Date of Award

Spring 1998

Document Type


Degree Name

Doctor of Philosophy in Chemical Engineering - (Ph.D.)


Chemical Engineering, Chemistry and Environmental Science

First Advisor

Basil Baltzis

Second Advisor

Gordon Lewandowski

Third Advisor

Piero M. Armenante

Fourth Advisor

Robert G. Luo

Fifth Advisor

Richard Bartha


This study dealt with the removal of vapors of volatile organic compounds from airstreams in biotrickling filters (BTFs). A detailed general model was developed for describing the process under steady-state conditions. The model accounts for mass transfer between phases (air, liquid, biofilm) and biodegradation of pollutants in the biofilm. It also accounts for potential kinetic interactions among pollutants as well as potential process limitations by oxygen availability.

The general model was experimentally validated using mono-chlorobenzene (m-CB) and ortho-dichlorobenzene (o-DCB) as model compounds either alone or in mixture with each other. Before BTF experiments were undertaken, a systematic kinetic study was performed with suspended cultures. Two microbial consortia, called m-CB and o-DCB consortium, were used. The o-DCB consortium could use both m-CB and o-DCB as sole carbon and energy sources whereas the m-CB consortium could not utilize o-DCB. In all cases it was found that self-inhibition (Andrews kinetics) takes place. When the two compounds are present in a mixture they are simultaneously used but are involved in a competitive cross-inhibition which is stronger from m-CB presence on o-DCB removal than vice versa. Studies on the effect of pH showed that a value of 6.8 is optimal. Some kinetic studies were repeated after the biomass had been used in a BTF for about 8 months and showed that the kinetics, i.e., the values of the kinetic constants remained unaltered.

Experiments in a BTF with the m-CB consortium and m-CB as model compound were performed with air residence times between 3.0 and 8.8 min, liquid flow rates between 0.7 and 5.7 Lh-1, and inlet m-CB concentrations between 0.4 and 4.4 gm-3. The percent m-CB removal observed ranged from 79 to 96% and the maximum removal rate was 60 gm-3-packing h-1. Removal of o-DCB vapor was found to be more difficult. In fact, using a BTF with the o-DCB consortium percent o-DCB removal ranged from 57 to 76% and the removal rate never exceed 30 gm-3-packing h-1 In these experiments, the air residence time, liquid flow rate, and inlet o-DCB concentration were in the range of 3.0-6.5 min, 1.2-5.2 Lh-1 , and 0.25-3.5 gm-3, respectively. In all cases, a very good agreement between data and model predictions was found. Regarding removal rates, the proposed model predicted the data with less than 10% error in most cases. Most experiments were performed in counter-current flow of liquid and air, but some were performed in co-current mode. Co-current operation was found to be slightly superior to the counter-current mode; this is also predicted by the model. The great majority of BTF experiments was performed at pH 6.8. Some experiments at lower pH values showed considerable VOC removal somewhat unexpected based on the suspended culture studies.

Experiments in a BTF with the o-DCB consortium and airstreams laden by both m-CB and o-DCB validated the proposed model for the case of mixtures. These experiments were performed in counter-current flow of air and liquid. The liquid flow rate was 6 Lh-1 whereas air residence time, and m-CB and o-DCB concentrations varied in the range of 3.2-5.9 min, 0.17-3.1 gm-3, and 0.1-0.8 gm-3, respectively. The agreement between model predictions and data was very satisfactory but not as good as in the case with single VOCs.

For removal of m-CB/o-DCB mixtures it has been shown that kinetic interference can be neglected because the expected VOC concentrations are low. Regarding oxygen, it was found that an oxygen-controlled zone exists in the BTF (close to the inlet of the polluted air) when the total VOC concentration is relatively high. For the hydrophobic compounds used in this study oxygen availability does not seem to play a crucial role. Model sensitivity studies have shown that at least two kinetic constants are important and thus, zero or first-order kinetic approximations cannot and should not be made.

The model developed in this study along with the computer code generated for solving the equations can be used in (at least preliminary) scale-up calculations for the design of BTFs.