Prediction of VIV for Risers With Effective Strakes

Abstract

Accurate prediction of straked riser response can be important for determining the minimum strake coverage needed to ensure adequate riser fatigue life in demanding high-speed current environments. Previously published results from ExxonMobil’s 2003 vortex-induced vibration (VIV) model tests on a long, flexible, straked cylinder showed that strakes can greatly reduce, but may not entirely eliminate flow-induced vibration. The data also showed that the low-amplitude vibrations that persist when the cylinders are straked are different in character from VIV on bare cylinders, suggesting that the strakes are effective in disrupting the alternating vortex shedding mechanism that excites bare-cylinder VIV and that other flow mechanisms may drive the straked-cylinder vibration. This further implies that classic VIV models that have been developed to predict bare-pipe response due to alternating vortex shedding may not be accurate for straked cylinders even if the excitation model (“lift curve”) is adjusted to represent lower excitation levels on strakes. Classic VIV prediction methods for bare risers rely on special-purpose theoretical models to account for the strongly nonlinear interactions between bare cylinder motion and hydrodynamic excitation. However, for the low amplitudes of vibration observed in tests with straked cylinders, the influence of the cylinder motion on the flow may be less important. In an effort to more accurately predict the performance of strakes, ExxonMobil has recently explored a new approach for predicting flow-induced vibrations for straked risers that neglects any influence of riser motion on excitation, and instead assumes that hydrodynamic excitation can be approximated to useful accuracy by excitation measured on a fixed straked cylinder. New experiments show that this excitation is stochastic, with broad band-width, and that it conforms to relatively simple scaling laws. The measured stochastic excitation data has been used with a conventional linear random vibration model to predict the response data acquired from the 2003 model tests on the long, flexible, straked cylinder. Comparison shows a dramatic improvement in prediction scatter compared to predictions using a classic bare-pipe VIV formulation with a “lift curve” model of the strake hydrodynamic excitation.