Please use this identifier to cite or link to this item:
Title: Prediction of Water Loss in Hydraulic Distribution System in Minna, Nigeria Using Artificial Neural Network
Authors: Yaba, T.
Jimoh, O. D.
Adesiji, A. R.
Keywords: Water Distribution System
Emitter Coefficient
Artificial Neural Network
Issue Date: 6-Feb-2023
Citation: Yaba, T., Jimoh, O.D., and Adesiji, A. R. (2023). Prediction of Water Loss in Hydraulic Distribution System in Minna, Nigeria Using Artificial Neural Network. The 4th School of Environmental International Conference (SETIC2023) Pp 513-519
Abstract: Water supply network are prone to leakages resulting to a loss of large volume of water. Hence it is required to implement a leak detection/prediction technique through water simulation and machine learning. The main objective of this study is to model water loss in the distribution network of Shiroro District Metered Area. This is important because leak is a measure of efficiency of water distribution network. The hydraulic machine, EPANET was used for the hydraulic modelling of the networks. Emitters were used to simulate leakages at thirty-seven nodes in water distribution system. Physical measurement was carried out also at thirty-seven nodes in the network using measuring can, hose, GPS, meter, stop clock. Nash-Sutcliffe simulation efficiency (ENS) indicates how well the plot of observed versus simulated value fits the 1:1 line. The value of efficiency of 1 (when E = 1) means there is a perfect match of modelled discharge relative to the observed data. The observed and model data were loaded into NSE model using coefficient of 0.1, 0.15 0.2 and 0.3. The performance of the model has suggested that using the emitter coefficient of 0.2 can model the study area. Having established this, the values of the model could be used to predict leakages in the DMA using Artificial neural Network, ANN. This study was based on Multi-Layer Perception which was trained and tested using DMA flow data. The objective was to develop an ANN-based model using flow data generated in the selected DMA in Minna, Niger State, Nigeria. The input variables are elevation, base demand, demand and pressure of the network. The data was trained tested and validated in neural network. The study has shown 17.15% of loss from the nodes in the network. The sum of square errors 13.4% and 5.1% respectively for training and testing of the variables in the machine learning. R square is 97%. The model developed can be used in any district metered area of a distribution network to estimate or predict loss. The developed model is expected to help set the direction of improvement of the analysis of water distribution system and optimal operation of water supply in the studied DMA and other DMAs.
Appears in Collections:Civil Engineering

Files in This Item:
File Description SizeFormat 
TANIMU-Adesiji_SETIC2022.pdf423.86 kBAdobe PDFView/Open

Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.