Self-Adaptive Control of Integrated Ultrafiltration and Reverse Osmosis Desalination Systems
- Author(s): Gao, Larry Xingming;
- Advisor(s): Cohen, Yoram;
- Christofides, Panagiotis D
- et al.
Water shortages in many areas of the world have increased the need for fresh water production through water desalination in applications such as the production of potable water, use in agricultural irrigation, and wastewater reuse. In this regard, reverse osmosis (RO) membrane desalination of both seawater and inland brackish water has emerged as the leading technology for water desalination, with a growing number of large-scale desalination plants in the planning and/or construction stages.
Currently, the design of a water desalination plant is typically tailored to the specific water source in terms of meeting productivity targets and pre-treatment requirements. The standard operating procedure is to determine one optimal operating state for an RO system (e.g., overall water recovery, membrane cleaning frequency) and maintain this specific operating point for the duration of operation. However, these methods do not adequately account for the variability in feed water salinity and fouling propensity, and may result in suboptimal operation with respect to excessive energy consumption, poor RO feed pre-treatment, and degradation of RO membrane performance. Therefore, it is crucial to develop effective process control approaches which can mitigate membrane fouling and reduce RO energy consumption in order to improve the robustness of the RO desalination process.
In order to reduce membrane fouling, several concepts which involve improvements to RO plant pre-filtration capability (e.g., the addition of a separate, modular ultrafiltration membrane process, the use of a transient high-flux “pulse” backwash) were developed. The concept of direct integration of ultrafiltration (UF) and RO was introduced, whereby the UF filtrate is fed directly to the RO and the RO concentrate is used for UF backwash. Additionally, a control system was designed for the UF pre-treatment unit whereby membrane fouling was reduced through optimization of backwash through a combination of varying the backwash frequency and varying the coagulant dose. This approach was shown to significantly reduce membrane fouling and significantly increased operation duration before chemical cleaning was required (~900% longer).
In order to reduce energy consumption of RO desalination, energy-optimal control systems featuring a novel two-layered controller architecture were developed and implemented using fundamental models of specific energy consumption (SEC) of single-stage and two-stage RO systems. The implemented control algorithms utilized extensive sensor measurements from the pilot plants (i.e., flow rate, pressure, conductivity, etc.) to determine the optimal operating set-points for the RO systems (e.g., system feed flow rate, system feed pressure, and overall system water recovery). Accordingly, the control system shifted the RO system operation to the operating conditions that resulted in the lowest energy consumption for a given feed salinity and for a given target product water productivity while accounting for system constraints.
The control and design concepts developed in this dissertation were tested on two water purification systems, constructed by a team at UCLA. The two pilot plants were the Smart Integrated Membrane System – Seawater (SIMS-SW) and the Smart Integrated Membrane System – Brackish Water (SIMS-BW). Field tests of the control systems were conducted and the results successfully demonstrated the ability for the control systems presented in this dissertation to reduce membrane fouling and RO energy consumption.