RAW MATERIALS FOR PORTLAND CEMENT: APPLICATIONS OF CONDITIONAL SIMULATION OF COREGIONALIZATION (GEOSTATISTICS)

Modern cement plants commonly rely on homogenization and proportioning facilities to reduce the natural variability of incoming raw materials and to blend different materials into plant feeds satisfying strict chemical specifications. To determine appropriate designs and capacities for such facilities, plant designers must know the statistical behavior of each raw material arriving at the plant. However, if the quarries have not been opened yet, or if historical data on raw-material quality have not been kept, or if the materials to be quarried in the future differ appreciably from those used in the past, then only data extracted from exploratory drilling of the materials in the ground will be available.;Improved methods have been developed for creating statistically and geologically realistic conditional simulations of multivariate, or coregionalized, data such as complete rock analyses. These methods have been used to simulate chemical analyses of two limestone deposits currently being mined for cement manufacture. Simulations of alternative mining and homogenization procedures applied to one of the simulated deposits illustrate how conditional simulations can be employed to select the best mining and homogenization procedures before mining of the deposit has actually begun.;Fortunately, useful information can be obtained by combining conditional simulations of the deposits to be mined with simulations of likely mining procedures. A conditional simulation of a mineral deposit consists of a large set of simulated raw-materials analyses distributed among the nodes of a fine two- or three-dimensional grid covering the region of the deposit that is to be mined. These simulated data should possess all statistical properties, including spatial properties, that are suggested by the available data from the real deposit; furthermore, they should equal the data obtained from the real deposit at actual data locations. The output of a mining simulation applied to a simulated deposit is a time-series simulation of material compositions that mimics the statistical behavior of the real materials to be delivered to the plant in the future.