Bayesian estimation and sensitivity analysis of toluene and trichloroethylene biodegradation kinetic parameters

Abstract

Parameter estimation is needed for process management, design, and reactor scaling when values from the literature vary tremendously or are unavailable. A Bayesian approach, implemented via Markov chain Monte Carlo (MCMC) simulations using SAS software, was used to estimate the kinetic parameters of toluene and trichloroethylene (TCE) biodegradation by the microorganism Pseudomonas putida F1 in batch cultures. The prediction capabilities of Bayesian estimation were illustrated by comparing predicted and observed data and reported in goodness-of-fit statistics. The sensitivity analysis showed that the parameters obtained using this approach were consistent under the designated toluene and TCE concentration range. Moreover, the impact of TCE on toluene degradation kinetics was numerically exhibited, verifying the fact that TCE was able to stimulate toluene degradation; hence, TCE's presence increased the apparent maximum toluene-specific rate. Various kinetic models were explored at different degrees of complexity. At a low TCE concentration range (e.g., <2 mg L-1 ), a simplified Michaelis-Menten model (i.e., substrate half-saturation parameters approximated the inhibition parameters) was adequate to describe the reaction kinetics. However, at a higher TCE range (e.g., 5 mg L-1 ), a full-scale Michaelis-Menten model was needed to discriminate among the inhibition parameters in the model. The results demonstrated that a Bayesian estimation method is particularly useful for determining complex bioreaction kinetic parameters in the presence of a small volume of experimental data.