CFD Simulation of Ocean Turbulence Interactions
With Spar Platforms
OBJECTIVE:Little quantitative information is available on ocean turbulence in storm and loop current environments at scales that might exert significant forces on offshore structures (say, 6 - 600 second time scales and 2 - 30 meter length scales). Field measurement programs currently being planned by others will provide this information. This project will address the forces that ocean turbulence exerts on offshore structures, initially under the assumption that ocean turbulence can be represented using simple isotropic geophysical turbulence models. Previously developed and validated CFD modeling techniques for studying interactions of non-turbulent incident flow with rigid and flexible cylinders will be extended to include the effects of upstream turbulence. The CFD model will be used to simulate the low frequency and VIV-induced motions of a spar platform in current environments as a function of the upstream turbulence structure.
APPROACH:
The general methodology behind the proposed work is to solve the incompressible Navier-Stokes equations numerically using techniques from the area of computational fluid dynamics (CFD). Previous work has demonstrated that this approach may be used to accurately simulate the interaction of non-turbulent incident flow with rigid or flexible cylinders of varying degrees of surface roughness, with and without strakes.
Extending the existing CFD methods to incorporate a specified upstream turbulence structure will be a challenge, not only because it is not obvious how to mechanically embed and maintain the upstream turbulence structure, but also because of the refined mesh and associated increase in computational burden required to track and evolve the more extensive turbulent domain. Consequently, initial work will focus on 2D simulations of bare cylinders at subcritical Reynolds numbers. The simulations will be set up to track the evolution of the turbulence of the current profile downstream and its interaction with both the boundary layer on the cylinder surface and the wake downstream of the cylinder, with and without VIV effects. To improve the computational efficiency, more sophisticated techniques for representing boundary roughness will be investigated. Rather than the brute force method of representing roughness geometrically, the use of special boundary conditions that emulate roughness-like effects through local generation of vorticity will be investigated. Once these new modeling techniques have been established, more complex and realistic 2D and 3D simulations at supercritical Reynolds numbers for bare and straked cylinders will be studied.
The first step will be to verify that it is possible to simulate numerically (i) turbulence in the current profile and (ii) surface roughness. Dr. Mercier will provide turbulent current profiles for this phase of validation.
In collaboration with Dr. Mercier, the CFD model will then be applied in a series of numerical experiments to investigate the effects of upstream turbulence on VIV of spar platforms. To the extent possible (timely delivery of the data, availability of experimental data that are in a form which is appropriate for CFD validation), the 2-D CFD modeling techniques will be validated using experimental data that may be available from DeepStar through a separate OTRC project. The possibility and number of such validation runs will also depend on the required computing resources for each such case.The 2-D analysis will occupy part of year 2003-2004 work primarily in the area of employing more realistic turbulence profiles. However, year 2003-2004 will focus on the 3-D work as originally planned.
DEPLOYMENT OF RESULTS:
Study results will be shared and validated through interactions with industry focal points as well as through conference and journal publications. Through the ability to directly incorporate turbulence information from field and laboratory measurement programs, the CFD model will provide a rational basis for determining how existing design practice for spars should be modified to include ocean turbulence effects.
ANTICIPATED PROJECT DURATION: 3 yearsPROJECT PLAN FOR YEAR 1 (2002-2003):
Scope of Work: The first year will focus on developing modeling techniques for simulating prescribed turbulence structures in the upstream flow field and for simulating boundary roughness through alternative vorticity-generating boundary conditions. The techniques will be developed and tested using 2D simulations of a bare cylinder (no strakes) that may be elastically restrained to allow VIV motions. The bulk of the simulations will be for subcritical Reynolds numbers with a limited number of simulations in the supercritical regime.
Anticipated Results: Numerical modeling techniques for simulating the effects of ocean turbulence in 2D. Comparisons will be made of simulation results with and without upstream turbulence to provide insight on interaction effects.
