Improving The Estimates Of Recoverable Reserves

Reconciliation of long-term, block-model reserves and mine production frequently shows significant discrepancies. This reconciliation can be done comparing the long-term, block-model reserves against a grade control model (a model based on blast holes or production data). In this article, this is called an Fl factor for tonnages, grade and ounces. This reconciliation should also be done by comparing the grade control model to the mill feed (or loaded to heaps) tonnages, grades and ounces (F2 factor). The long-term model usually contains blocks with dimensions larger than the size of the selective mining unit (SMU) of the operation. The smoothing effect of an estimator, such as kriging, will generally result in a grade-tonnage curve that does not match the grade-tonnage curve of the SMUs to be recovered. This problem is partly to blame for the reconciliation discrepancies found. In addition, in-pit perfect selection is impossible. Therefore, the grade-tonnage predictions based on blastholes also need to be assessed - considering unplanned dilution and the errors of estimation of the grade control model. This article focuses on using conditional simulations to assess the uncertainty of the predicted grade-tonnage curve in terms of minable reserves and in the framework of long-term mine planning. Quantifying the uncertainty in no uncertain terms is done in conjunction with a popular pit optimizer. It yields several possible minable reserves and several possible minable tonnages and grades, per pit shell or scheduled phase. This is a step further from the usual sensitivity analyses that are performed for deposit evaluation or development planning. Specifically, this approach takes into account the volume-variance relationship, the selectivity of the mining operation and the planned and unplanned dilution. It also carries the local uncertainty implicit in grade estimation into the mine design and mine planning phases. The estimation of recoverable reserves revolves around estimating the block distribution of metal grades from the drill hole information available, typically composites of a fixed length. Ideally, this estimation should accurately predict the ore sent to the processing facilities of an operation. Here are some definitions used in this article: ? Recoverable reserves are those where an attempt has been made to estimate the actual distribution of grades (usually by a block model) that would be found in the ground at the time of mining, relative to the mine equipment size and other operational characteristics of the mine SMU. In particular, it incorporates the effect of internal dilution for each block how the grade distribution changes in variability when comparing a composite grade distribution to the "mining unit" grade distribution. In geostatistical jargon, this is known as "change of support effect," or volume-variance effect. ? Minable reserves are those recoverable reserves defined by a mine plan. In an open-pit scenario, they would incorporate pit optimization in addition to the sequencing and scheduling of production through mine life. ? All other block model estimates from drill hole data that do not explicitly attempt to estimate or measure the volume-variance effect are called in situ. It can be argued that these models may be inappropriate tools for predicting the grade-tonnage curve of the material to be sent for processing (minable reserves). The purpose of estimating recoverable reserves can be thought of as estimating the appropriate amount of dilution that should be incorporated into the model so that it adequately predicts (long-term and on average) the grade-tonnage curve of the ore sent to the processing facilities (mean grades and ton-