(444c) Fault Detection And Control Of A Reverse Osmosis Desalination Process

System automation and reliability are crucial components of any modern reverse osmosis (RO) plant [1]. The operational priorities are personnel and product water safety, while also meeting environmental and economic demands, and power input constraints [3]. Automated RO plants, however, can be vulnerable to faults in several process components that can effect plant operation. Examples of faults can include valve failure, membrane fouling or scaling, sensor data loss, and pump or variable frequency drive failure. Because RO plants run at high pressures, these failures may cause immediate safety risks to plant personnel. These failures can also lead to a decline in the production of quality water, rendering it unsafe for public consumption or leading to shortages. These safety issues provide strong motivation for the development of fault-detection and isolation (FDI) and fault-tolerant control (FTC) structures that can quickly identify failed actuators and make effective decisions to maintain safe plant operation as developed in [4], [2].

A detailed mathematical model has been developed on the basis of a physical RO system being designed and constructed at UCLA. The manipulated inputs for this system include actuated control of the pump motor, and two actuated control valves. The measured outputs of this system include pressure, conductivity, and flow velocity at several points in the system. In this work the mathematical model is utilized in simulations to explore FDI and FTC control structures that may be implemented on the physical system when it is complete. A family of control configurations that will provide adequate and independent control of internal pressure and membrane cross flow have been identified, where some of these configurations will act as fall-backs in the case of a failure.

A fault-detection and isolation filter has been developed for the RO system. This FDI filter incorporates real-time measurements and uses the process model to provide instantaneous information about faulty actuators [4]. A supervisory switching law has been derived that guarantees closed-loop stability by determining the activation of fall-back control configurations in the presence of faults in the primary control configuration. The proposed fault-tolerant control scheme is demonstrated in the context of several reverse osmosis system simulations. The simulation examples demonstrate the effectiveness of the proposed methodology considering practical issues such as variation in feed-water salinity, noise on the measurements, sample-and-hold control action, model parameter errors, and controller actuator / measurement sensor dead-time.

[1] I. M. Alatiqi, A. H. Ghabris and S. Ebrahim. System identification and control of reverse osmosis desalination. Desalination, 75:119-140, 1989.

[2] N. H. El-Farra and P. D. Christofides. Bounded robust control of con- strained multivariable nonlinear processes. Chem. Eng. Sci., 58:3025- 3047, 2003.

[3] C. K. Liu, Jae-Woo Park, R. Migita and G. Qin. Experiments of a prototype wind-driven reverse osmosis desalination system with feedback control. Desalination, 150:277-287, 2002.

[4] P. Mhaskar, C. McFall, A. Gani, P. D. Christofides, and J. F. Davis. Isolation and Handling of Actuator Faults in Nonlinear Systems. Automatica, regular paper in press, 2007.