Modelling and visualization of failure rate of a water supply network using the regression method and GIS
B. Kubat , Marian Kwietniewski
In the study, the authors have attempted to develop a failure rate prediction model for a selected water supply network using linear regression, taking into account the simultaneous impacts of many factors on the failure rate; also, possible methods of failure rate visualization for the examined network were presented. In the regression model, the dependent variable is the unit failure frequency (fail/(km year)), whereas independent variables (explaining this frequency) are: pipe material, age and diameter, soil type (impermeable, permeable) and soil moisture content (dry, wet). Statistical analyses have shown that the failure rate of the examined water supply network is affected only by soil conditions. Spatial distribution of the failure number and frequency was visualized, paying special attention to the parts of the network where the spatial concentration of failures was the greatest.