An extension of lognormal theory and its application to risk analysis models for new mining ventures

An extension of lognormal theory and its application to risk analysis models for new mining ventures by B. M. WAINSTEIN*, B.Sc., M.B.A. (Witwatersrand) (Visitor) The lognormal distribution, which remains the most popular distribution model for ore values, is investigated. The T-distribution, an integral part of this theory, is dependent on the sample size, n, and the unknown population parameter, o2, which cannot be integrated out. Hence, o2constitutes a nuisance parameter. In this study, the robustness of the T-distribution to changes in o2 is examined. It is found that the T-distribution is robust for 0,1 <= o2 >= 2,5, a domain covering most practical applications of the lognormal theory. By the use of o2 = 0,7, multiplier factors, which facilitate the computation of the confidence limits, are derived for a comprehensive set of confidence levels.