PROJECT PLAN FOR YEAR 2 (2003-2004):
Scope of Work: Complete 2D simulations with upstream turbulence. Work will then focus on developing modeling techniques for simulating prescribed turbulence structures in the upstream flow field and for simulating boundary roughness through alternative vorticity-generating boundary conditions in 3D. The techniques will be developed and tested using 3D simulations of a bare cylinder that may be elastically restrained to allow VIV motions. The simulations will be for subcritical Reynolds numbers.
Anticipated Results: Numerical modeling techniques for simulating the effects of ocean turbulence in 3D. Comparisons will be made of simulation results with and without upstream turbulence to provide insight on interaction effects. Preliminary models to account for the effects of ocean turbulence in the design of spar platforms (e.g. through drag & lift coefficients).
PROJECT PLAN FOR YEAR 3 (2004-2005):
If field measurements of ocean turbulence are available, incorporate such information in 3D simulations and compare with results from previous simulations using assumed isotropic turbulence models. Refine preliminary engineering design analysis procedures developed in Year 1. Share results with industry designers to initiate the use of improved modeling procedures that include the effects of ocean turbulence.
SPONSORSHIP: MMS and OTRC Industry Sponsors
PRINCIPAL INVESTIGATOR (S) & OTHERS INVOLVED IN PROJECT:
PI(s): Dr. Spyros Kinnas
Collaborator: Dr. Richard MercierOthers: 1 graduate student
Date: June 2005
Project Title: VIV CFD Analysis
MMS Project: 483 TO Numbers: 73611/35984
PI: John Kallinderis
COTR: A. Konczvald
Estimated Completion Date: August, 2005
Project Description:
This project is addressing the forces that ocean turbulence exerts on offshore structures, initially under the assumption that ocean turbulence can be represented using simple isotropic geophysical turbulence models. Previously developed and validated CFD modeling techniques for studying interactions of non-turbulent incident flow with rigid and flexible cylinders will be extended to include the effects of upstream turbulence. The CFD model will be used to simulate the low frequency and VIV-induced motions of a spar platform in current environments as a function of the upstream turbulence structure.
Extending the existing CFD methods to incorporate a specified upstream turbulence structure is challenging because (1) it is not obvious how to mechanically embed and maintain the upstream turbulence structure, and (2) the refined mesh requires increased computational time to track and evolve the more extensive turbulent domain. Initial work focused on 2D simulations of bare cylinders at subcritical Reynolds numbers. The simulations will track the evolution of the turbulence of the current profile downstream and its interaction with both the boundary layer on the cylinder surface and the wake downstream of the cylinder, with and without VIV effects. To improve the computational efficiency, more sophisticated techniques to represent the boundary roughness by boundary conditions that emulate roughness-like effects through local generation of vorticity will be investigated.
Progress:
Required Time-Step Size for current with turbulence
The time step size required for the simulation of turbulent flow was investigated. In order to determine the necessary time step size to accommodate turbulence disturbances, a uniform inflow case with initial disturbance (rotating the cylinder) was employed and the effect of the initial disturbance was observed. Figure 1 shows that the initial disturbance persists for long and destroys the unsteady solution behavior as the time step becomes large ( ). Excessive fluctuations of and are observed when the time step of is used. This is because the effect of the initial disturbance persists for very long and it affects the overall response of hydrodynamic forces. For the time step , the effect of the disturbance is limited to the initial stage, and does not corrupt the time-accurate solution further. Hence, in order to resolve turbulent fluctuation accurately, all the simulation results are produced with in this work.
High Reynolds number flows with inflow turbulence
The inflow turbulent profile is applied to a series of higher Reynolds number flows, and the hydrodynamic force responses are compared in Figure 2. For high Reynolds number flow simulations, the Spalart-Allmaras one-equation eddy-viscosity model is used. For trans-critical Reynolds numbers (Re=100,000~1,000,000), a reduction of and amplitudes is observed compared to the sub-critical Reynolds number case (Re=1,000). The same inflow turbulence profile is used for all three cases.
