A total of 148 raw aerobic biodegradation data sets from batch and continuous stirred-tank reactors were extracted from the open literature and previous NJIT MS theses. Kinetic analysis of each of these data sets was performed with respect to the following commonly used empirical models: (1) zero-order, (2) first-order, and (3) Monod. Two constant-biomass versions of each model were evaluated; one in which So (i.e., the boundary condition for substrate concentration at time equal to zero) was assumed to be equal to the measured value of the initial substrate concentration and the other in which So was treated as a regressable parameter. Where adequate biomass concentration data were available, variable-biomass versions of each model, in which So was assumed to be equal to the initial substrate concentration, were also evaluated. Each data set was categorized within one of nine different biodegradation data types and discussed with respect to the advantages and disadvantages of each model evaluated for the given data type. Model selection recommendations were given for each data type.

A theoretical analysis of the effects of variations in raw biodegradation data on the corresponding regression results was performed for the constant- and variable-biomass models. The effects of random experimental error, number of data points, sampling regularity and substrate concentration range were evaluated. The impact of erroneous models on reactor sizing was also demonstrated.