Estimation of uncertainties in the performance indices of an oxidation ditch benchmark

Estimation of the influence of different sources of uncertainty is very important in obtaining a thorough evaluation or a fair comparison of the various control strategies proposed for wastewater treatment plants. This paper illustrates, using real data obtained from a full-scale oxidation ditch wastewater treatment plant, how the effect of the various uncertainty sources can be quantified. A Monte Carlo simulation analysis method was preferred over a first-order variance analysis method because it is more reliable and it provides the complete probability distribution. For various sources of uncertainty, except for the additive modelling error, samples were selected using the efficient Latin Hypercube Sampling technique. Large deviations in the benchmark performance indices from the nominal values, due to uncertainty in influent loads and parameter values, were found, in particular, for effluent quality and total sludge production indices. However, relatively smaller deviations are found due to uncertainty in the states' initial conditions. The effect of the model structural uncertainty on the performance indices was found to be negligible. (C) 2002 Society of Chemical Industry.