Increasingly, those who work in the field of drinking water have demonstrated an interest in developing models for evolution of water quality from the treatment plant to the consumer's tap. To date, most of the modelling efforts have been focused on residual chlorine as a key parameter of quality within distribution systems. This paper presents the application of a conventional approach, the first order model, and the application of an emergent modelling approach, an artificial neural network (ANN) model, to simulate residual chlorine in a Severn Trent Water Ltd (U.K.) distribution system. The application of the first order model depends on the adequate estimation of the chlorine decay coefficient and the travel time within the system. The success of an ANN model depends on the use of representative data about factors which affect chlorine evolution in the system. Results demonstrate that ANN has a promising capacity for learning the dynamics of chlorine decay. The development of an ANN appears to be justifiable for disinfection control purposes, in cases when parameter estimation within the first order model is imprecise or difficult to obtain.
First-Order Model Checking on Nested Pushdown Trees is Complete for Doubly Exponential Alternating Time
