Optimal Control of Coal Segregation using On-line Analyzers

R. Ganguli, J. C. Yingling, J. Sottile, Dept. of Mining Eng. and R. Kumar, Dept. of Electrical Eng. University of Kentucky, Lexington, KY 40506 ABSTRACT The capability for on-line measurement of the quality characteristics of conveyed coal now enables mine operators to take advantage of the inherent heterogeneity of those streams and split them into wash and no-wash stocks. Relative to processing the entire stream, this reduces the amount of coal that must be washed at the mine and thereby reduces processing costs, recovery losses, and refuse generation levels. In this paper, new approaches for optimal control of the segregation process, using time series models and empirical distributions, are presented and demonstrated using field data. INTRODUCTION Coal washing is one of the major cost components of the coal production process. Moreover, because of the imperfect nature of the coal/ash separation, it is also a major source of yield losses. To avoid washing large volumes of coal, mine operators like to physically segregate run-of-mine (r.o.m.) coal into stocks of different quality such as a no-wash stock (typically low ash) or a wash stock (typically high ash). Advancements in on-line quality analyzers have made it easier to segregate r.o.m. coal. On-line analyzers measure the quality in real time, and a typical segregation strategy is that if the percentage ash is above a certain cut-off level or grade, it is sent to the wash pile. Otherwise, it is sent to the no-wash pile. The cut-off levels are based on the customer specifications and are often determined by trial and error. A hardware arrangement for effecting such segregation is shown in Figure 1. Since the coal in the no-wash pile is usually sent directly to the customer, it is imperative that, in aggregate, it meets quality specifications.