Kinetics of thermophilic, anaerobic oxidation of straight and branched chain butyrate and valerate

The degradation kinetics of normal and branched chain butyrate and valerate are important in protein‐fed anaerobic systems, as a number of amino acids degrade to these organic acids. Including activated and primary wastewater sludge digesters, the majority of full‐scale systems digest feeds with a significant or major fraction of COD as protein. This study assesses the validity of using a common kinetic parameter set and biological catalyst to represent butyrate, n‐valerate, and i‐valerate degradation in dynamic models. The i‐valerate degradation stoichiometry in a continuous, mixed population system is also addressed, extending previous pure‐culture and batch studies. A previously published mathematical model was modified to allow competitive uptake of i‐valerate, and used to model a thermophilic manure digester operated over 180 days. The digester was periodically pulsed with straight and branched chain butyrate and valerate. Parameters were separately optimized to describe butyrate, i‐valerate, and n‐valerate degradation, as well as a lumped set optimized for all three substrates, and nonlinear, correlated parameter spaces estimated using an F distribution in the objective function (J). Each parameter set occupied mutually exclusive parameter spaces, indicating that all were statistically different from each other. However, qualitatively, the influence on model outputs was similar, and the lumped set would be reasonable for mixed acid digestion. The main characteristic not represented by Monod kinetics was a delay in i‐valerate uptake, and was compensated for by a decreased maximum uptake rate (km). Therefore, the kinetics need modification if fed predominantly i‐valerate. Butyrate (i‐ and n‐) and n‐valerate could be modeled using stoichiometry consistent with β‐oxidation degradation pathways. However, i‐valerate produced acetate only, supporting the stoichiometry of a reaction determined by other researchers in pure culture. Therefore, lumping i‐valerate stoichiometry with that of n‐valerate will not allow good system representation, especially when the feed consists of proteins high in leucine (which produces i‐valerate), and the modified model structure and stoichiometry as proposed here should be used. This requires no additional kinetic parameters and one additional dynamic concentration state variable (i‐valerate) in addition to the variables in the base model. © 2003 Wiley Periodicals, Inc. Biotechnol Bioeng 84: 195–204, 2003.