Author(s): Nangigadda Mowlali, Ruben Nerella, Venkata Rathnam Erva
Linked Author(s): Venkata Rathnam Erva
Keywords: Air chamber, transient pressures, mitigation, pumping main, ANN
Abstract: Transient mitigation in pumping mains may require the use of devices such as air chambers/ vessels, open surge tanks, air/vacuum valves, pressure relief valves, etc. Selection and design of suitable transient mitigation devices are dictated by the severity of transient causing events. The design of these systems is a challenging problem and selection, installation, and operation of these hydraulic devices depend on the layout, alignment, pipe and pump characteristics and flow rates. The paper presents a regression based artificial neural network (ANN) model for investigating optimized design variable of air chamber (air volume and chamber volumes) from the system parameters viz. , pipeline length, pipe diameter, flow velocity, friction factor, wave celerity, maximum and minimum pressure heads. The system parameters were used as input variables and the corresponding air chamber volume as output variable to train the neural network model. The training has been done by feed forward, back propagation algorithm. The developed model has one input layer (8 system parameters), ten hidden layers (log sigmaoid function) and one output layer (air chamber volume). A case study of pumping main of J. C. R. Devadula Lift Irrigation Project, India is presented. The neural network model simulations were compared with the sizes obtained from the results of commercial water hammer software and observed that ANN models provide economical sizes
Year: 2017