Mobile Edition
Print
Get the Mag
Weekly UpdatesINTRODUCTION
Economic decision analysis models provide a robust framework for understanding complex problems and improving the quality of decision-making under uncertainty. Economic decision analysis can provide a foundation to support strategic investment and operational decisions, particularly when they are characterized by significant uncertainty. In this case study, we implement a value of information methodology to analyze a mine company's decision to purchase ore grade scanners. We demonstrate that the expected financial benefits of installing ore grade scanners far outweigh the costs. In addition, the scanner purchase enables the world's largest underground iron ore mine to better meet long-term contractual agreements.
The Loussavaara-Kiirunavarra Aktiebolag (LKAB) company operates an underground iron ore mine north of the Arctic Circle in Kiruna, Sweden. The mine produces three ore types, each with different processing requirements and monthly production targets. Kiruna's mining method, sublevel caving, leads to a high degree of ore dilution during recovery, which, in turn, creates significant uncertainty in the quality of the extracted ore. We evaluate an ore grade scanner technology that provides information that may improve the quality of mine operators' decisions with respect to ore type identification and subsequent classification. We utilize a value of information framework that can potentially enable a decision-maker to make better choices in the context of the underlying uncertainties. The results of our work yield the following benefits: (1) a methodology for estimating the costs associated with ore misclassification errors in the mining sector; (2) a real-world application of the value of information framework as it relates to the mining sector; and (3) a prescriptive approach to improving operating decisions in mineral production.
This article is organized as follows: first, we review the relevant literature. Then, we discuss the issue of ore misclassification and identify the causes and magnitude of misclassification errors over a three-year period using ore extraction data obtained directly from the company. Next, we approximate the cost of these misclassification errors to the Kiruna mine. We then use a value of information framework to analyze the impact on decision-making of obtaining additional information on extracted ore quality. We compare the expected value of additional information to the cost of purchasing the source of information. We conclude with a summary of our findings and the managerial implications.
LITERATURE REVIEW
A variety of deterministic and stochastic decision models have been applied to complex problems in the mining sector. Deterministic models--e.g., static net present value calculations--compare costs and benefits of alternative mining methods given assumptions regarding ore body size and shape, reserve quantity and quality, and market prices (Boshkov and Wright, 1973). Laubscher (1981) outlines factors affecting underground mining method selection including regional rock stresses, rock mass classification, and location and layout of ore body geometry. Sevim and Sharma (1991) select least-cost transportation options for surface coal mines. Nicholas (1981, 1992) ranks different mining methods based on a set of critical decision factors and then performs a cost analysis on the highest-ranked methods to determine a least-cost implementation. Celebi (1998) uses an integer program to select the optimal mix of equipment to strip Turkish surface mines.
Stochastic modeling methods have also been utilized in the mining sector as a means to support decision-making. Kappas and Yegulalp (1991) use queuing theory to analyze a truck-and-shovel system in an open pit mine to determine optimal operating and dispatching policies. Zhonghou and Qining (1988) use a similar methodology to select trucks and shovels for mining operations. Dimitrakopoulos, Farrelly and Godoy (2002) apply uncertainty and risk analysis in open-pit design and production scheduling. Magalhaes et al. (1996) utilize simulation techniques to set operational policies for trains in underground mines. Sturgul (1997) provides a comprehensive review of mine simulation literature.
Our emphasis in this work is on the application of a decision science approach known as value of information (VOI) methodology. Previous applications of this stochastic approach to support decision-making are limited in the mining sector. Peck and Gray (1999) make no explicit reference to VOI, yet they discuss the potential benefits to decision-makers of gathering information in the mining industry. Barnes (1986) uses VOI to incorporate geostatistical estimation into mine planning. Typical estimates done via kriging provide not only a parameter estimate but a measure of the uncertainty associated with this parameter, the parameter variance. The author investigates geologic delineation sampling as a technology that has a cost and information value associated with it.
