Risk based dam safety in Namibia : a quantitative approach

Cloete, Gert Christiaan (2015-03)

Thesis (PhD)--Stellenbosch University, 2015.

Thesis

ENGLISH ABSTRACT: A flood event in the town of Mariental, in 2006, raised a sudden awareness regarding the state of dam safety in Namibia. Although damage was caused at the town, the flood was not extreme; it was approximately a one in fifty year event. The concern, however, was the increase in risk imposed on the town due to the temporary malfunctioning of the back-up power system: should the secondary back-up system also have failed, the embankment could have overtopped with subsequent failure; a catastrophe. The Rational Quantitative Optimal (RQO) approach, presented in this dissertation, provides a robust risk evaluation model which produces a definitive result for the reduction of risk from the overtopping of earth-fill dams. The model is based on principles of risk, but an assessment of a portfolio of dams provides discrete optimal results, not expressed in terms of probability. All the steps that the methodology comprises have been developed exhaustively and propose to address concerns raised by dam owners and decision makers regarding risk-based dam safety: a transparent framework for decision making related to public safety, which will also appeal to the technically minded portfolio manager looking for a purely quantitative procedure to assist in the decision making process. The RQO process is applied mechanistically, not requiring judgement from the decision maker. It thus addresses the concern raised by dam owners regarding the probability of risk assessment being judgmental. Risk in this dissertation is associated with embankment dams and concomitant external erosion, which globally is the single largest cause of failure of these dams. This specific failure mechanism, in particular, is a threat in Namibia, since other mechanisms, such as internal erosion, poses very little risk to the type of embankment dams typically found in Namibia. Therefore, for practical purposes, the extreme flood hydrology in Namibia is revisited and applied to real dams in the RQO model. Extreme flood hydrology in Namibia has, for the past thirty years, largely been based on the South African Department of Water Affairs Technical Report 137 (TR 137) of 1988; This report proposes an empirically established upper limit of flood peaks, called the Regional Maximum Flood (RMF), which is associated with an annual recurrence interval of 10 000 years, as shown in this study from probabilistic analysis which included palaeoflood data. The updated flood model incorporates thirty years of additional systematic data, as well as palaeoflood data that has resulted from a new approach. The new data have provided an increase in the K‒value boundaries for some of the regional flood zones. A revised graphical distribution of the K‒value zones for Namibia is presented and is proposed as a replacement for the current model.

AFRIKAANSE OPSOMMING: ‘n Vloed in die dorp Mariental, in 2006, het belangstelling in damveiligheid in Namibia aangewakker. Alhoewel skade aangerig is aan die dorp, was dit nie ‘n besonderse groot vloed nie; dit was ‘n vloed met ongeveer ‘n vyftig jaar herhalingsperiode. Kommer met betrekking tot die voorval spruit uit die toename in risiko weens die bystand kragaanleg wat gefaal het toe dit nodig was. Indien die tweede bystand stelsel ook nie gewerk het nie, kon die dam se wal oorstroom het, wat tot katastrofiese faling van die dam kon gelei het. Die Rasionele Kwantitatiewe Optimale benadering (RQO) vir damveiligheid, wat verlaging in risiko teen oorstroming van grondvul damme teweegbring word hier voorgestel. Die model is gebaseer op beginsels van risiko analise, maar die resultaat vir ‘n portefeulje van damme word uitgedruk nie in waarskynlikheidsterme nie, maar in terme van ‘n diskrete optimale antwoord. Die metode is in diepte ontwikkel en spreek onsekerhede aan waarvoor dam-eienaars en besluitnemers te staan gekom het; ‘n deursigtige besluitnemings proses wat die veiligheid van die publiek eerste stel, en wat ook aanklank sal vind by ‘n tegnies georiënteerde bestuurder wat ‘n kwantitatiewe oplossing soek vir besluitneming by ‘n portefeulje van damme. Die RQO proses is meganisties in sy toepassing; dit verg geen oordeel van die besluitnemer nie. Sodoende spreek dit ‘n bekommernis aan wat menige dam-eienaars het oor die onpartydigheid of onbevooroordeeldheid in risiko besluitneming. Risiko word in hierdie studie geassosiëer met grondvul damme en eksterne erosie. Eksterne erosie is op internasionale vlak die grootste enkele oorsaak van faling van grondvuldamme. Hierdie falingsmeganisme is ook die grootste risiko van faling wat in Namibia voorkom, aangesien interne erosie nie by rotsvul damme, wat tipies in Namibia gebou word, ‘n groot risiko inhou nie. Dus, vir die praktiese toepassing van die RQO metode, word die ekstreme vloedhidrologie van Namibië ook onder oënskou geneem. Die afgelope dertig jaar is vloedhidrologie in Namibie hoofsaaklik gebaseer op die streeksmaksimum vloed metode wat deur Kovács (1988), van die destydse Suid Afrikaanse Departement van Waterwese, opgestel is vir lande in suidelike Afrika. Dit is beskryf in die tegniese verslag, die TR 137 van 1988, van die Departement van Waterwese, Suid Afrika. TR 137 stel ‘n streeksverbonde empiries-gebaseerde boonste limiet vir vloede voor, die sogenaamde Streeks Maksimum Vloed (SMV). Hierdie studie het gevind dat die SMV vloede tipies ‘n 10 000 jaar herhalingsperiode het deur ‘n waarskynlikheidsontleding te doen van die vloeirekords en palaeovloeddata. Die opgedateerde SMV vloedmodel vir Namibie sluit in dertig jaar se addisionele aaneenlopend-gemete data, asook nuwe palaeovloed data. Die nuwe data vergroot die areas van sommige van die die K-sones, wat die streeksvloed sones voorstel. ‘n Hersiene kaart met die nuwe K-sone grense daarop aangedui word deur hierdie navorsing aangebied en word voorgestel as ‘n vervanging van die Kovács-SMV-kaart van Namibië van 1988.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/96702
This item appears in the following collections: