An improved differential evolution method for efficient parameter estimation in biofilter modeling

Abstract

The task of mathematical modeling of biofilters is weighed down by a lack of reliable information about some parameters, which decisively influence the substrate biodegradation process. Biofilm thickness is one such parameter. Biofilters handle diverse substrate loading rates and gas phase velocities, which have a significant effect on the biofilm thickness. The problem of evaluation of this parameter is further compounded by the fact that there are other related issues, which need to be addressed. Secondly, information about biokinetics is rather hard to get because of problems with experimentation. Thirdly, air/water distribution coefficients are used as a rule for predicting the amount of substrate going into the biolayer from the gas phase, when, in fact, air/biofilm distribution coefficients should be used. Henry's law distribution coefficients for the substrate in actual conditions are equally difficult to determine. Inverse modeling is a viable strategy for estimation of these critical parameters. An important component of the inverse modeling is an efficient optimization technique. In this study, an improved differential evolution method (IDE) has been proposed and applied for the evaluation of the parameters mentioned above, in the framework of the solution of an inverse problem. It has been demonstrated that, the IDE, which is a variant of the original DE method, is twice as efficient as the DE method in terms of computational speed for the same level of accuracy. A case study involving the biodegradation of phenol in air streams has been investigated to validate the models proposed.