Characterisation of the South African extreme wind environment relevant to standardisation

Bakker, Frederik Pierre (2021-03)

Thesis (PhD)--Stellenbosch University, 2021.

Thesis

ENGLISH ABSTRACT: To ensure adequate structural resistance, engineers estimate the largest loads that windcan reasonably be expected to induce in a structure. In South Africa, operational designpractices are described in a set of standards, which include a wind loading standard (SANS10160-3) that provides guidance to engineers in estimating wind loads. The estimationprocedure described in the standard uses a basic wind speed, provided in a characteristicmap, that accounts for what the climate could produce at a particular location. This mapis based on inference from available wind speed observations using stochastic models.In an effort to improve the background information on which the standard is basedand in response to greater volumes of wind speed data, this dissertation presents a setof investigations which incorporate a greater quantity of relevant data, strive to improvethe stochastic model, and make specific recommendations that could be included in thenext revision of the South African wind load standard.A background study that summarises some of the most salient theory is also presented.This includes an overview of the South African extreme wind climate, extreme valuetheory, wind load standardisation, and reliability-based design.The quantity of available data collected by the South African Weather Service hasincreased substantially since the current standard was derived. This data had to bechecked for errors and classified by climatic mechanism. Following an attempt to correctfor the dynamic response of outdated instruments, the inclusion of historical data datingas far back as 1948 was also investigated. To be relevant for use in the standard, allthe data had to be measured according to World Meteorological Organisation standards.This was typically not the case, and so surface roughness correction was applied using aGeographical Information System technique. The expanded dataset includes data from an extensive network of 132 stations through-out South Africa, but the use of only annual maxima resulted in insufficient samplesizes at individual sites, potentially introducing substantial model-variance. In response,improved estimates by incorporating data from multiple sites within the region of studywere determined in this thesis. This included regional estimation of the shape of theprobability distribution, based on preconditioning the data by an exponent; and a separateinvestigation on optimal estimation of variability. For estimation of the variability, it wasshown that there was a balance between site and regional estimation, which was used todevelop an optimal estimator based on the bias variance trade-off.The expanded dataset and the insights into regional estimation were then applied toestimate design wind speeds that should satisfy the target reliability specified in the SouthAfrican standard. These estimates were incorporated into a new characteristic wind speedmap using a Monte Carlo simulation-based hypothesis test.

AFRIKAANSE OPSOMMING: Om ’n struktuur te onwerp met voldoende weerstand, voorspel ingenieurs die kritiesewindlading wat op die struktuur kan inwerk. In Suid-Afrika word die ontwerp proseduresbeskryf in ’n stel standaarde wat ’n windlading standaard insluit (SANS 10160-3) waarindie voorspelling van windlading beskryf word. Hierdie voorspellings prosedure is gebaseerop ’n basiese wind spoed, wat op ’n windkaart beskyf is, wat voorsiening maak vir dieklimaat by spesifieke liggings. Hierdie windkaart is gebaseer op afleidings van beskikbarewind spoed waarnemings en stogastiese modelle.In ’n poging om die Suid-Afrikaanse windlading standaard te hersien, voorsien hierdieverhandeling ’n stel studies wat meer data gebruik om die stogastiese model te verbeter,asook nodige aanbevelings wat in die volgende hersiening van die Suid-Afrikaanse wind-lading standaard inkorporeer kan word. n Agtergrond studie wat ’n opsomming van die belangrikste teorie ̈e bevat wordook aangebied. Dit bevat ’n oorsig van die Suid-Afrikaanse uiterste/sterk windklimaat,statistiese benaderings om uiterstes te beraam, standaardisering van windladings asookbetroubaarheids-gebasseerde ontwerpe.Die hoeveelheid data wat deur die Suid-Afrikaanse weerdiens versamel is, het aan-sienlik toegeneem sedert vorige studies wat gebruik was om die huidige standaard teontwikkel. Hierdie data moes deur ’n kwaliteitsbeheer proses gaan en volgens verskillendetipe uiterste/sterk windklimate geklassifiseer word. Na ’n poging om die dinamiese reaksievan ouer instrumente in ag te neem, is die insluiting van windspoed meetings van so verterug as 1948 ondersoek. Om vir die standaard gebruik te kan word, moet al die datavolgens die Wˆereld Meteorologiese Organisasie se standaarde gemeet word. Omdat ditnie gewoonlik die geval was nie, moes ’n korreksie van die grondwrywing toegepas wordmet behulp van Geografiese-Inligting-Stelsel tegnieke.Die finale datastel bestaan uit data vanaf ’n netwerk van 132 weerstasies regoor Suid-Afrika. Die gebruik van jaarlikse maksima beteken dat individuele weerstasies problemeervaar met klein steekproef grotes, wat onaanvaarbare model-variansie kan lewer. Asgevolg van die bogenoemde, is data van verskeie weerstasies in die omgewing gebruikom die voorspellings te verbeter. Die toepaslike streeksverspreidings is ook bepaal, watgebaseer is op die metode van maksimum waarskynlikheid. In terme van die beramingvan veranderlikheid, dit getoon dat daar ’n balans tussen individuele en streeks beramingbestaan, wat gebruik is om die optimale beramer te ontwikkel.Die verlengde datastel en insigte oor streeks voorspelling is gebruik om die ontwerpwind snelhede te bepaal wat die betroubaarheid van die Suid-Afrikaanse standaarde salonderhou. Hierdie voorspellings is dan opgeneem in ’n nuwe windspoedkaart met behulpvan ’n hipotesetoets gebaseer op Monte Carlo simulasies.

Please refer to this item in SUNScholar by using the following persistent URL: http://hdl.handle.net/10019.1/109823
This item appears in the following collections: