Development of a two-tier prioritisationalgorithm for the replacement of water reticulation pipes

[摘要] ENGLISH ABSTRACT: Water pipe replacement in ageing water networks needs to be prioritised within constraints of limitedmunicipal budgets. Relatively higher water pipe failure frequency in a distribution zone could point toa higher replacement priority. Priorities are typically determined based on historically recorded pipefailures, but actual pipe failure data is often not available – especially in developing countries. Pipefailure records may be available for certain zones in a particular system, while no data may be availablein other zones of the same system. Replacement priority cannot be limited exclusively to zones withfailure data, so a method was devised to spatially extrapolate pipe failures from zones with failuredata to other zones where no knowledge of historical failures is available. An algorithm was developedfor this purpose to prioritise pipe replacement based on a two-tier structure, comprising physical andhydraulic characteristics. The following model parameters were incorporated: pipe material,diameter, remaining useful life, static pressure, residual pressure and reserve pressure ratio. Actualpipe failure frequency data for a South African study site with 2021 km of pipes and 12802 reportedfailure events over a period of 180 consecutive months was obtained and used to devise the model.Actual pipe failures were linked to the different model parameters, with all parameter values knownper pipe in the case study area. Pipe failure likelihood index values were then calculated for each pipe element in the water network model (as failure/year/meter). Each pipe was then prioritised forreplacement in terms of a failure likelihood index, and grouped per water distribution zone. The waterdistribution zones were ranked for replacement prioritisation. The model was verified by evaluatingfailure likelihood index values and comparing replacement priority per zone based on actual data tothe model results (for those zones with known data). The model was subsequently used to extrapolatethe replacement priority to other zones without failure records in the case study area, withacknowledgement of in accuracy due to the lack of model validation. The model results are illustrativeand apply to the specific study site – results should not be generalised. The results were representedspatially in GIS format, allowing the user to visually identify the most critical areas for pipereplacement. Future research could involve model validation and possible application beyond thestudy sample.

[发布机构] Stellenbosch University