Author

Sofia Tosso

Date of Award

Summer 2021

Document Type

Open Access Thesis

Department

Chemical Engineering

First Advisor

James A. Ritter

Abstract

A heterogeneous extended Langmuir model (HEL) that uses a truncated frequency energy distribution based on a one, two or three normal distributions is presented to predict gas adsorption mixtures of single, binary, and ternary gas experimental data involving CO2, H2S and C3H8 on H-modernite reported in the work of Ritter et al (1986) in terms of both solid phase molar fractions and total loadings. These models were initially fitted against single gas data and then used to predict binary data and to establish the type of perfect correlations, whether positive or negative, exist amongst these three species. The models then used to predict the ternary data using the perfect correlations previously determined. A dual Langmuir process (DPL), the parameters of which were already determined elsewhere (Ritter et al., 2011) was also used against all experimental data for comparison. In general, the HEL models proved to be fundamentally correct mathematically and able to predict viable correlations among CO2, H2S and C3H8 on H-modernite, with CO2-H2S following a perfect positive correlation, CO2- C3H8 following a perfect negative correlation and H2S-C3H8 following a perfect negative correlation, which is consistent with the fact that both CO2 and H2S are polar gasses while C3H8 is nonpolar. The quality of the predictions among the HEL models were mixed and despite their complexity, none of them were able to match an apparent superior predictive ability of the DPL model, a result that was quite surprising because of the over simplicity of the latter. This reveals the strong ability of the DPL model to predict mixtures and that complex models such as the HEL presently discussed may not lead to better results.

Share

COinS