On the use of hindcasts in SPM design:

by K. Oostra

Institution: Delft University of Technology
Year: 2015
Keywords: Atmospheric models; Wind-Wave models; OGCM; Response based approach; Extreme Value theory; Single point mooring (SPM); Peak Over Treshold (POT); MetOcean practices
Record ID: 1251747
Full text PDF: http://resolver.tudelft.nl/uuid:10466713-f265-4046-b02d-779d25300582


Due to confidentiality only the cover and abstract of the thesis are available. In offshore design 100yr RP (Return Period) extrapolations for wave-height, wind- and current-speed serve as ULS (Ultimate Limit State) design-parameters. These extrapolations are produced by performing statistical inference on comprehensive hindcasted MetOcean time-traces. Nowadays, offshore contractors like Bluewater receive these time-traces with MetOcean-reports. Up to now Bluewater only used these time-traces for fatigue analysis. In this study it has been examined if and how they can also be of use in ULS design. The study roughly consists of two parts. The first part being an analysis of the MetOcean models and general practices of MetOcean-specialists. Wind-wave, current- and atmospheric-models, often combined in coupled atmosphere-ocean GCM’s, are assessed on their interaction, workings and underlying assumptions. In the analysis of MetOcean practices attention is directed at the way MetOcean-specialists perform inference on the data produced by the models. Special attention is paid to the verification/validation of MetOcean-data, the use of Extreme Value (EV) theory and how covariate effects like directionality, temporal- and spatial-variability are dealt with. With the aim to understand how MetOcean-specialists translate the information-rich time-traces in a very limited number of ULS environments. In the second part of the study knowledge obtained in the elaboration of part one was used to see how MetOcean time-traces could be of use in ULS design. For this several variations of threshold models were used to select relevant extreme cases from a 30yr MetOcean hindcast for Pierce Field (UK sector North Sea), where Bluewater’s FPSO Haewene Brim is located. This field was chosen for analysis because a detailed time-trace was available from a reputable MetOcean-office (Shell Global Solutions (UK) Ltd.) and because DYNFLOAT-models (modelling software developed by MARIN) of the FPSO Haewene Brim were available for this field. Choosing this location would also enable comparison to loads resulting from the current design method, since design-documents for Pierce Field were available. Following the Peak Over Treshold (POT) method, as described by (Coles, S. “An Introduction to Statistical Modeling of Extreme Values” Springer Series in Statistics (2001)), a Generalized Pareto Distribution (GPD) was fitted to the response-data that resulted from modelling the FPSO Haewene Brim in the sea-states selected by the various threshold models. Subsequently the fitted GPD’s were extrapolated to find 100yr RP maximum Line Tensions (LT’s). These extrapolated LT’s were compared to the maximum LT’s that occur when modelling the FPSO in the governing 100yr RP environment (taken from mooring design-documents developed by Bluewater’s Mooring & Subsea department). Differences are particularly small (-3.83%) for the threshold models that model entire storms and take the maximum LT of the storm. Results indicate that maximum LT’s often occur in…