Probabilistic design of gravity based foundations
Translated title
Probabilistisk design af gravitationsfundamenter
Authors
Poulsen, Anders Søndergaard ; Randers, Mads Bredsgaard ; Sørensen, Torben
Term
4. term
Education
Publication year
2012
Submitted on
2012-06-01
Abstract
This thesis investigates the potential benefits of using a probabilistic approach to design gravity-based foundations compared with conventional deterministic methods, and examines the effect of accounting for spatial soil correlation. The study focuses on a surface footing for an onshore wind turbine and uses two test sites representing sand and clay. Soil properties are collected and analyzed, and a spatial analysis indicates short correlation lengths of approximately 0.1–0.5 m vertically and 1.5–2.5 m horizontally. Deterministic designs are performed for both soil types, followed by calculation of reliability indices, which show compliance with current requirements. Building on this, a probabilistic analysis is used to explore calibration and reduction potential for partial safety factors. The findings indicate that partial safety factors depend strongly on the specific design context and the available information on soil characteristics, yet there is a substantial potential for reduction when spatial variability is included.
Dette speciale undersøger, hvilke gevinster der kan opnås ved at dimensionere gravitationsfundamenter probabilistisk i forhold til de traditionelle deterministiske metoder, samt hvilken betydning det har at inddrage jordens rumlige korrelation. Udgangspunktet er et overfladefundament til en onshore vindmølle, analyseret for to testlokaliteter med henholdsvis sand og ler. Jordens egenskaber indsamles og analyseres, og en rumlig analyse estimerer korte korrelationslængder på ca. 0,1–0,5 m vertikalt og 1,5–2,5 m horisontalt. For begge jordtyper gennemføres deterministiske dimensioneringer, hvorefter pålidelighedsindekser beregnes; disse viser, at kravene i gældende praksis overholdes. Dette danner grundlag for en probabilistisk analyse, der belyser kalibrering og reduktionspotentiale for partialkoefficienter. Resultaterne peger på, at partialkoefficienterne er stærkt afhængige af den konkrete designsituation og af tilgængelig viden om jordens karakteristika, men at der trods dette kan være et betydeligt potentiale for reduktion, når rumlig variabilitet inddrages.
[This apstract has been generated with the help of AI directly from the project full text]