As shown in Figure 2, as the Reynolds number falls in the trans-critical regime (Re=100,000~1,000,000), a reduction of and amplitudes is observed. This seems due to the boundary layer transition from laminar to turbulent, and the accompanying delayed separation of boundary layer and narrowed wake region. The general patterns of the hydrodynamic forces exerted on the cylinder are the same as those for the low Reynolds number cases. There is a slight phase shortening observed for the trans-critical Reynolds numbers compared to the Re=1,000 case, and this phase shift may be attributed to the slight increase in the Strouhal number as the Reynolds number increases. However, further investigation of this issue is needed.Reports & Publications:
Y. Kallinderis and H.T. Ahn, “Incompressible Navier-Stokes Method with General Hybrid Meshes,” Journal of Computational Physics. Accepted.
H.T. Ahn and Y. Kallinderis, “A Geometrically Conservative ALE Scheme for Strongly Coupled Fluid Structure Interactions.” Computer Methods in Applied Mechanics and Engineering. In Review.
H.T. Ahn and Y. Kallinderis, “CFD Investigation of the Effect of Ocean Turbulence on the Hydrodynamic Forces on a Cylinder.” 24th ASME International Conference on Offshore Mechanics and Arctic Engineering, June 12-17, 2005, Halkidiki, Greece.
Y. Kallinderis and H.T. Ahn, “Strongly Coupled Fluid-Structure Interactions via a New Navier-Stokes Method for Prediction of Vortex-Induced Vibrations,” 24th ASME International Conference on Offshore Mechanics and Arctic Engineering, June 12-17, 2005, Halkidiki, Greece.
H.T. Ahn., “A new incompressible Navier-Stokes Method with General Hybrid Meshes and its application to fluid-structure interactions,” PhD dissertation, Dept. of Aerospace Engineering & Engineering Mechanics, The University of Texas at Austin, Austin, TX. May 2005.
Date: December, 2004
Project Title: VIV CFD Analysis
MMS Project: 483 TO Numbers: 73611/35984
PI: John Kallinderis and Spyros Kinnas
COTR: A. KonczvaldEstimated Completion Date: August 2005
Project Description: The effect of ocean turbulence on the hydrodynamic forces is currently an important issue for offshore installations in the GOM. There are two focus areas this year: (i) completion of the CFD study in 2-D of the effect of upstream turbulence on the hydrodynamic forces exerted on cylinders, and (ii) study of the effect of upstream turbulence using 3-D CFD simulations. Another issue is the emulation of surface roughness via special boundary conditions. Work is being performed on developing techniques for simulating prescribed turbulence structures in the upstream flow field, as well as on simulations of flows around cylinders with and without turbulence being present in the current profile. One part of the work regards the specification of realistic profiles of turbulence for the CFD simulations. This part of the work is being addressed by a separate project led by Dr. Mercier. Numerical implementation of turbulence necessitates also investigation of required computational meshes and time-steps for appropriate resolution of spatial and time scales of the problem.
Progress:
1. Time-marching accuracy required for current with turbulence
The time scales of the inflow prescribed turbulence require increased accuracy of the time-marching process. There are two parameters governing the time-accuracy; the size of the time step, and the number of sub iterations required by the artificial compressibility method at each time step. Convergence of the Cd, Cl time series with increased number of sub iterations revealed that approximately three times as many are needed compared to the uniform inflow cases.
.
2. Strong vs weak coupling between solution of the turbulence model and the solution of the Navier-Stokes equations
Weak coupling involves staggering the solutions of the Navier-Stokes equations and the Spalart-Allmaras equation, while strong coupling is simultaneous solution of both. Weak coupling incurs non-physical overshoots in the Cd, Cl time series which are not observed when the strong coupling is used. For uniform (no turbulence) inflow there is no such difference between the two coupling techniques.
OTRC PROJECT STATUS REPORTDate: June 2004
Project Name: CFD Simulation of Ocean Turbulence Interactions with Spar Platforms
Project Number: 443 Task Order: 73611
Investigator: Spyros Kinnas
Estimated Completion Date: August 2005
Project Description:
The effect of ocean turbulence on the hydrodynamic forces is currently an important issue for offshore installations in the GOM. There are two focus areas this year: (i) completion of the CFD study in 2-D of the effect of upstream turbulence on the hydrodynamic forces exerted on cylinders, and (ii) study of the effect of upstream turbulence using 3-D CFD simulations. Work is being performed on developing techniques for simulating prescribed turbulence structures in the upstream flow field, as well as on simulations of flows around cylinders with and without turbulence being present in the current profile. One part of the work regards the specification of realistic profiles of turbulence for the CFD simulations. This part of the work is being addressed by a separate project led by Dr. Mercier. Numerical implementation of turbulence necessitates also investigation of required computational meshes and time-steps for appropriate resolution of spatial and time scales of the problem.
