NIST ThermoData Engine: Extension to Solvent Design and Propagation of Uncertainties for Process Simulation

ThermoData Engine (TDE, NIST Standard Reference Databases 103a and 103b) is the first product that implements the concept of Dynamic Data Evaluation in the fields of thermophysics and thermochemistry, which includes maintaining the comprehensive and up-to-date database of experimentally measured property values and expert system for data analysis and generation of recommended property values at the specified conditions along with uncertainties on demand. The most recent extension of TDE covers solvent design and multi-component process stream property calculations with uncertainty analysis.
Solvent Design function serves three tasks: (1) selection of best solvent for a solid solute, (2) search for a selective solvent for a solid binary mixture, and (3) selection of best solvent for extraction. Solvents are selected from the list of registered compounds encountering more than 27,000 entries. Selection is made by best efficiency (depending on the task, solubility, selectivity, or distribution coefficient, etc.) and matching other requirements requested by the user. At user’s request, efficiency criteria are evaluated based on experimental data for binary mixtures or predictive models (UNIFAC variations). Predictions can be compared to the available experimental data, and uncertainties are estimated for all efficiency criteria.
Calculations of the properties of multi-component streams including composition at phase equilibria (flash calculations) are at the heart of process simulation engines. However, the accuracy of such calculations are generally unknown that often leads to overdesign of the operational units and results in significant additional cost. TDE provides a tool for the analysis of uncertainty of property calculations for multi-component streams. A process stream in TDE can be either a pure substance or mixture of chemical compounds under bubble, dew, or specified P-T conditions. VLE compositions and single-phase properties are calculated for process streams. Uncertainties are evaluated with the use of the covariance method where covariance matrices account for experimental uncertainties, curve deviations, and inadequacies of the models. Uncertainty analysis shows relative contributions to the total uncertainty from each component and pair of components.