Simulating the capture of CO2 from natural gas: New data and improved models for methane + carbon dioxide + methanol

• The vapor liquid equilibrium of methane + CO2 + methanol, important for capture of CO2 from natural gas, is investigated.
• New binary and ternary mixture VLE data are reported at pressures to 11.4 MPa and temperatures down to 183 K.
• Peng Robinson equation of state binary interaction parameters are optimized for methane + carbon dioxide + methanol.
• Calculated liquid mole fractions of methane, CO2 and methanol are improved by factors of 2.5, 4.4 and 2.5, respectively.

The simulation of carbon capture unit operations often involves predicting the vapor liquid equilibrium (VLE) for mixtures containing polar, non-polar and quadrupolar compounds. In this work, we investigate how well a simple cubic equation of state (EOS) can predict the results of new low temperature, high-pressure VLE measurements of the ternary methane + carbon dioxide + methanol system, which is important to the Rectisol process used for capturing CO2 from natural gas. The ternary p, T, x measurements presented here are the first such data for this system reported in the open literature. First, predictions made with the Peng Robinson (PR) EOS as implemented in a commercial process simulator were compared to binary p, T, x data measured in this work and also taken from the literature. Significant deviations were found between the measured liquid mole fractions and those predicted with the EOS using the default binary interaction parameters (BIP): the relative standard errors were 39%, 77% and 17% for the methane + methanol, methane + carbon dioxide and methanol + carbon dioxide binaries, respectively. Regression of the PR EOS to the binary VLE data by adjusting the BIPs improved the liquid phase mole fraction predictions for the ternary mixture data by a factor of about 2.5 for methane and methanol. However, improvement by a factor of 4.4 in the carbon dioxide liquid mole fraction was achieved by describing the carbon dioxide + methanol binary with an asymmetric composition and temperature dependent mixing rule and tuning the BIPs therein to VLE data for this binary over a wide temperature range. This reduced the standard error in the liquid phase CO2 mole fractions predicted for the ternary mixture using the optimized model by 79% relative to the default model.