Progress:
1. Study of the effect of the mesh scale The 2D mesh was adapted (locally refined) upstream of the cylinder. The two inflow turbulence profiles (Re=150 and Re=1,000) were employed for this finer mesh.
For the low Reynolds number case (Re = 150), both the adapted and original meshes give identical results in terms of Cd and Cl, which means the original mesh is already fine-enough to resolve most of the turbulent eddies which are significant to the hydrodynamics forces (Cd and Cl).
For the higher Reynolds number flow (Re = 1000), the simulations using the adapted mesh show more fluctuations in the Cd and Cl histories when compared to the original (coarser) mesh case. The smaller turbulence scales captured by the finer mesh seem to play a role in the resulting hydrodynamic forces.
2. Effect of inflow turbulence on the frequency of the forces Multiple frequencies appear in the time series of the hydrodynamic forces when the current profile is turbulent.
For the Re = 150 case, both the original and adapted meshes "capture" two major frequencies (two peaks in the frequency spectrum) around the St = 0.2 (one peak is located higher than 0.2 and the other is lower than 0.2).
For Re = 1000, the original mesh captures only two peaks, but the adapted mesh captures four peaks around the St = 0.2 value. It appears that the smaller eddies in this higher Re case do contribute to the frequencies of the force response.3. Inflow boundary condition for the turbulence model The RANS models are based on the assumption that the Navier-Stokes equations are being solved for the mean flow, which means the velocity components in the N-S equations are mean-velocities and the fluctuations are accounted for by the turbulence model equation. However, the current inlet velocity is the mean velocity plus the turbulence fluctuations (u = U + u'), which implies that the N-S equations already include the effect of turbulent fluctuation at the inflow boundary of the computational domain. The inflow boundary condition for the turbulence model should be compatible to the turbulence inflow fluctuations already specified for the solution of the RANS equations. Different models of the turbulence eddy viscosity (basically the unknown of the S-A equation) at the inflow have been studied and tested. It was found that the influence of the value of the eddy viscosity coefficient at the inflow was not significant downstream.
4. Current work deals with the 3D simulations
The 3D RANS method is complete and deforming cylinder flows have been simulated. 3D profiles of inflow turbulence should now be used in order to study the effect of turbulence on both amplitudes and frequencies of the hydrodynamic forces.Reports & Publications:
W. Jester and Y. Kallinderis,
“ Numerical Study of Incompressible Flow about transversely Oscillating Cylinder Pairs”,
J. of Offshore Mechanics and Arctic Engineering, Vol. 126, February 2004.H. Ahn, Y. Kallinderis and R. Mercier,
“ Numerical Study of the Effect of Upstream Turbulence on the Hydrodynamic Forces exerted on Cylinders”, report, in preparation.
Date: December, 2003Project Name: CFD Simulation of Ocean Turbulence Interactions with Spar Platforms
Project Number: 32558-60260 MMS Task Order: 73611 MMS Project Number: 483
Investigator: Spyros Kinnas
Estimated Completion Date: August 2005
Project Description:
The effect of ocean turbulence on the hydrodynamic forces is currently an important issue for offshore installations in the GOM. There are two focus areas this year: (1) completion of the CFD study in 2-D of the effect of upstream turbulence on the hydrodynamic forces exerted on cylinders, and (2) study of the effect of upstream turbulence using 3-D CFD simulations. Work is being performed on developing techniques for simulating prescribed turbulence structures in the upstream flow field, as well as on simulations of flows around cylinders with and without turbulence being present in the current profile. One part of the work regards the specification of realistic profiles of turbulence for the CFD simulations. This part of the work is being addressed by a separate project led by Dr. Mercier. Numerical implementation of turbulence necessitates also investigation of required computational meshes and time-steps for appropriate resolution of spatial and time scales of the problem.
