Decision enhancement for the management of apparent water losses

Abstract

Non-revenue water (NRW) is a prevalent and pervasive problem in both the developed and developing nations of the world. Coupled with increasing water demands that are outstripping avail-able supply, the need to significantly improve water use efficiency and water loss management has never been more urgent. Of the two main components of water losses, real (physical) losses have enjoyed greater attention in both research and practice that apparent (commercial) losses despite that latter potentially being more impactful as it is measured in terms of the retail cost of water. Research in apparent water loss has shown that metering errors are the leading contributor but also that traditional assessment methods are costly, laborious and out of reach for many utilities. There is also limited adoption of data-driven discovery processes to glean insightful trends from existing data to inform improved apparent loss management. Existing management tools are also frag-mented and do not provide for the integrating assessment of the various component of apparent water losses. There is therefore a clear need to improve the status quo in both the assessment methodologies and the management tools for the improved and integrated assessment and man-agement of apparent water losses, particularly within the context of the developing world. This thesis contributes towards the quantification and the management of apparent losses, with the City of Johannesburg, South Africa as the study area. Representative consumption profiles for the city were determined using results of the meter logging of 408 properties in Johannesburg and ad-ditional third-party data for 256 properties from various areas within South Africa. Residential con-sumers have up to 37% of flow occurring at less than 0.03m3/hr, indicating a substantial proportion of low flow rates that might not be adequately measured by water meters. A 63% occurrence of on-site leakage with a significant monthly leakage of 11 – 41m3/month/property was found in the Johan-nesburg. Decision Enhancement for the Management of Apparent Water Losses Meter testing at a SANAS accredited flow laboratory was successfully done for 123 small diameter meters at 10 test flowrates to better represent the meter curve and characterise the meter accuracy of Johannesburg water meters. This was supplemented by the analysis of over 3,000 historical of meter testing records to determine the meter accuracy degradation rates at various flowrates. The starting flowrate showed the expected highest degradation rate of up to 0.25% per every 1,000kℓ and 0.32% per year of meter age, with other flow rates exhibiting lower degradation rates. Apparent water losses within Johannesburg were evaluated using three different methodologies, two of which were alternative methods that had not been previously validated against the best practice weighted accuracy method. Using the weighted meter accuracy method, apparent losses due to meter inaccuracy estimated to range from 9.4% to 14.6%, on average, depending on the actual ratio of meter sizes used. The comparative billing analysis method estimate was closer to the worst-case scenario at 14%, while the meter change analysis method understated the losses at 4.7% for a subset of the data. Both alternative methods were found to be unreliable in their current form for the estimation of apparent losses although they do provide some insight into the meter management practices of a utility. A proposal is made on how the weighted error methodology can be also modified and enhanced to better quantify errors at low flows. The annual apparent losses were found to be location specific and reliant on prevailing meter regulation and therefore the default reference annual apparent losses of 5% will need to account for this. Proposals on how this can be done are included in the research. An improved comparative billing methodology was developed using a combination of data prepara-tion techniques, clustering analysis and classical regression analysis. Consumption was found to be reducing on at a rate of 0.7% per annum, similar to the weighted error degradation rate found in other studies using the weighted accuracy methodology. Apparent losses due to metering error was estimated to be 8.1% of billed consumption for the residential sector against 11.2% estimate from the weighted accuracy method for the same sector. The difference is fully accounted by meter under-registration, shown to be higher than 4% for new meters, which the developed methodology does Abstract not account for. A general relationship is also proposed for similar utilities to estimate their apparent losses using the results of the developed method. Based on the results of this research, a simplified meter management framework is proposed and is made up of four main components namely Systems, Audits, Analysis and Metering (SAAM). The main objective of the framework is to simplify the main requirements for achieving improved assess-ment and management of apparent water losses. Based on the SAAM meter management frame-work, an Apparent Loss Management Tool (ALMAT) was also developed to enable utilities to per-form, among other things, meter audit analysis to derive consumption profiles; meter error analysis; optimum meter replacement period determination; and to assess apparent losses and meter accu-racy degradation analysis using the weighted accuracy method and the improved comparative billing analysis method developed in this study. Keywords Apparent water loss, water metering, water meter management, non-revenue water, water demand management, decision support