Physical Property Modeling of Concentrated Cesium Eluate Solutions—Part I: Derivation of Models

Abstract

Both experimental and process simulation studies were performed to develop physical property models of the concentrated cesium ion-exchange eluate solutions in one of the Hanford Waste Treatment Plant evaporators. The physical properties of interest included the bulk solubility, density, viscosity, and heat capacity of the evaporator bottoms, and the proposed model of each response was a linear mixture model containing 12 coefficients. A unique feature of this work is that the values of these coefficients were determined by the regression of the “virtual” experimental data, which were not measured but were calculated using a computer process model that simulated the semibatch evaporation of cesium eluate solutions. To improve the accuracy of calculated virtual experimental data and the resulting physical property models, a series of benchscale evaporation tests was also conducted to provide the necessary experimental data for the development of a multielectrolyte thermodynamic database on which the computer process model was built. Specifically, the solubility and other physical properties of selected binary, ternary, and higher-order systems were measured to support the optimization of a sexenary database for the Na-K-Cs-Al-HNO3-H2O system. As the input to the virtual experimental runs, a matrix of cesium eluate simulants was designed within the bounding concentrations of the major analytes identified in radioactive samples. The computer process model was then run in conjunction with the sexenary thermodynamic database to calculate the physical properties of each matrix solution concentrated to the target end points of 80 and 100% saturation. The calculated physical properties were analyzed statistically and fitted into the 12-coefficient mixture function of temperature and the concentrations of major analytes in the unevaporated eluate. Over the concentration and temperature ranges considered, the resulting empirical physical property models were found to correlate the computer-generated data well without significant bias.