Dear Dr. Jonkman,
I am currently attempting to reproduce the visous damping coefficients of the UMaine VolturnUS reference platform in the NREL Report (NREL/TP-5000-76773, Allen et al., 2020) using OpenFOAM. Based on the brief description in the report, my goal is to apply the same methodology (or approach) in OpenFOAM to obtain values consistent with those listed in the attached table.
I am using simpleFoam, a steady-state incompressible solver, to apply unit translational (1 m/s) and rotational velocities (1 degee/sec) to calculate the damping coefficients for each degree-of-freedom (DOF).
However, with the limited information provided in the report, it has been challenging to reproduce the consistent results shown in the table above. So I would be greatly appreciate if you could share more details on the methods used to calculate the damping coefficient in OpenFOAM. Since the OpenFOAM results can vary substantially with domain boundary conditions, meshing quality, convergence behavior, and other input settings, additional modeling details or information would be extremely appreciated.
Especially,
-
the specific convergence criteria and procedure used to determine the damping coefficients,
-
critical simulation settings or conditions (e.g., domain boundary conditions for velocity(U) and pressure (p), how to impose the rotational unit flow (angular velocity), meshing or key input parameters, etc.).
-
How was the rotational unit flow (angular velocity) imposed? When using a steady-state solver (simpleFoam), I understand that the unit flow imposed at the “inlet” is used to calculate the forces and moments acting on the surface of the stationary platform body, and the surge motion can be induced by applying a flow through the inlet in the direction normal to the surface. However, to induce the heave or sway motion, under the assumption that the flow is applied normal to the inlet surface, does this mean that the platform inside the domain must be rotated? Also, how was the rotational (angular) velocity imposed from the inlet to induce the pitch, roll, or yaw motions? I was wondering if methods like “codedFixedValue” or “MRF” should be used for simulating the rotational behaviors in the steady-state solver.
-
Could you please share the procedure used to convert the force/ moment from OpenFOAM for each DOF into the damping coefficients shown in the table? Based on the units in the table, it appears that the forces or moments were divided by the square of the velocity (or angular velocity (omega)), but I would like to confirm whether a particular governing equation was used for this calculation.
I apologize for the questions and any requests I have made, and any suggestions or comments you may be willing to share would be greatly appreciated.
Best regards,
Jeongjoo Kim
Dear @JeongJoo.Kim,
I am not able to answer your questions because this quadratic drag matrix was developed by the University of Maine, not NREL. Moreover, we at NREL would generally recommend avoiding the use of a quadratic drag matrix, and instead, employing the strip-theory solution of the HydroDyn module to properly capture viscous effects in OpenFAST. While NREL has developed a HydroDyn model with strip-theory viscous terms for the VolturnUS semisubmersible, this is not currently uploaded to a public repository. However, we are intending to public release such a model soon.
Best regards,
Dear Dr. Jonkman,
Thank you so much for your kind clarification on the quadratic drag matrix and for recommending the use of the strip-theory solution in HydoDyn to capture viscous effects. I will certainly take this into account in my ongoing work. I also greatly appreciate you letting me know about the forthcoming public release of the VolturnUS model. I will look forward to its availability.
Thank you again for your guidance and for taking the time to provide such a thoughtful response.
Best regards,
Jeongjoo Kim
Dear Dr. Jonkman,
Thank you so much for sharing the updated HydroDyn input file. I truly appreciate your continued help and support. I have a question regarding the strip-theory approach. Since each member (circular columns and pontoons) is treated individually, it seems that the hydrodynamic interactions among the members, such as the wake-shielding effect, where the upstream column shields the downstream center column and reduces its local flow velocity, may not be directly captured in the model. Is my understanding correct that such interaction effects need to be implicitly embedded in the calibrated member drag coefficients? I am asking because I have been working on identifying the viscous damping of the VolturnUS-S platform (i.e., IEA-15-240-RWT-UMaine semi-submersible platform) using CFD simulations (i.e., Fluent) under steady surge flow. In the CFD results, the wake formed behind the upstream column significantly reduces the incident velocity on the center column, and the resulting effective drag coefficient for the columns is less than the typical value for an isolated circular cylinder (0.5-0.8). This suggests that the inter-member interactions have a considerable influence on the effective coefficients.
I would appreciate any thoughts on how these interaction effects are typically handled when setting the member coefficients in the strip-theory model.
Thank you again for sharing the files and references.
Best regards,
Jeongjoo Kim
Dear @JeongJoo.Kim,
Indeed the strip-theory formulation of HydroDyn assumes no hydrodynamic interaction between members. The potential-flow solution of HydroDyn can calculate hydrodynamic interaction between bodies, but for hybrid potential-flow + strip-theory solutions, the potential-flow solution does not impact the strip-theory solution. With this simplification, I agree that hydrodynamic interaction between members, when important, would have to effectively be taken into account through the hydrodynamic coefficients themselves.
Best regards,
1 Like
Dear Dr. Jonkman,
Thank you so much for the clarification and explanation. Your response provides meaningful support for the approach I have been taking, which is deriving effective drag coefficients from CFD simulations that directly resolve the inter-member interactions, such as the wake shielding among the columns, so that these effects are inherently reflected in the calibrated coefficients.
Your response has been very helpful in clarifying the modeling assumptions and strengthening the basis of my work. Thank you again for your time and kind support.
Best regards,
Jeongjoo Kim
