Predicting Spar VIM Using CFD

Recent work toward predicting spar vortex induced motion (VIM) with computational fluid dynamics (CFD) suggests that such simulations can anticipate many aspects of spar response and thus supplement tow tank experiments and other design methods. However, the results also highlight a number of challenges as well. The spar VIM problem is characterized by very high Reynolds numbers, geometric complexity including the presence of numerous external appendages and the presence of very rough surfaces. In this paper, we first review recent work on spar VIM where CFD was used to simulate tow tank experiments. This work suggests that CFD methods give good results in most cases but also points to some exceptions. In particular, in simulations of small scale vortex induced motion tests of spars, good agreement between analysis and experiments is usually obtained when the flow separates from the spar hull at the strakes. The CFD simulations are sometimes less successful at predicted VIM when flow separation occurs at the spar hull. We then examine our own recent practice in simulating tow tank experiments with CFD with the objective of finding possible modeling deficiencies. The focus is on the resolution of the large eddies in the wake which most influence the fluctuating loads on the spar, but we are also concerned with the use of wall functions to model the boundary layer. All of the calculations use detached eddy simulation (DES). In order to test the method, we make use of wind tunnel experiments at on a fixed truncated cylinder without strakes. The wind tunnel experiments are performed at Reynolds numbers (Re) that are about the same as those used in scale model spar VIM experiments. Wake particle image velocimetry (PIV) and other data from wind tunnel experiments published in the open literature are used for comparison. The comparisons are used to examine requirements for grid resolution in the wake. Finally, it is suggested that specific wind tunnel experiments might be used to gather needed data on the effects of rough walls and appendages at very high Reynolds numbers. BACKGROUND Spars and other floating platforms are geometrically complex structures which are often festooned with external pipes, chains and anodes as shown in Figure 1. The wetted surfaces of a full scale spar are also typically rough although the model spars may be smooth. Typical tow tank tests of model spars are undertaken at Reynolds numbers on the order of 1e5 so a model spar with a smooth surface will probably have a laminar boundary layer up to the point of flow separation. A full scale spar may operate at Re in the tens of millions and will have a turbulent boundary layer before flow separation.