INVESTIGATIONS ON GALLOPING OF NON-CIRCULAR CROSS SECTIONS

Bundle arrangements are currently used in the design of underwater riser towers or oil export lines. A bundle is made of several parallel pipes linked together at intervals. Even if individual pipes are of circular section, the global external cross-section seen by the fluid is non-circular. When placed in a current, a bundle may be therefore prone to plunge instability [6], also know as "galloping". When designing such bundle's section, it is important to be able to predict its susceptibility to galloping and what are the implications on the whole structure. Galloping is taking place in a low frequency range compared to VIV but with larger amplitude, up to several diameters. Instability can also occur in torsion through a coupling effect with transverse oscillations. Riser Vortex Induced Vibrations have been studied for decades, and numerous experiments have been performed both in-situ and in model test facilities to understand and predict the instability of a slender cylindrical structure in current. The main motivation is the consequences of VIV on riser fatigue life. If galloping and related instabilities are well known in aerodynamics [9], no large specific experiment or study exists for hydrodynamic flows [1], [7]. Therefore no guidelines exist to help prevent or predict galloping while designing cross-sections and pipe arrangements. Until recently, only the Blevins criteria [1] were available to predict the risk of instability. Based on recent examples of riser tower, experimental and numerical investigations are carried out within the "Gallopan" project in the frame of CITEPH (Concertation pour l'Innovation Technologique dans l'Exploration Production des Hydrocarbures) [11]. The main objective is to propose guidelines to avoid or reduce the risk of galloping in bundle cross section design. Two cross section shapes are investigated, a square cross section for which results are available in the literature [1], [10], and a bundle cross section specifically designed to be unstable. Model tests are performed in two steps: Captive tests and transverse forced oscillation tests in steady current to derive hydrodynamic coefficients; Free oscillations tests in steady current to identify the range of reduced velocity where instability occurs as well as the response amplitudes. A specific experimental arrangement, based on a vertical pendulum system is used. Numerical investigations are focused on the use of a standard riser analysis tool. Hydrodynamic coefficients issued from experiments are introduced. Model test set-up is reproduced for comparison purpose.