| one publication added to basket [355966] | Evaluation of Gerris flow solver for the computation of wind coefficients: parameter variations and validation
Van Hoydonck, W.; Verwilligen, J.; López Castaño, S. (2022). Evaluation of Gerris flow solver for the computation of wind coefficients: parameter variations and validation. Version 3.0. FHR reports, 16_058_2. Flanders Hydraulics Research: Antwerp. XII, 60 + 35 p. app. pp. https://dx.doi.org/10.48607/128
Deel van: FHR reports. Flanders Hydraulics Research: Antwerp, meer
|
|
| Beschikbaar in | Auteurs |
|
Documenttype: Projectrapport
|
| Trefwoorden |
Harbours and waterways > Manoeuvring behaviour > Wind
Numerical calculations
|
| Author keywords |
CFD; Ship wind coefficients; Gerris flow solver; unsteady computation |
| Auteurs | | Top |
- Van Hoydonck, W., meer
- Verwilligen, J., meer
- López Castaño, S., meer
|
|
|
| Abstract |
The objective of project 16_058 is to evaluate the open‐source Gerris flow solver for the computation of ship wind coefficients.
In the original project plan, it was envisaged to only execute a parameter variation and grid convergence study with Gerris. While this research was started in 2016, due to shifting priorities, it has only been finalised in 2022. Not only was time spend to validate the Gerris flow solver, but time was also used to automate tasks that are executed before this type of Computational Fluid Dynamics (CFD) computations are executed. In the first report (Van Hoydonck et al., 2021), automation of tasks to prepare geometry for CFD computations were reported. The current report will present details on validation and use of Gerris for the determination of wind coefficients.
Multiple sources of experimental (wind tunnel) data have been used to validate Gerris for the computation of wind coefficients on ship‐like structures, ranging from simple wall‐mounted cubes and rectangular prisms to ship scale models. Before executing a validation, the effects on the solution of certain choices in the configuration of the flow solver are determined. The maximum refinement level (both in the wake and at the ship structure), the length of the refinement region and the blockage are all subject to parameter variations to determine their respective influence on the solution. Gerris is configured such that the viscous interaction of the flow with the ship structure (friction) is not included in the solution, only the pressure part is taken into account. The inherent unsteady nature of the flow is taken into account by solving the Navier‐Stokes equations in a time‐accurate manner. Overall, the predicted hull forces agree well with experimental data, following the trends for the complete range of incidence angles. In some cases, the peak values of the yawing moment shown significant deviations from the experimental values, the cause of this is unknown. A grid convergence study shows that predicted forces converge as the grid is refined.
One of the biggest disadvantages of Gerris is that in order to capture small details in the ship geometry, the maximum refinement level has to be set fairly high, resulting in a large cell count. Combined with the poor parallel scalability of the solver on the Flanders Hydraulics Research (FHR) cluster, this means that the required time to execute computations may be more than what was expected. The successor of Gerris (Basilisk1) has been reported to show significantly better parallel scalability, which may make it more suitable for computations where a large refinement is required on the hull. This will have to be verified in future research.
In an appendix of this report, a discussion is held about a typographical error that was discovered in a peerreviewed publication related to the experimental determination of wind coefficients Andersen (2013). It is found that the error affects the accuracy and reliability of steady Reynolds‐Averaged Navier Stokes (RANS) CFD methods reported in literature that used Andersen’s data as validation material. This second report also contains a critical review of past research related to the simulation of atmospheric boundary layers in FINE/Marine. |
|
