Quadratic Programming For The Multi-Variable Pre-Homogenization And Blending Problem

Kumral, M.
Organization: The Southern African Institute of Mining and Metallurgy
Pages: 6
Publication Date: Jan 1, 2005
Raw material fed into a processing or refining plant is required to be uniform in composition for several reasons. When the mined ore is highly variable in quality the only way to ensure consistency is to homogenize the ore prior to the process. The problem is more complicated in the case of multiple ore sources. The idea behind the research is that theoretical blending ratios can serve not only to meet predefined specific criteria but also to reduce grade fluctuations of variables under consideration. In this paper, the problem is formulated as a quadratic programming problem, whose objective function is in quadratic form and constraints are linear. A case study was conducted on a data set from an iron orebody to demonstrate the technique. The objective was to minimize the blend variability in terms of each variable (in this study, iron, silica, alumina and lime) grade of ores extracted in three different production faces. The stockpile capacity, lower and upper limits of each variable satisfying operational requirements and non-negativity were constrained to the model. A modified simplex method developed by Wolfe was used for solving the blending and homogenization problem. The promising results can be used as apart of the stacking and reclaiming design. Keywords: blending and homogenization problem, quadratic programming, Wolfe method.
Full Article Download:
(581 kb)