Assessment of the Uncertainty of Material Differences

In a fuel processing facility, material may become unaccounted for in different process stages, such as blending, pelletizing, sintering, pellet grinding, and waste disposal. If the transitions between each stage j and a stage where material is accounted for are assumed to be discrete, the probabilities pj per unit mass that material is lost in stage j can be modelled with a discontinuous Markov analysis. The results from calculations of this type are used with principles of information theory to determine the average uncertainty of a) whether or not the material is lost, and b) the stage in which material becomes unaccounted for. A simple example calculation for steady-state plant operations illustrates the possible reduction of these uncertainties obtained with improvements in the performance of instrumentation monitoring material losses. Results show that if the number of different process stages J is large then there is little difference between the two uncertainties because they both approach log2(l+J). Furthermore the ratio of the average rate per unit time at which material is lost to that recovered becomes important only if J is small.