Offshore gravity foundations are often designed with complex geometries. Such structures interact with the local hydrodynamics and generate enhanced bed shear stresses and flow turbulence capable of scouring the seabed or destabilizing bed armour where deployed. In the present study a novel bed shear stress measurement method has been developed from the camera and laser components of a Particle Image Velocimetry (PIV) system. The bed shear stress amplification was mapped out around six models of gravity foundations with different geometries. Tests were repeated for two bed roughness conditions. The structures tested included uniform cylinders, cylindrical base structures and conical base structures. The flow field around the models was also measured using PIV.
The results of this study reveal that the conical base structures generate a different hydrodynamic response compared to the other structures. For uniform cylinders the maximum bed shear stress amplification occurs upstream, at an angle of 45° relative to the flow direction, and measurements were found to agree well with numerical results obtained by Roulund et al. (2005). In the case of the cylindrical base structure the maximum amplification occurs upstream at a similar location to the uniform cylinder case. For the conical base structures the maximum amplification of the bed shear stress occurs on the lee side of the structure, with the magnitude dependent on the side slope of the cone. The bed shear stress results were validated against stresses derived from analysis of the flow fields obtained by the PIV measurements performed under the same test conditions.
Conclusions from the study are that the structure with the cylindrical base foundation produces the lowest bed shear stress amplification and that an increase in the bed roughness results in an increase in the amplification of the bed shear stress. These findings have direct implications for design of scour protection. In addition the flow reattachment point behind the foundation is dependent on pile Reynolds number (ReD). This suggests that the results of this study may be extrapolated for higher pile Reynolds using the method described in Roulund et al. (2006).