Defining Blending Classes to Solve Open Pit Scheduling Problems

For this paper, blending is conceived as a linear process that takes a proportion from one stockpile or location and mixes it with another proportion from another stockpile or location, with the aim of obtaining a set of attributes in the final combination that satisfies a given quality standard. A blending class is the combination of ore attributes and grade limits that should be stockpiled in order to achieve the target quality for the process Plant. Current literature considers the number of blending classes in the Open Pit Scheduling Problem as fixed; however, this paper shows how to define specific blending classes for a given deposit so that the schedule can be solved with Sequential Linear Programming. Finally, a practical application in a Peruvian copper mine is shown, involving four ore attributes: alteration code, arsenic, oxide content and copper grade. INTRODUCTION In copper ore deposits, blending for quality has not been a typical issue as it was for coal and iron ore; however, current copper deposits in the world are becoming geologically and chemically more complex than in the past and is not uncommon to have quality standards involving up to four attributes, introducing an additional complexity to the mine scheduling process [1]. In order to solve the Open Pit Scheduling Problem with multiple ore quality constraints, two assumptions are necessary [2]: - Grade constraints are defined only by specifying an upper and lower limit; - Ore attributes in stockpiles are pre-determined, i.e., the upper and/or lower grade limits defining each stockpile are already known at the time of setting the schedule. Assuming each ore attribute is classified as high, medium or low, and there are four attributes to be controlled in the target ore quality, the number of possible blending classes would be 3^4 = 81. This number is impractical to manage in Operations, and increases the complexity at the time of running the schedule; therefore, a more efficient way of classifying ore is required. Note that in practice, one blending class can be split into more than one stockpile location due to physical capacity constraints. Therefore, the number of stockpiles could even be more than 81. BLENDING CLASSES DETERMINATION For a given set of ore grades with lower and upper limits to be achieved, one has to find all feasible ore classes such that when blended, their grades will lie within those limits.