N.W. Rees

Models of Large Boiler-Turbine Plant

Summary During the next decade the demand on power system generating units will increase as loading patterns require more and more flexible operation. To meet this need for improved performance a much better knowledge of the dynamics of the units will be needed both at the design stage and for control studies. In this paper models of a 500MW unit of the Electricity Commission of N.S.W. at their Liddell Power Station are developed. To obtain these models, data has been collected using a Leeds & Northrup LN5200 digital computer logging 124 variables around the plant at various sampling rates (6, 12, 30 secs.). The logged variables have been chosen to allow mass, energy, and momentum balances to be written for each of 11 subsections that make up the plant. Two types of models have been developed. The first is a nonlinear model requiring only steady-state operating and design data to fix parameters and the second is a linear statistical model with parameters estimated from dynamic data obtained from field testing with the data collected when the plant is operating under closed loop conditions. The advantages and disadvantages of both models is discussed including ease of producing the model, usefulness for design, usefulness for control and so on. The transient responses are compared with the actual plant operating data for both small and large disturbances so that the simulation capabilities can be judged.

Fuzzy Logic Control of Vertical Spindle Mills

Abstract In this paper we develop a new systematic procedure for controlling non linear plant based on fuzzy modelling and control. The fuzzy controller design is based on local linear models smoothly connected by fuzzy membership functions all of which can be determined directly from logged data. The practical value of the method is tested on the simulation of a vertical spindle coal mill

Practical Optimal Predictive Control of Power Plant Coal Mills

Abstract A procedure is described for the computation of optimal receding horizon control sequence for a power plant coal mill whose dynamic behavior is described by a set of nonlinear differential equations. The derivation of the mill model equations and the control algorithm are described briefly. The derivation of the control sequence algorithm is based on the local linearisation of the plant describing model and its approximation by a series expansion in terms of impulse response kernels. A series of simulation test results are presented which demonstrate the effectiveness of the technique.

Modelling and Identification of a 500MW Drum-Type Boiler-Turbine System

Abstract This paper is concerned both with the development of a mathematical model of a 500MW boiler-turbine system, using statistical identification techniques, and with the more general problem of developing a methodology for the statistical identification of a large scale MIMO industrial plant from input/ output data. The work described is carried out using a MIMO modelling and identification program (SIPAC) developed in-house and consisting of data conditioning routines, correlation analysis routines and systems fitting procedure which is an extension of the Zellner and Theil 3SLS estimator. Validation of the resulting model is carried out using residual analysis and model simulation. The C.A.D. package will be described and the boiler-turbine modelling discussed in detail.

MODELLING, SIMULATION AND CONTROL OF BOILER-TURBINE PLANT

Abstract This paper describes a microprocessor-based simulator of a Boiler-Turbine Plant, and control studies that have been carried out on the simulator. The simulator adequately represents the major plant dynamics and dynamic interactions and special attention has been paid to modelling the existing plant automatic control systems. Speeds of simulation of up to 20 x real time have been achieved, and operation of the system is simply obtained from a computer keyboard. The major simulator output is a graphic display of up to seven system variables against time.

Muscle Modelling and Force Estimation

Abstract Studies of human gait and the control of prothetic devices both require knowledge of the forces and torques developed by the muscles. Physiologically-based muscle models are extremely complex and are still poorly understood, so it is not uncommon to find statistically-based models being used by researchers. Such models typically attempt to relate muscle force with the electrical signals associated with muscle activity (the EMG). In this paper we examine some of the problems associated with force-EMG modelling using statistical methods and we show some results obtained from data collected using a special rig designed to study ankle movement. Some properties of the EMG signal and the force signal are discussed in terms of an ARMA analysis and it is suggested that modifications to the standard linear analysis may be needed to cope with the essentially nonlinear EMG-force relationship, underlying the data.

Optimal Co-ordinated Control of Boiler Turbine Plant

Abstract Using mathematical models developed from real plant data, optimal control schemes for Boiler-Turbine plant are developed and their performance compared with existing boiler-follow control. Advantages and disadvantages of the schemes are discussed.

Identification of a Boiling Channel and Comparisons with a Theoretical Model for Coolant Flow Stability Analysis

Abstract Traditional methods used to develop models for system identification and control studies invariably yield models which give little understanding of the physical structure of the process, and are therefore of limited value to the process design engineer. This paper demonstrates how noise analysis methods together with a comprehensive theoretical model of a boiling channel may be used to give a more detailed understanding of the channel dynamics and its coolant flow stability. Results are presented from an experimental and theoretical study carried out on a boiling channel situated in a small scale test rig.

A Knowledge Based Operator Support/Control System (KBOSS) for Coal Mills Used in Power Plants

Abstract In this paper, we describe the development of a knowledge based mill operator support/control system for coal pulverising mills. The work described is being performed as part of a research project. The target of the project is to develop a knowledge based expert mill operator support/control system (KBOSS) which is not only able to on-line monitor the process and provide the operator with the necessary information and advise for mill control, but is also able to directly intervene in the mill control thus running the mill and the overall power generation plant more efficiently and economically

Modular Simulation System

Abstract It is now accepted by much of industry that simulation is a valuable method for improved plant understanding and control. This paper describes a Modular Simulation System which runs in UNIX using the C language and supplies a user friendly software in which the user does the modelling and simulation. This software the simulation SHELL and the slructured simulation program, SSIM, allows the user to build up a simulation from a library of submodules and each submodule can be reused as often as required There is also a limited degree of run tune reconfiguration. The SHELL incorporates a graphic display that extensively uses multiple windows and pull- down menus and mouse-assisted user input. The Simulator has been tested on a number of applications, including power plant, sewerage works and steel rolling. Its use here will be described on a 3rd order system, and an outline will be given of its use for simulating a 660MW boiler.