DYNAMIC MODELING OF BACTERIA IN A PILOT DRINKING-WATER DISTRIBUTION-SYSTEM

Computer modeling can be a useful tool in understanding the dynamics o f bacterial population growth. Yet, the variability and complexity of biological systems pose unique challenges in model building and adjust ment. Recent tools from Bayesian statistical inference can be brought together to solve these problems. As an example, the authors modeled t he development of biofilm in an industrial pilot drinking-water networ k. The relationship between chlorine disinfectant, organic carbon, and bacteria concentrations was described by differential equations. Usin g a Bayesian approach, they derived statistical distributions for the model parameters, on the basis of experimental data. The model was fou nd to adequately fit both prior biological information and the data, p articularly at chlorine concentrations between 0.1 and 2 mg/liter. Bac teria were found to have different characteristics in the different pa rts of the network. The model was used to analyze the effects of vario us scenarios of water quality at the inlet of the network. The biofilm appears to be very resistant to chlorine and confers a large inertia to the system. Free bacteria are efficiently inactivated by chlorine, particularly at low concentrations of dissolved organic carbon. (C) 19 97 Elsevier Science Ltd.