Daniela Hossu, Ioana Fagara§an, Andrei Hossu and Sergiu Stelian Iliescu

University Politehnica of Bucharest, Faculty of Control and Computers

Romania

1. Introduction

Poor control of steam generator water level is the main cause of unexpected shutdowns in nuclear power plants. Such shutdowns are caused by violation of safety limits on the water level and are common at low operating power where the plant exhibits strong non­minimum phase characteristics. In addition, the steam generator is a highly complex, nonlinear and time-varying system and its parameters vary with operating conditions. Therefore, there is a need to systematically investigate the problem of controlling the water level in the steam generator in order to prevent reactor shutdowns.

Difficulties on designing a steam generator (SG) level controller arise from the following factors:

— nonlinear plant characteristics. The plant dynamics are highly nonlinear. This is reflected by the fact that the linearized plant model shows significant variation with operating power.

— nonminimum-phase plant characteristics. The plant exhibits strong inverse response behavior, particularly at low operating power due to the so-called "swell and shrink’ effects.

— dynamics uncertainties,

— corrupted feed-water flow measurement signal with biased noises.

At low loads (less than 15% of full power) the non-minimum phase behavior is much more pronounced.

Various approaches have been reported in the literature: an adaptive PID level controller using a linear parameter varying model to describe the process dynamics over the entire operating power range (Irving et al. 1980); a model of the steam generator water level process in the form of a transfer function, determined based on first-principles analysis and expert experience has been presented in (Zhao et al., 2000); LQG controllers with "gain­scheduling" to cover the entire operating range (Menon & Parlos, 1992); a hybrid fuzzy-PI adaptive control of drum level, a model predictive controller to identify the operating point at each sampling time and use the plant model corresponding to this operating point as the prediction model (Kothare et al., 2000). Paper (Park & Seong, 1997) presents a self organizing fuzzy logic controller for the water level control of a steam generator. A

nonlinear physical model with a complexity that is suitable for model-based control has been presented by Astrom and Bell (Astrom & Bell, 2000). The model describes the behavior of the system over a wide operating range.

With the advent of the current generation of high-speed computers, more advanced control strategies not limited to PI/PID, can be applied (Hirota, 1993), (Pedrycz & Gomide, 2007), (Yen et al., 1995), (Ross, 2004).

Model predictive control (MPC) design technique has gained wide acceptance in process control applications. Model predictive control has three basic steps: output prediction, control calculation and closing the feedback loop (Camacho & Bordons, 2004), (Demircioglu & Karasu, 2000), (Morari & Lee, 1999).

In this chapter, we apply MPC techniques to develop a framework for systematically addressing the various issues in the SG level control problem.

Fuzzy models have become one of the most well established approaches to non-linear system modeling since they are universal approximations which can deal with both quantitative and qualitative (linguistic) forms of information (Dubois & Prade, 1997), (Zadeh, 2005), (Zadeh, 1989) This chapter deals with Takagi-Sugeno (T-S) fuzzy models because this type of model provides efficient and computationally attractive solutions to a wide range of modeling problems capable to approximate nonlinear dynamics, multiple operating modes and significant parameter and structure variations (Kiriakidis, 1999), (Yager & Zadeh, 1992), (Ying, 2000). Takagi-Sugeno (T-S) fuzzy models have a good capability for prediction and can be easily used to design model-based predictive controllers for nonlinear systems (Espinosa et al., 1999).

The objective of this chapter is to design, evaluate and implement a water level controller for steam generators based on a fuzzy model predictive control approach. The chapter includes simulations of typical operating transients in the SG. A new concept of modular advanced control system designed for a seamless and gradual integration into the target systems is presented. The system is designed in such a way to improve the quality of monitoring and control of the whole system. The project targets the large scale distributed advanced control systems with optimum granularity architecture.