Abstract A178

CFD Simulation of Riser VIV
by
Hamn-Ching Chen, Ocean Engineering Program, Department of Civil Engineering,
Texas A&M University
Chia-Rong Chen, Department of Mathematics, Texas A&M University
Richard S. Mercier, Ocean Engineering Program, Department of Civil Engineering, Texas A&M University
Final Project Report
December 2006

Offshore oil/gas drilling and production operations face hazards due to the exposure of submerged rig components to underwater sea currents. Critical among these components are the marine risers, consisting of a series of long steel pipes of circular cross-section, used for deep-water extraction of oil and/or natural gas. These long cylindrical structures, exposed to strong sea currents, induce the flow around them to separate and initiate vortex shedding – whereby vortices of opposite sign are shed synchronously from the aft of the structure. The resultant lift and drag forces excite forced oscillations of the riser, known as vortex-induced vibrations (VIV). When the VIV frequency is close to one of the natural frequencies of the structure, a resonance phenomenon popularly known as “lock-in” occurs, resulting in enhancement of the vibration amplitude of the structure – and thus its destructive potential.

Since the VIV phenomenon is quite complex to model, the analysis of fluid-induced motions in deepwater has remained a somewhat empirical science. Currently, the design criteria and tools for the operation of marine risers under lock-in conditions rely on simplified mathematical models and/or anecdotal evidence. Increasing development costs and increasingly hostile field environments demand more robust, reliable, and refined design strategies and tools. This need has renewed interest in the study of the VIV phenomenon by experimentalists and numerical analysts.

Accurate fatigue life prediction requires knowledge of the frequency and magnitude of the vibration, in addition to average forces acting on the moving structure. Improved design principles for risers require an approach that can supplement existing empirical knowledge and experimental data. A promising approach is to use computational fluid dynamics to provide studies on fluid-induced motions of structures.

In operating conditions, the marine risers are subjected to flow conditions that correspond to high Reynolds number flows, typically O(105) - O(107), and low structural mass ratio and damping. These conditions are difficult and expensive to achieve in an experimental setup, and thus an attractive alternative is to use computational fluid dynamics (CFD) tools to study/predict the response of the riser undergoing VIV. Unfortunately, well-established numerical procedures for CFD (e.g., finite elements, finite differences, and control volumes) are not robust with respect to high Reynolds number flows. As the Reynolds number increases, these methods are prone to developing spurious node-to-node oscillations – which eventually cause the numerical solution procedure to break down. The oscillations can be suppressed by refining the mesh, but the necessary degree of refinement is often prohibitively expensive. In view of this drawback, most of the previous work on simulations of the VIV phenomena have been limited to low Reynolds number flows, O(102) – O(103).

The objective of this research is to develop advanced CFD capabilities for improving the prediction of riser VIV responses at high Reynolds number. The performance of VIV suppression devices such as helical strakes and fairings is also evaluated. This report documents the following tasks accomplished under this project:
• Development of advanced CFD capabilities: numerical method and parallelization
• 2-D simulations of flow past a fixed riser at high Reynolds numbers
• Surface roughness effects
• 2-D simulations of elastically mounted risers undergoing VIV at high Reynolds number: single isolated riser and arrangements of multiple risers
• 3-D Large Eddy Simulation of flow past a fixed riser
• 3-D Large Eddy Simulation of an elastically mounted riser undergoing VIV
• Simulations of an elastically mounted riser outfitted with a fairing
• Simulations of an elastically mounted riser with helical strakes

The main thrust of this work is use of a novel and robust discretization procedure whereby the Navier-Stokes equations are discretized using local analytic solutions of the linearized problem (Pontaza, Chen and Reddy, 2005; Chen, Patel and Ju, 1991). The formulation is robust with respect to high Reynolds number flows, strong mesh skews, and is asymptotically second-order accurate. In addition, the formulation is implemented alongside the ability to treat embedded and non-matching grids, allowing for relative grid motion. This capability makes fluid-structure interaction problems easy to handle, as tedious grid re-generation is avoided by virtue of the ability to treat embedded meshes moving in fixed background grid components. Thus, the formulation is suitable to simulate the VIV phenomena – inherently a fluid-structure interaction problem at high Reynolds numbers.

In practical applications, the Reynolds number is large (typically O(105) - O(107)) and flow conditions are transitional or turbulent. Thus, the flow exhibits complex three-dimensional features, making analysis difficult and computationally expensive for direct numerical simulations of the governing incompressible Navier-Stokes equations. A more practical first-stage approach is to assume that the flow can be treated as two-dimensional and perform high resolution simulations at this level. These simulations are of great practical importance as they shed light on first-principles of fluid-induced motions and may improve design principles for deepwater risers. Two-dimensional simulations are presented in the first part of this work. In later stages of this work we will discard the two-dimensionality assumption, and perform fully three-dimensional simulations of short aspect ratio risers (L/D = 3 and 9). In real world applications, deepwater risers can easily exceed aspect ratios of L/D = 1000. For example, Holmes, Constantinides and Oakley (2006) presented fully-3D simulation results for a bare riser with L/D = 1400 and a straked riser with L/D = 1151. The computer resources currently available at Texas A&M University do not permit adequate resolution of fully 3D VIV of such L/D ratios since each user is allowed to use no more than 16 processors at a time. However, it should be remarked here that the present method is currently being generalized in a new research project (2006-2008) for the simulation of flexible long risers.

In a recent study, we have successfully performed fully 3D simulations of a flexible riser with L/D = 1400 using very large grid spacing (L/D ~ 50) in the spanwise direction. These preliminary 3D simulations clearly demonstrated the feasibility of using the present CFD code in conjunction with a finite element tensioned beam model for simulation of flexible long risers even though the fluid interactions in the spanwise direction were not properly resolved with such a large spanwise spacing. With the availability of additional computer resources in the future, it is anticipated that the parallel computer program developed under this project can be readily employed for truly 3D simulations on large computer clusters with hundreds of dedicated processors.