Evaluation of Precision being Obtained by a Three-Stage Sampling System for Coal

摘要

The primary objective of this paper is to demonstrate a different approach to predicting and verifying mechanical sampling system precision. Early in the year 2011 an experiment was conducted to evaluate the sampling precision being obtained by a three-stage falling-stream mechanical sampling system for coal owned and operated by Peabody Energy. This system collects samples from lots averaging approximately 140001 being loaded onto rail cars at a flow rate averaging about 9000 t/h. In phase 1 of the experiment, a total of 60 increments were collected in a two day period at preassigned intervals using a cross-belt sampler located on a conveyor belt feeding a silo upstream of the sampling system under evaluation. Coal variability parameters were calculated from the resulting laboratory analysis data using the gauge-invariant fractal model developed by the author. Estimates of precision of the three-stage system as a function of the number of increments were calculated using the variability parameters determined in phase 1. In phase 2 of the experiment, duplicate samples of 30 lots of coal were taken by the three-stage sampling system, enabling calculations of the precision being achieved by the system. Phase 2 precision estimates are in reasonable agreement with those of phase 1. Conclusions drawn from the experiment include the following: 1. Experimental variograms or the equivalent correlograms are essential in understanding sampling issues and for designing a sampling system to meet specific precision requirements. 2. Consensus coal sampling standards provided by ISO and ASTM may be improved by treating all stages of a mechanical sampling system as 'sampling' rather than considering downstream sampling stages as 'sample preparation', to be treated in a separate part of the document. 3. The coal industry would profit from use of methodology for routinely monitoring the overall measurement precision and the precision of the individual components of measurement: sampling, sample preparation and laboratory testing. This would not only give necessary credence to the measurement results but would also no doubt, lead to more efficient sampling system designs.