Uncertainty analysis applied to thermodynamic models and process design—2. Binary mixtures

Abstract

A simple procedure is proposed to evaluate the uncertainty on binary interaction parameters propagated from the uncertainties present in the physical properties and equation parameters applied for their calculation using a simple cubic equation of state. A useful database containing 87 binary mixtures present in natural gas processing was constructed through critical collection of available experimental vapour–liquid equilibrium (VLE) data and their associated uncertainties.A thermodynamic consistency test was performed on each isothermal dataset to investigate the quality of the associated VLE data. Upon acceptance of the VLE data based on its quality and consistency, binary interaction parameters and associated uncertainties were determined using a combination of nonlinear regression and Monte Carlo simulation, taking into consideration the uncertainties of pure components, equation of state parameters, and VLE data. The Monte Carlo simulation was also used for the error propagation to estimate the uncertainty on the calculated VLE. Sample calculations were presented illustrating the effect of uncertainties on the PXY and TXY diagrams of ethane/propane and methane/hydrogen sulfide binary mixtures. The required minimum number of stages for a simplified de-ethanizer was calculated taking into account uncertainties of basic input parameters. In addition the effect of uncertainties on the position of calculated cricondenbar and cricondentherm was investigated.