Performance evaluation and physical property study of structured packing HETP correlations
Abstract
Structured packing HETP correlations play a significant role in the design andoperation of packed distillation columns. A model derived from inadequate HETP data will not only affect where it may perform poorly but will also impact the relationshipbetween specific physical properties and the model’s predicted HETP. This study focused on the impact of physical properties on structured packing HETP correlations in the literature, as well as their HETP prediction capabilities. The evaluated models included Gualito (1997), Billet & Schultes (1999), Delft(2004, 2014), Aspen (2012) and Song (2017). Multiple hydrocarbon and aqueous systems were analyzed, with pressures ranging from 0.130 – 27.6 bara. The sensitivity testing wasmeant to highlight the differences in model relationships with physical properties, causedby assumptions, different physical property ranges used for development, etc. As a result of the analysis, the counter-intuitive relationships between HETP and liquid viscosity were observed with the Billet & Schultes and Gualito correlations. For these two correlations, HETP decreased with increased liquid viscosity. In general, all the modelsshowed varying degrees of sensitivity to liquid viscosity, some were more impacted atlow pressures, while others at high pressures. In addition to liquid viscosity, thedifferences in dependency on surface tension were highlighted between all the models.Some correlations demonstrated predicted HETP changes of 15% or more, while other correlations were impacted by less than 3%, with the same variation of surface tension.Overall, the physical property sensitivity results quantified the impact of surface tensionand liquid viscosity with multiple types of systems. Monte Carlo methods were alsoutilized to determine the uncertainty of each model with different physical propertyinputs. In addition to the physical property study, a HETP performance evaluation wasconducted with various types of packings and test systems. Billet & Schultes correlationgenerally performed the best, due to regressed packing-specific constants, but other correlations such as Song and Aspen performed nearly as well, and in some cases better. Other aspects of the models such as effective area predictions were discussed in detail and observations were made about model strengths and weaknesses.
Collections
- OSU Theses [15752]