Model for Methanol Partitioning in Fractionation Column
Rameshwar Hiwale* and Kathy Bigger
Linde Process Plants, Inc., Tulsa, USA
Natural gas hydrates are formed in the presence of moisture, low temperatures and high pressure operations. Hydrate formation is a potentially serious problem in natural gas flow lines starting at the production well all the way through to the customer delivery system. Methanol is an effective hydrate inhibitor exhibiting low viscosity, high water affinity, and relatively low cost, thus making it a versatile solvent. The injection of methanol into pipelines has become common practice in the natural gas industry to prevent hydrate formation and prevent corrosion. The use of methanol has been increasing recently, specifically due to the enactment of environmental restrictions limiting the benzene, toluene, ethylbenzene and xylene (BTEX) emissions of Triethylene Glycol (TEG) Dehydration plants. For these reasons, methanol has been and will frequently be present in feed streams to natural gas liquid (NGL) fractionation plants. However, then total level of oxygenates in NGL streams is limited by specification. Therefore, the accurate prediction of methanol partitioning in a distillation column is critical to ensure that NGL products maintain their maximum value.
Apart from its desirable characteristics, methanol also exhibits an elevated vapor pressure and complex polarity. Thus methanol distribution can become a concern, particularly in fractionation columns such as a Deethanizer where product impurities are of high interest. The accuracy of simulation tools to predict the methanol partition coefficient is critical to maximize return on investment. Existing thermodynamic models are tailored toward systems of high methanol content and do not predict methanol distribution accurately when methanol is present as a low concentration contaminant.
In this study, in the first phase, the model is developed to predict methanol distribution along with column. The accuracy of a vapor-liquid equilibrium (VLE) model for methanol-hydrocarbon mixtures has been improved through modification of the simulation-embedded binary interaction parameters. The modified model accuracy to predict the methanol partition coefficient was confirmed by comparing model predictions for wide range of operating conditions with published VLE data from GPA Research Report 219.
In the second phase, the improved thermodynamic model results will be confirmed by comparing experimental results for methanol concentration in the overhead ethane product stream from Plant in 2016.