Bauxite Mine Abstract
Bauxite is mined and transported by conveyor to a processing plant, screened and washed, then placed into blended stockpiles to feed the alumina refinery. While being stacked to the stockpile, the ore is sampled. Completed stockpiles must be acceptably close to target grade (composition), not only in alumina, but also in residual silica, carbon and sodium carbonate.
The mine is an open-cut pit. Each day the choice of ore to mine, from multiple locations in the pit, is based upon estimates of grade. Estimated grade, from exploration drilling of the area before mining , has both systematic and random error. This paper describes an information system to guide the daily choice of ore to mine. Continually updating the comparison between forecasts and sampled product, the system provides adjusted forecasts. Ore is selected to bring the exponentially smoothed grade to target, in each of the control minerals.
Bauxite Mine Introduction
Bauxite is the raw material for alumina and aluminium production. In 2004, 56.6 million tonnes of bauxite were mined in Australia, the world’s largest producer. The Australian aluminium industry directly employs over 16,000 people. (Australian Bureau of Statistics, 2006). The International Aluminium Institute, (2000) provides a summary description of the bauxite mining and refining processes.
Figure 1 shows the flow of ore from mine to refinery. The ore is mined from an open-cut pit, from the various pit faces available at the time. For each 12-hour period, the mine planner must decide the amount of ore to be selected from each available source. Each available source has an estimated grade vector in a range of minerals, of which the four most important are alumina, reactive silica, carbon and oxalate. The grade estimate for each source is the “mine forecast”, based upon sampling from exploratory drill holes before mining.
The mined ore travels over 50 km on a conveyor belt to be crushed and sampled before being used to build stockpiles, typically of 200kt. The stockpiles are built by “chevron” stacking back and forth, and reclaimed from one end in such a way that each stockpile can be assumed to be of uniform grade.
The refinery uses the Bayer process (Habashi, 2005) to convert the bauxite ore to pure alumina. Efficient operation of the refinery depends upon the bauxite feed having consistent grade. It is therefore important that the stockpiles from which the refinery feed is drawn each have grade within a tight range of target, in each of the four minerals. To achieve the required stockpile grade, the ore selected each day must be chosen appropriately. This is a difficult task, because the ore grade is not known accurately until the ore has been mined, crushed, sampled and assayed. Before mining, the grade of each available source must be forecast from the assays of samples previously taken from quite widely spaced exploration drill holes. An adjusted forecast is calculated, by comparing the recent forecasts with corresponding production sample assays.
Selecting ore to bring each individual stockpile to target requires very tight operational coupling between the mining operation and the building of stockpiles, to ensure that material designed for a particular stockpile goes to that stockpile. The improved system to be described here is operationally decoupled by instead selecting ore each shift so as to bring the exponentially smoothed grade back to target. The Continuous Stockpile Management System (CSMS) being applied was originally developed for iron ore production (Everett, 1996, 2001; Kamperman, Howard & Everett, 2002).
The CSMS achieves operational decoupling but tight information coupling between mining and stockpile building. In selecting ore to be mined, it is chosen to have adjusted forecast grade such that the exponentially smoothed continuous stockpile is as close as possible to target grade. As the ore is mined, the smoothed stockpile grade is updated to incorporate the adjusted forecast grade of the newly
mined ore. When that ore is sampled and assayed, the exponentially smoothed continuous stockpile grade is revised accordingly. The continuous stockpile grade, and the forecast adjustments are both calculated using exponential smoothing, a widely used forecasting technique (Hanke & Reitsch, 1998). However, the two applications of exponential smoothing differ from each other, and from the usual published applications, as will be described below.