Adsorption technologies are widely employed in many important separation processes, especially in fine chemistry and environmental control. Thus, the increasing pressure for cost reduction in the operation of industrial plants, which calls for the use of efficient design techniques based on scientifically advanced methods, has led to the development of sophisticated physical chemical models for the description of adsorption equilibrium parameters. To account for the complex phenomena that take place at the microscopic level in the adsorption process, the most recently developed models require the estimation of a number of parameters higher than the ones present in the traditional Langmuir and Freundlich models. On the other hand, the presence of an increased number of strongly correlated parameters requires the use of suitable statistical methods for the information contained in the experimental results to be utilized efficiently. In this article we present a method that generalizes previous identification procedures to complex models containing an arbitrary number of parameters. The sensitivity of the resulting estimates on error distributions assumed and theoretical models chosen is examined using both simulated and experimental data.
A Maximum Likelihood-based Method for the Nonlinear Estimation of Equilibrium Adsorption Parameters
SOLISIO, CARLO;LODI, ALESSANDRA;DOVI', VINCENZO;REVERBERI, ANDREA
2015-01-01
Abstract
Adsorption technologies are widely employed in many important separation processes, especially in fine chemistry and environmental control. Thus, the increasing pressure for cost reduction in the operation of industrial plants, which calls for the use of efficient design techniques based on scientifically advanced methods, has led to the development of sophisticated physical chemical models for the description of adsorption equilibrium parameters. To account for the complex phenomena that take place at the microscopic level in the adsorption process, the most recently developed models require the estimation of a number of parameters higher than the ones present in the traditional Langmuir and Freundlich models. On the other hand, the presence of an increased number of strongly correlated parameters requires the use of suitable statistical methods for the information contained in the experimental results to be utilized efficiently. In this article we present a method that generalizes previous identification procedures to complex models containing an arbitrary number of parameters. The sensitivity of the resulting estimates on error distributions assumed and theoretical models chosen is examined using both simulated and experimental data.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.