You are here:
Revisiting Isotherm Analyses Using R: Comparison of Linear, Non-linear, and Bayesian Techniques
Citation:
Fairey, J. L., D. WAHMAN, A. D. Pifer, AND G. V. Lowry. Revisiting Isotherm Analyses Using R: Comparison of Linear, Non-linear, and Bayesian Techniques. Presented at AEESP 2009, Iowa City, IA, July 26 - 29, 2009.
Impact/Purpose:
To inform the public.
Description:
Extensive adsorption isotherm data exist for an array of chemicals of concern on a variety of engineered and natural sorbents. Several isotherm models exist that can accurately describe these data from which the resultant fitting parameters may subsequently be used in numerical adsorption models and design of treatment processes. However, various techniques have been used for analyzing adsorption data, determining uncertainty, and incorporating this uncertainty into the design process. Log-linear least squares (LLLS), geometric mean functional regression (GMFR), non-linear least squares (NLLS), and Bayesian analyses have all been applied to isotherm data, but the suitability of these techniques and their relationships with one another have precluded wide-scale adoption of a unified approach. In this poster, we compare the fitting parameters and confidence intervals of select isotherm data using LLLS, GMFR, NLLS, and Bayesian techniques using the freeware program R. The goals of this research were to (1) determine if differences existed between these four analysis techniques, (2) assess the practical importance of any differences, and (3) demonstrate the usefulness of R in analyzing isotherm data.