Estimating sanitary flows using neural networks

The estimation of sanitary sewer flows is required to effectively plan, des ign, build, operate, and maintain sewerage facilities. Existing flow estima tion methods are crude and often not intended to represent actual condition s but rather to act as guidelines, often for design purposes. This study pr oposes a neural network approach to estimating actual sanitary flows under dry weather conditions. Development as well as validation showed the neural network model to produce results with an average error less than 16% when compared to measured data. (C) 1998 IAWQ Published by Elsevier Science Ltd. All rights reserved.