Progress:
Inlet turbulence profiles for two different Reynolds numbers of Re = 150 and Re = 1,000 were tested. The first case essentially represents laminar flow with low upstream turbulence. For both cases, the simulations give similar pattern of response to the inlet turbulence. Strong influence on the Cd values is observed. The peak Cd value is about 3 for the Re=1000 case. The time series of Cd “follows” the time series of the inlet turbulence profile (i.e. the stronger the turbulence, the higher the Cd gets).
Investigation of the appropriate upstream boundary condition for the turbulence flow variable (i.e. the intensity of local flow turbulence) was initiated by studying the influence of different values of this variable. The research on this topic is continuing.
Current work is focusing on the 3-D simulations of the “turbulence-structure” interaction. The same turbulence profiles which were employed in the 2-D simulations are being used for the 3-D study which will allow a conclusion on the effect of three-dimensionality on the hydrodynamic forces. In addition, implementation of truly 3-D turbulence profiles is being investigated.
Lastly, 3-D turbulence will be simulated and the corresponding hydrodynamic forces will be compared to those obtained from (i) 2-D simulations with 2-D turbulence profiles, and (ii) 3-D simulations with 2-D turbulence profiles.
Reports & Publications:W. Jester and Y. Kallinderis, “Numerical Study of Incompressible Flow about fixed Cylinder Pairs”, J. of Fluids and Structures, Vol. 17, pp. 561-577, 2003
W. Jester and Y. Kallinderis, “Numerical Study of Incompressible Flow about transversely Oscillating Cylinder Pairs”, J. of Offshore Mechanics and Arctic Engineering, accepted for publication, 2003
H. Ahn, Y. Kallinderis and R. Mercier, “Numerical Study of the Effect of Upstream Turbulence on the Hydrodynamic Forces exerted on Cylinders”, report, in preparation.
OTRC PROJECT STATUS REPORT
Date: June, 2003
Project Name: CFD Simulation of Ocean Turbulence Interactions with Spar Platforms
Project Number: Industry Funded
Principal Investigators: Spyros Kinnas
Estimated Completion Date: 8/31/05
Project Description:
This year’s focus is on two issues: (i) the effect of current profile turbulence on the forces on a cylinder, and (ii) investigate techniques to simulate roughness on cylinders. Both items are being investigated via CFD simulations in two dimensions during the current phase of the work. The effect of turbulence on the forces is currently an important issue for offshore installations in the GOM. A separate project led by Dr. Mercier is addressing the specification of realistic profiles of turbulence for the CFD simulations.
Numerical implementation of turbulence necessitates also investigation of required computational meshes and time-steps for appropriate resolution of spatial and time scales of the problem. Simulation of roughness is very expensive if one tries to model it geometrically. Special boundary conditions are being studied to see if they can be used to emulate roughness on cylinders.
Progress:
Part I (study of turbulence)
We started with current profiles that are spatially varying only. The two parameters of this initial investigation were (i) the amplitude of the disturbance of the uniform current, and (ii) the frequency of the disturbance. We found that increase in the level of turbulence (amplitude of superimposed waves) resulted in increase in both Cd and Cl. The higher the frequency, the smaller this increase was. The next immediate step is to try time-varying current emulating turbulence fluctuations.
This initial study was also used to determine the level of spatial and temporal resolution needed for such simulations.Part II (emulate wall roughness)
It is not clear how one can simulate roughness without actually resolving geometrically the roughness elements. The present study focuses on altering the surface boundary conditions. Three techniques were investigated: (i) surface pressure perturbations, (ii) near-surface velocity perturbations, and (iii) small jets distributed on the surface of certain strength. So far, it appears that the third special boundary condition is promising. The main parameter we varied was the strength of those jets. We found that by increasing the strength (i.e. the roughness gets larger) the drag coefficient increased but the lift coefficient decreased. The next immediate step here is to try a distribution of these jets with a random variation of their strength.
