Component Trapping with Vapor–Liquid Equilibrium Uncertainty: Principles, Design, and Troubleshooting

Industrial & Engineering Chemistry Research

5 years ago

Component Trapping with Vapor–Liquid Equilibrium Uncertainty: Principles, Design, and Troubleshooting

Bruce Parker, Lydia Narvaez, Thomas Schafer, Paul M. Mathias, Alan Erickson, Henry Z. Kister

Intermediate components tend to accumulate near the middle of a distillation column in many chemical and petroleum separations. Their accumulation may lead to off-spec products, corrosion, plugging, or periodic cycling. The most common method of removing these intermediate components is by taking them out in one or more side draws. The addition of such side draws may not go far enough to achieve the desired product specs. If the side-draw approach fails, the expensive addition of additional towers may be required. The question of whether a side draw is sufficient to remove the accumulating intermediate components or whether additional columns are needed depends on the relative volatilities of the intermediate components at tower conditions. Often, there is a high degree of uncertainty in the model used for correlation of the phase equilibrium of these components. The Margules Uncertainty Analysis method is valuable in analyzing these situations and in guiding engineers to the rational solution. The case analyzed here is an acetonitrile + water separation tower with t-crotonaldehyde and propionitrile as the intermediate components. There are uncertainties in the correlation of the volatilities of these intermediate components in this system. In light of these uncertainties, our analysis addresses the question of whether the tight product specs can be achieved in one tower with a side draw or whether more than one tower is needed. The analysis shows that depending on the product specs and the tolerance for a high reflux ratio, there would be situations where one can be certain that one tower will work, while in other cases more than one tower is needed to ensure achievement of the required product specs. The analysis enables quantification of the risk of the single-column solution—specifically the quantification of confidence limits.

Publisher URL: http://dx.doi.org/10.1021/acs.iecr.7b02812

DOI: 10.1021/acs.iecr.7b02812