STOCHASTIC REAL TIME CONTROL OF WASTEWATER TREATMENT PLANT OPERATION

ABSTRACT Stochastic operation of wastewater treatment plants, especially of those receiving wet weather transient flows, relies on predictions of the input and of the system conditions during and at the end of the operation period. The predictive models can be extracted from available knowledge about the system and from the past operational and input data. ARMA and AR-transfer function models are the most appropriate models. These models can also contain known deterministic relationships. In the stochastic mode of operation the predictions carry an error which should be an unpredictable random noise. Knowledge of the noise characteristics, which can be extracted, enables to estimate the probability of success of the operation and operate the system according to the desired risk of failure. A computerized operational system is also described.