CFD benchmark study from the Wind Load Joint Industry Project

AuthorsWilde, J. de, Schrijvers, P., Witz, J. (Witz Ltd.)
Conference/JournalOffshore Technology Conference (OTC), Houston, Texas, USA
DateMay 1, 2017
DOIhttps://doi.org/10.4043/28832-MS
Wind loads represent an important contribution to the design loads of large floating offshore structures during installation, normal operation and in extreme conditions. It impacts the intact stability of the floating structure as well as its mooring loads. In many cases the wind loads are even the dominant load factor. Wind loads can cause large overturning moments, which may have important safety implications during installation and operation. Planning of required tug force capacity for offshore installation, maneuvering, berthing and offloading activities may often strongly depend on reliable wind load predictions. Traditionally, the wind loads acting on floating offshore structures are studied experimentally in wind tunnels, but recent developments in Computational Fluid Dynamics (CFD) make it possible to also compute the wind flow around structures and to also derive the wind loads numerically. In general, the numerical and experimental approaches to establishing wind loads may be considered to be complimentary methods.

The Wind Load JIP, in 2014, was initiated with the aim of investigating the accuracy level and consistency of wind tunnel tests and CFD predictions for offshore applications. Wind tunnel tests are commonly required towards the end of the design process and provide the final wind loads for qualification. However, there is today no common industry guidance for setting-up a wind tunnel test experiment for large blunt bodies in an atmospheric boundary layer flow. There is particular uncertainty on how to set up the inflow conditions for the undisturbed atmospheric boundary layer velocity profile and associated turbulence intensity. End users of wind tunnel tests have reported occasions where unexpected relatively large variation in forces was observed for fairly similar geometries. For example, large variations in wind forces have also been reported by Croonenborghs et al. (2013) and Andrillon et al. (2015).
The Wind Load JIP facilitated a carefully planned benchmark study for wind tunnel tests. Contributions were obtained from three established large boundary-layer wind tunnel test facilities in Europe. A scale 1:230 model of a modern FPSO design with complex topside geometry was tested on the force balance in the wind tunnel. The tests were conducted with carefully adjusted ISO/Frøya velocity profiles, for a target wind speed of respectively 25 m/s and 45 m/s at 10 m reference height. The ‘blind’ wind tunnel tests were planned in such a way that the wind tunnel test facilities were not informed of each other’s 2 OTC-28832-MS results until after handing in their contribution. The scale 1:230 model was shipped around to be sure of testing the exact same geometry.

The JIP also facilitated a CFD benchmark study for the same geometry. The CFD benchmark study was performed ‘blind’ as well and the CFD providers were not informed of each other’s results and also not of the wind tunnel test results. The different CFD participants used different choices for the numerical settings in their calculations. Commonly they opted for steady RANS, mesh size around 30M cells, unstructured meshes, meshes generated with mesh wrapper functionality, low y+, avoidance of wall functions on the test geometry, Menter SST k- turbulence model and about ten thousand iterations for convergence. The CFD participants had the choice between modeling the exact FPSO geometry as tested in the wind tunnel, or a simplified geometry with equivalent wind area. Both geometry variations were used.

The CFD benchmark study showed that CFD can predict the wind tunnel test results fairly well. The shape of the wind load coefficient curves was adequately predicted by all CFD contributions for all six degrees of freedom (CX, CY, CZ, CNX, CNY and CNZ) and for all wind headings between 0 to 360 degrees. The scatter between the CFD contributions was about 10%, which is not too much different from the scatter in the wind tunnel test results. Apart from the random scatter, a 5 to 10% systematic error was observed between the CFD results and the wind tunnel test results. The CFD results slightly under predicted the wind tunnel test results.

This paper is copyrighted material and cannot be shared. Due to copyright policy MARIN is not allowed to reproduce and distribute papers written for and presented at the OTC (Offshore Technology Conference), ISOPE (International Ocean and Polar Engineering Conference) and SPE (Society of Petroleum Engineers). You can order these papers through
www.OnePetro.org.

For this specific paper follow this link.
Tags
cfd