Development and application of a multi-period pipe replacement model utilising risk-based prioritisation for water distribution systems

De Klerk, Altus Hugo (2016-03)

Thesis (MEng)--Stellenbosh University, 2016.

Thesis

ENGLISH ABSTRACT: The decision on whether to replace a pipe, with specific reference to a water distribution network, can be complicated by several aspects in the decision making process, such as available funding and required system performance. Long term budget requirements need to be assessed for the effective management of an existing water distribution network to find balance between the return on investment and customer satisfaction. Various failure prediction models are available to calculate the probability of failure of each pipe in a water network. The probability of failure is then used to determine a replacement priority for all pipes in the network accordingly. Research has shown that the choice and implementation of failure prediction models are sensitive to the availability of data and in many cases a high degree of expertise is required to sufficiently understand the results. Semi-quantitative risk assessments provide a structured way to rank pipes by accounting for likelihood and consequence of failure while providing adaptability to the availability of data. In order to utilise the advantages of the risk-based approach a multi-period replacement model was developed to determine a suitable long term investment strategy, while taking some practical considerations into account. A model was developed which utilised a risk-based approach to determine the pipe replacement priority. The model considers each pipe in a pipe inventory database based on several contributing pipe attributes and the available budget. A failure forecasting algorithm was also included in the model. The model could be used to determine the required budget based on certain fixed input parameters such as the total length of pipe to be replaced or the total allowed number of failures per year. Four hypothetical investment scenarios were analysed for a case study. The results were compared to a fifth scenario, noted as the reactive strategy, which involved no pipe replacement. For the specific case study that was analysed the reactive strategy involved the lowest total cumulative expenditure. Additional investment was required to improve the performance indicators for the number of failures, service interruption duration, estimated remaining useful life and estimated remaining asset value. This research presented a methodology across the different performance indicators noted above, wherein the relative weights of the performance indicators were used to calculate a best-fit index.

AFRIKAANSE OPSOMMING: Die besluit om ‘n pyp te vervang, met spesifieke verwysing na ‘n water verspreidingsnetwerk, kan ingewikkeld wees as gevolg van verskeie aspekte in die besluitnemingsproses, soos beskikbare befondsing en vereistes vir stelsel prestasie. Langtermyn vereistes vir befondsing moet geassesseer word vir die effektiewe bestuur van ‘n bestaande water verspreidingsnetwerk om ‘n balans te vind tussen die opbrengs op belegging en verbruiker tevredenheid. Verskeie voorspellingsmodelle is beskikbaar om die waarskynlikheid van faling vir elke pyp in die water netwerk te bereken. Die waarskynlikheid van faling word dan gebruik om dienooreenkomstig die prioriteit van vervanging vir alle pype in die netwerk te bepaal. Navorsing toon dat die keuse en uitvoering van voorspellingsmodelle, vir faling, sensitief is ten opsigte van die beskikbare data en in baie gevalle word ‘n hoë graad van kundigheid verlang om die resultate voldoende te verstaan. Semi-kwantitatiewe risiko assesserings bied ‘n gestruktureerde wyse om die pyprang te bepaal deur die waarskynlikheid en gevolge van faling in ag te neem en terselfde tyd aanpasbaarheid tot die beskikbaarheid van data te bied. Ten einde die voordele van die risikogebaseerde benadering te gebruik, was ‘n multi-tydperk vervangingsmodel ontwikkel om ‘n geskikte langtermyn beleggingstrategie te bepaal, inaggenome sommige praktiese oorwegings. Die ontwikkelde model het gebruik gemaak van die risiko-gebaseerde benadering om die vervangingsprioriteit van elke pyp te bepaal. Die model neem elke pyp teenwoordig in ‘n pyp inventaris databasis in ag, gebaseer op verskeie bydraende pyp eienskappe en die beskikbare begroting. ‘n Algoritme vir die vooruitskatting van falings was ook by die model ingesluit. Die model kan gebruik word om die benodigde begroting te bepaal gebaseer op sekere vaste inset parameters soos die totale lengte pyp wat vervang moet word of die totale aantal falings wat toegelaat word per jaar. Vier hipotetiese belegging scenarios was geanaliseer vir ‘n gevallestudie. Die resultate was dan vergelyk met ‘n vyfde scenario, bekend as die reaktiewe strategie, waar geen pyp vervanging plaasgevind het nie. Vir die spesifieke geanaliseerde gevallestudie het die reaktiewe strategie die laagste kumulatiewe uitgawes getoon. Addisionele belegging was benodig om die restasieaanwysers vir die aantal falings, diens onderbrekingsduur, geskatte nuttige oorblywende lewensduur en verwagte oorblywende batewaarde. Hierdie navorsing het ‘n metode aangebied waar die relatiewe gewig toegeken aan die verskeie prestasie-aanwysers, soos hierbo genoem, gebruik word om ‘n beste-pas indeks te bereken.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/98606
This item appears in the following collections: