Development and application of self-optimising control to coal-fired steam-generating plant

Abstract

The objects of the experimental programme, which is the subject of the paper, were to assess the technical feasibility and economic return that could be expected from applying an automatic self-optimising control system to coal-fired steam-generating plant. The work centred on the control of combustion, this being directly exposed to disturbances arising from variability of the fuel quality. A hill-climbing control system was used which maximised combustion efficiency by manipulating the air/fuel ratio. As efficiency could not be measured directly, an inferential method based on changes in boiler pressure or turbine output was adopted. Two control strategies, utilising, respectively, sinusoidal and step perturbations of air flow, were investigated. The paper contains estimates of the disturbance spectrum, based on measurements of both coal-quality and boiler-output fluctuations, and shows this to be equivalent to the addition of coloured noise to the system input. The methods of obtaining optimum discrimination against this noise for each of the two strategies are discussed. Experimental data obtained by using the control system to track movements of the peak of the combustion hill are given, a range of 2% excess oxygen being observed. It is concluded that, although the factors governing this movement have not been fully identified, the fineness of grinding is of great importance in determining the optimum combustion conditions. It is also concluded that this form of control is feasible and could lead to an improvement in efficiency of about 0.25%.