In comparison to the mining sector, the value of information approach has been researched more extensively in a related industrial sector--the oil and gas industry. Grayson (1960) was first to demonstrate the application of VOI to information purchases that may aid drilling decisions. Newendorp (1975) also discusses value of information in his classic petroleum decision analysis text. More recently, there have been a number of illustrations and applications of value of information applied to seismic information purchases. Seismic data represent an essential source of information utilized to characterize geological and/or geophysical features and to assist in hydrocarbon reservoir characterization and management. Seismic data can have significant economic benefit and cost implications. Much of the previous research and many of the applications of VOI techniques in the oil and gas sector focus on the value of seismic information. For a more extensive literature review of the theory and application of value of information concepts in the oil and gas industry, see Bratvold and Bickel (2007). Other works include the examination of the accuracy of that information (Stibolt and Lehman, 1993; Houck, 2004; Steagall et al., 2005; Pickering and Bickel, 2006). The issue of information accuracy and its impact on the value of information is an important element of our work that concerns ore collection and classification procedures.
This research advances the use of economic decision analysis methodologies (VOI) as they apply to the mining sector. This work contributes to the economic decision analysis literature by (1) providing a sound and practical technique for the application of the value of information approach in a complex mining decision context; (2) informing the academic community about the practice of economic decision analysis in the mining sector; and (3) describing an actual application with demonstrable value to the participating organization.
THE KIRUNA MINE AND ORE MISCLASSIFICATION
The Kiruna ore body is a high-grade magnetite iron ore deposit approximately 4 km long and 80 km wide (Kuchta, 2002). Mine operators extract the iron ore via a mass mining method known as sublevel caving that employs the concept of gravity flow to assist in ore recovery. Drifts are drilled horizontally into the ore body and then charged and blasted, which causes the ore to filter down into the drift. After a drift is blasted, load haul dump units transport the ore to an ore pass. Ore from an ore pass fills a 455-ton capacity train on the main haulage level. The train then transports the ore to a set of four crushers, three of which are operational at any point in time. After the ore is crushed, it is hoisted to the surface and is processed at one of four mills before the product is shipped to markets (steel mills) in Europe and the Middle East. Figure 1 depicts a typical sublevel caving operation.
Extracted Ore Grade and Ore Grade Uncertainty
There are two main ore types located in situ. About 80% of the ore body contains a high-iron, low-phosphorus B-type ore and the remaining 20% is a high-phosphorus D-type ore. Extraction of the two in situ ore types yields three ore types that are then processed: B1, B2, and D3. B1 ore is characterized by having a high iron content (~68% on average), a low phosphorus (P) level (~0.06%), and a potassium ([K.sub.2]O) level lower than 0.15%. Raw B1 ore is enriched simply by crushing and grinding it to a small particle size and then using magnetic separation to obtain the ore (fines), leaving the impurities behind. Although this processing method is relatively cheap, the product cannot be ground to too fine a granularity, because it would become too difficult to handle. (One can handle sand, but not dust.) However, the coarser granularity of the final product limits the degree to which the impurities can be removed. Hence, ore classified as B1 can contain only limited amounts of the two impurities found at Kiruna, P and [K.sub.2]O. B2 ore is typically formed during the extraction process when the high-iron content, low-phosphorus-content B1 ore mixes with waste rock, raising the phosphorus content of the ore. On average, the B2 ore contains approximately 0.2% P, but the [K.sub.2]O level is irrelevant. Raw B2 ore is enriched in much the same way that B1 ore is, only the ore is ground to a much smaller particle size. The smaller size allows a higher level of impurities to be extracted via magnetic separation, but the resulting product, of the consistency of dust, is difficult to handle and must be transformed (at an expense) into pellets suitable for transportation. D3 ore has the highest level of phosphorus, greater than 0.9% P (Topal, 2003). Raw D3 ore is also crushed and ground, but a more expensive flotation process is used to remove all contaminants, regardless of their levels. Table 1 illustrates the average content of the key elements. Note that phosphorus is the primary distinguishing factor. Potassium content is important only when categorizing B1 ore.
FEED