Evaluating the geological risk associated with ore deposits is becoming common practice in the mining industry._x000D_
While geostatistical simulation is a useful tool for evaluating this risk, the effect of applying different types of simulation is less understood._x000D_
For simulation of categorical variables like rock types, available methods perform simulation in fundamentally different ways._x000D_
In this study, the geological risk relevant for informed mine planning is studied with two categorical simulation methods, PluriGaussian Simulation (PGS) and Sequential Indicator Simulation (SIS)._x000D_
Risk is expressed as a probability of repeated occurrence of lithological units in the same location across simulations._x000D_
Significant differences are observed between the methods in terms of spatial distribution of risk._x000D_
To identify the most plausible risk model, the input parameters at every step of both techniques are scrutinised. PGS and SIS inputs are classified as four parameters that are: fundamental to implementation of the method fixed to enable comparison of PGS and SIS determined by user interpretation, hence variable normally optimised during application of PGS and SIS. The effect of parameter selection from categories 1 and 4 on the risk model is investigated using drill core data from a leucogranite deposit. A corresponding block model is created for assigning outputs from PGS and SIS. Risk models produced by either method are illustrated in a cut-out from the block model to enable bench-by-bench comparison. For PGS, risk associated with relatively rare lithologies is relatively sensitive to variation of parameters. Independent of the abundance of lithology, a location-dependent effect of variogram and vertical proportion curve parameter selection is observed, with risk increasing towards the deposit edge. Although SIS requires fewer inputs, single lithologies must be modelled in turn. Using small numbers of samples in the search neighbourhood, all risk models differ substantially. For larger numbers of samples, the risk models are similar. Overall it appears that the risk model produced using SIS is less sensitive to changes than PGS but further work is required to confirm this.CITATION:Palmer, L W and Glass, H J, 2014. Risk associated with rock type prediction using simulation techniques, in Proceedings Ninth International Mining Geology Conference 2014 , pp 217-228 (The Australasian Institute of Mining and Metallurgy: Melbourne).