(438f) Modeling and Control of An Experimental Reverse Osmosis Water Desalination System

Authors: 
Bartman, A., University of California Los Angeles
Cohen, Y., UCLA


Reverse osmosis (RO) membrane desalination has emerged as one of the leading methods for water desalination due to the low cost and energy efficiency of the process. Lack of fresh water sources has necessitated further development of these desalination plants, especially in arid climates. Despite advances in reverse osmosis membrane technology, there remains a challenge of maintaining robust process operation when systems experience fluctuations in water feed quality. Seasonal, monthly, or even daily changes in feed water quality can drastically alter the conditions in the reverse osmosis membrane modules, leading to decreased water production, sub-optimal system performance, or even permanent membrane damage. In order to account for the variability of feed water quality, a robust process control strategy is necessary. Traditional control algorithms for RO desalting do not take the variability of feed water quality into account, thus raising the potential for the crippling impact of mineral scaling, fouling, and eventual plant shutdown. Traditional process control schemes are also unable to monitor plant energy usage and make dynamic adjustments toward energy-optimal operation. Model based control is a promising alternative to traditional RO plant control strategies in that it is able to account for the simultaneous effect of many system variables (feed quality, system pressure, stream flow rates, etc.). The goal of this work is to develop and evaluate a feedback-based nonlinear model-based controller through experience with an experimental reverse osmosis desalination system. The RO desalination model was first derived based on the relevant mass and energy balances. Mode parameters were then computed based on experimental data gathered from the experimental RO desalination system in order to minimize the error between model and experimental system responses. The resulting dynamic nonlinear model was then used to derive a nonlinear feedback linearizing controller to conduct set-point transitions of the retentate flow rate. Accurate control of the retentate flow rate and system pressure is integral to effective reverse osmosis system operation. It is shown that the nonlinear controller is well suited for dealing with the highly coupled system dynamics during set-point transitions. This control approach was shown to outperform the traditional (proportional and proportional-integral) control schemes. The model-based nonlinear controller also is shown to perform well when the reverse osmosis system operates in a cyclic feed flow-reversal mode. This mode of operation, which is accompanied with large step changes in feed salt concentration, was shown to be effective in mineral scale mitigation.