Membrane process for biological treatment of contaminated gas streams

Citation
Sj. Ergas et al., Membrane process for biological treatment of contaminated gas streams, BIOTECH BIO, 63(4), 1999, pp. 431-441
Citations number
36
Categorie Soggetti
Biotecnology & Applied Microbiology",Microbiology
Journal title
BIOTECHNOLOGY AND BIOENGINEERING
ISSN journal
00063592 → ACNP
Volume
63
Issue
4
Year of publication
1999
Pages
431 - 441
Database
ISI
SICI code
0006-3592(19990520)63:4<431:MPFBTO>2.0.ZU;2-8
Abstract
A hollow fiber membrane bioreactor was investigated for control of air emis sions of biodegradable volatile organic compounds (VOCs). In the membrane b ioreactor, gases containing VOCs pass through the lumen of microporous hydr ophobic hollow fiber membranes. Soluble compounds diffuse through the membr ane pores and partition into a VOC degrading biofilm. The hollow fiber memb ranes serve as a support for the microbial population and provide a large s urface area for VOC and oxygen mass transfer. Experiments were performed to investigate the effects of toluene loading rate, gas residence time, and l iquid phase turbulence on toluene removal in a laboratory-scale membrane bi oreactor. Initial acclimation of the microbial culture to toluene occurred over a period of nine days, after which a 70% removal efficiency was achiev ed at an inlet toluene concentration of 200 ppm and a gas residence time of 1.8 s (elimination capacity of 20 g m(-3) min(-1)). At higher toluene load ing rates, a maximum elimination capacity of 42 g m(-3) min(-1) was observe d. In the absence of a biofilm (abiotic operation), mass transfer rates wer e found to increase with increasing liquid recirculation rates. Abiotic mas s transfer coefficients could be estimated using a correlation of dimension less parameters developed for heat transfer. Liquid phase recirculation rat e had no effect on toluene removal when the biofilm was present, however. T hree models of the reactor were created: a numeric model, a first-order fia t sheet model, and a zero-order flat sheet model. Only the numeric model fi t the data well, although removal predicted as a function of gas residence time disagreed slightly with that observed. A modification in the model to account for membrane phase resistance resulted in an underprediction of rem oval. Sensitivity analysis of the numeric model indicated that removal was a strong function of the liquid phase biomass density and biofilm diffusion coefficient, with diffusion rates below 10(-9) m(2) s(-1) resulting in dec reased removal rates. (C) 1999 John Wiley & Sons, Inc.