Limitations of oxygen transfer in fermentation can be solved using auxiliary liquids immiscible in the aqueous phase. The liquids (called oxygen-vectors) used in this study were hydrocarbon (n-dodecane) and perfluorocarbon (forane F66E) in which oxygen is highly soluble (54.9 mg/L in n-dodecane and 118 mg/L in forane F66E at 35 degrees C in contact with air at atmospheric pressure). It has been demonstrated that the use of n-dodecane emulsion in a culture of Aerobacter aerogenes enabled a 3. 5-fold increase of the volumetric oxygen transfer coefficient(k(L)a) calculated on a per-liter aqueous phase basis. The droplet size of the vector played a crucial role in the phenomena. When a static contact between gas bubble and vector droplet was established in water, the vector covered the bubble, in agreement with positive values of the spreading coefficient for these fluids. The determination of the oxygen transfer coefficients (k(L)) in a reactor with a definite interfacial area enabled the main resistance to be located in the boundary layer of the waterside either for a gas-water or a vector-water interface. Because oxygen consumption by weakly hydrophobic cells can only occur in the aqueous phase, the oxygen transfer is achieved according to the following pathway: gas-vector-water-cell. Finally, a mechanism for oxygen transfer within this four-phased system is proposed.

No abstract

A cognitive model for anaerobic digestion in fluidized bed reactors is developed. The general pathway of the process is divided into five main reactions performed by different bacterial groups. Molecular diffusion of each substrate involved in the reaction scheme is described. Effectiveness factor calculations are performed in steady state for each bacterial group taken into account in the process. The case of a single substrate removal is discussed, and optimal biofilm sizes are found. Sequential substrate removal is investigated, and different kinetic regimes are characterized. The influence of biofilm size and primary substrate removal is discussed in the case of standard concentrations in the liquid phase. This study shows that, according to the theoretical model the limiting step of the process may be different and depends in a large way on mass transfer effects. Finally, importance of biofilm size is compared for acidogenic and methano-genic steps: each reaction is found to be optimized for different biofilm thicknesses. This result may be of interest for design purposes and further dynamic modeling. Concluding remarks concerning the validation of the model are made, and a comparison to experimental data from the literature is presented. (c) 1995 John Wiley & Sons, Inc.