, we develop a neural networkapproach for the prediction and optimization of process performance parameters of large-scale commercial desalination plants. In contrast to previous studies, this work utilizesactual operating data (not simulated data) from a Multistage Flash (MSF) DistillationPlant (48 Million Gallons Per Day, MGPD) and a Reverse Osmosis (RO) Plant (15MGPD) located in Kuwait, and Saudi Arabia, respectively.Our completed work has demonstrated the accuracy and efficiency of neural networkmodels in conjunction with engineering knowhow and statistical techniques to predict andoptirnize the operating variables of the processes. The neural network model alsooutperforms the statistical regression technique in accurately and efficiently predicting theperformance parameters of the plants. This work suggests that ANN’s are particularlyappropriate as the basis for the development of tools to aid in the various phases ofoperating a desalination plant.Part two of this research deals with the development of both steady-state and dynamicmodels for simulating the operational performance of a large-scale commercial multi stageflash (MSF) desalination plant. This development utilizes advanced commercial softwaretools, such as ASPEN (Advanced System for Process Engineering) PLUS and SPEEDUP(Simulation Program for Evaluation and Evolutionary Design of Unsteady Process) beingmarketed by Aspen Technology, Cambridge, MA.In part three, we apply the simulation models developed in part two and investigate theuse of multivariable feedbacklfeedforward control of large-scale commercial MSFdesalination plant. Out’ goal is to identify the best control approachefor wide rangesof setpoints and disturbances.To our knowledge, this research represents the first comprehensive study of predictivemodeling, simulation, and control of large-scale commercial desalination plants usingartificial intelligence techniques and advanced software tools.