[cyclogyrocrasher]

p { margin-bottom: 0.1in; line-height: 120%; }Thank you for your help in advance! I've tried my best to avoid taxing the time of others with my problems, but I've come across a problem which is beyond my expertise. I've beaten my head against it for too long. I hope the responses can benefit others.

Simulation Physics Description: I am evaluating the interaction between the freestream flow and cycloturbines (variable pitch vertical axis wind turbines) for different blade pitching motions (performing simulations on different blade pitching motions for cycloturbines . The blades are rotating about the rotor at 7-28m/s in a freestream flow of 7m/s, and in sea level standard air density/viscosity. Since there is no heat transfer, and relatively low velocities I selected an incompressible solver (pimpleDyMFoam). I am interested in the unsteady interaction of the blades so I selected an unsteady simulation. However I need to reduce the computaitonal demand to perform many simulations so I selected a 2D simulation. A number of other similar investigations suggest this is adequate for my purposes.


Simulation Description: I have selected openFoam so that I can easily pass on the model to once I get it working


Solver Characteristics

Solver: pimpleDyMfoam
Pimple characteristics:
Outer correctors: 1
Correctors: 6
Non-Orthogonal Correctors: 3
Maximum Courant Number: 10
Turbulence Model: k-epsilon



Tolerances and Linear Matrix Solvers
[Variable][Solver][Smoother][Residual Tolerance][Relaxation Factors]
Pressure: GAMG, DIC Gauss-Seidel, 10^-6, 0.3
Velocity: Smooth Solver, Symmetric Gauss-Seidel, 10^-6, 0.3
k: Smooth Solver, Symmetric Gauss-Seidel, 10^-6, 0.3
epsilon: Smooth Solver, Symmetric Gauss-Seidel, 10^-6, 0.3

Numerical Methods
Gradient: [Pressure=Gauss Linear][Cell Limited Gauss Linear]
Divergence: Gauss Upwind
Laplacian: Gauss Linear Limited Corrected
Interpolation: Linear
d/dt: Euler

Mesh Characteristics
Cell Type Hexahedra
Number of Cells: 153675
Maximum Cell Aspect Ratio: 29.74
Maximum Non-Orthogonality [deg]: 64.19
Maximum Cell Skewness: 1.26
Height of first cell [Y+]:16
Dynamic mesh Interface: Sliding Mesh (each blade slides in the rotor mesh which slides in the domain mesh. Sliding mesh interfaces depicted as green lines on the mesh figure)


Boundary Conditions
The boundary conditions are shown in the attached boundary conditions figure.
Initial Conditions:
A potential flow solver generates a steady flow field for a stationary rotor, which then switches to the N-S solver and spins up to full speed.


Question/Problem: The pressure field (especially in the domain prior to the turbine) experiences wild unrealistic pressure oscillations with the passage of each blade (shown in the final figure). This seems to be a result of implementing zero gradient pressure boundary conditions upstream. Is there a different boundary condition that could be implemented? How could I improve my boundary conditions (listed in the boundary condition figure)? Could my numerical methods/solver be exacerbating the problem?


I spoke with a professional (CFD, not openFoam) in the field and he recommended implementing a wave transmissive (characteristics based) boundary condition so that the boundaries would not reflect pressure waves. However it does not appear that openFoam has any such boundary condition for incompressible solvers. Moreover it seems like others got similar simulations to work with exactly these boundary conditions. Any thoughts?


Thanks again! I welcome your advice and criticism.



-Cyclogyrocrasher

[cyclogyrocrasher]

I apologise for the low quality figures! They were the best I could upload with the restrictions.