Many water utilities are still grappling with unprecedented Non-Revenue Water (NRW) challenges in their distribution networks worldwide. Current urban population growth rates are high, thus influencing rapid increases in water demand against the fact that freshwater resources are diminishing. High levels of NRW demonstrate a lack of planning and inefficiency in utility management. In Malawi, through network rehabilitation activities, the Lilongwe Water Board (LWB) planned to reduce NRW from 38.9% to 28% by 2020. Despite huge investments, the reduction has not been significant. The underlying NRW component specific drivers remained unclear. This study analyzed NRW component-specific drivers for the Lilongwe Water Board (LWB) water distribution system in Lilongwe City, using Fixed Effects Regression and the feedforward backpropagation Artificial Neural Network based Improved Garson algorithms. One month of primary data on water production and consumption maintained by the LWB for two District Metered Areas (DMAs) (SZA1 and SZD3) was used to assess NRW trends. Flow measurements were done using flow data loggers. Real system loss data was collected through the Minimum Night Flows (MNF) model. Historical data on the physical characteristics of the distribution system was also collected. The water balance for the two DMAs confirmed that post-rehabilitation NRW (38.95%) was still too far above the set target of 28%. The water loss components analysis showed that Apparent losses (AL) (21.18%) were higher than Real losses (RL) (16.85%) and Unbilled authorized consumptions (UAC) (0.92%). Components specific analysis showed flushing during maintenance works as the main driver for UAC, while accounting errors, illegal connections and customer non-payments drove AL. Background leakages and bursts, connection density, type of pipe materials and population density were the main RL drivers. In conclusion, this study provides insights into the specific NRW drivers for targeted interventions by LWB to address the identified NRW drivers and meet the utility’s goals and global standards.
Keywords: Non-revenue water, Apparent Losses, Real Losses, Artificial Neural Network, DMA, Lilongwe City
