Preliminary Feed Test Algorithm for the Hanford Waste Vitrification Plnat product composition control system

摘要

The Feed Test Algorithm (FTA) will test the acceptability (conformance with requirements) of process batches in the Hanford Waste Vitrification Plant (HWVP). Although requirements and constraints will be imposed on properties of the material in the melter and the resulting glass, the FTA must test acceptability while the batch is still in the Slurry Mix Evaporator (SME), i.e., before material is transferred to the Melter Feed Tank. Hence, some properties upon which requirements will be imposed must be estimated from data available on the feed slurry. The major type of data to be used in this estimation is feed composition, usually expressed in terms of nine oxide mass fractions and a catchall tenth category, Others. Uncertainties are inherent in the HWVP process. The two major or types of uncertainty are composition uncertainty (that related to measurement and estimation of feed composition and other quantities) and model uncertainty (uncertainty inherent in the models developed to relate melt/glass properties to feed composition). Types of uncertainties, representation of uncertainty, and a method for combining uncertainties are discussed. The FTA must account for these uncertainties in testing acceptability; hence it must be statistical in nature. Three types of statistical intervals (confidence, prediction, and tolerance) are defined, and their roles in acceptance testing are discussed.