Conditional Simulation is used in the mining industry to quantify uncertainty and to minimize risk. Whereas an estimator reproduces average properties, simulation reproduces variability. The potential errors in any estimate, both the upside and downside, are functions of the variability and are critical factors in the decision-making process.

Point or block estimates, such as those produced by kriging or inverse power of distance, are not sufficient to answer all the questions that are raised regarding grade, tonnes and metal content.

For example;

• What is the probability that the reserve grade will exceed a particular threshold?
• What is the likely daily variation in mill head grade?
• What is the impact of changing the bench height or changing the sampling density
• What are the best and worst case scenarios and how do they affect the mine plan?

These are just some of the questions that can be answered using Conditional Simulation.

Conditional Simulation is an important part of Risk Management, and although the subject is more than 30 years old, it has taken a long time to become available for general use rather than just for highly qualified consultants or academics. The main reasons for this have been:

• The absence of suitable software that doesn’t require a degree in computing and geostatistics to run.
• It’s all very well producing a matrix of conditionally simulated points, but they are not much use to anyone. What is also needed is easy-to-use software to analyze and interpret the results.
• Even with the speed of computers today, simulation can be a time-consuming process; one needs powerful and targeted software to enable decisions to be made in a timely manner.

The Datamine Solution

The most widely used algorithms for Conditional Simulation are presented by Deutsch and Journel in their book, GSLIB Geostatistical Software Library and User’s Guide. There are a variety of simulation methods, with the most popular one being Sequential Gaussian Simulation which has been incorporated into the Datamine SGSIM simulation process. Two other GSLIB routines for transforming to and from Gaussian distributions have also been included.

A simulation study involves creating many equi-probable realizations, which are conditioned to the local data, and then analyzing the variability of the results. Datamine Studio provides a powerful set of tools for interpreting the results of the simulation. In addition to the standard data transformation and manipulation commands, two new processes have been added which deal specifically with the results of the simulation. One provides a detailed statistical analysis of the conditional distributions of the simulated model cells allowing confidence limits to be assigned, and the other optimizes ore and waste block selection in a model by minimizing the loss due to misclassification.

Conditional Simulation Study

The Conditional Simulation module combined with the existing statistical, geostatistical and data manipulation processes in Datamine Studio provide an ideal environment for any conditional simulation study. As well as producing invaluable results in its own right, the output can be easily fed into the Mineable Reserves Optimizer or NPV Scheduler for further investigation and analysis.