Hi, I am trying to run a simulation in fluent for 2D Wind Turbine using Dynamic meshing.
I read somewhere that time step required for convergence in dynamic meshing has to be very small for very fine grids because at any time step, the displacement shall not be more than twice of minimum cell length.
Does not this limit the use of dynamic mesh for a coarser mesh?
Hi Shashank,
Choosing a right time step to suit your case would rely on the CFL study. Try limiting your Courant number close to 1 and based on that choose a time step. Your reading meant that time step should be chosen in such a way that the displacement should not exceed one cell in one time step. It doesn't limit your mesh type if you play around with time step unless you are using LES or DES solver where mesh refinement is of prime consideration. Just have a go with CFL study and I feel it would work for you.
I am working on turbine blade modeling using DES solver wherein even I had to play around with time steps. In order to achive good convergence, I could have refined my mesh or reduced my time step size. But, I chose the latter. See what works for you better. Good luck!
Hi VInit, thanks for the reply.
I am doing is using 6 dof modelling for running the simulation so that I dont have to give rpm of blade as boundary condition but with the help of a udf , rpm is calculated at each time step.
Since I have 2 D Mesh, therefore 2.5D , local face, layering option ( i have triangular mesh ) is not applicable. When I use smoothing and remeshing (size function ), it gives me negative volume error no matter how small time step I use. I use time step as small as e-07 but still it didnt work. I guess may b there is an issue with my udf. Is that possible?
Hi Shashank,
I recommend you to use "Coupled" solver and the "Cycle Type" of your "Flow" should be "Flexible". I always start the analysis with steady solver first and then move on to the unsteady solver. After I move on to the unsteady solver, I select my momentum and turbulence values as first order and after the convergence is achieved for these, then I make them second order. It is hard to converge analyses involving dynamic mesh so you should watch your case carefully and select your preferences accordingly. I hate to say this thing because everybody keeps telling it but "take a closer look at your mesh". Be especially careful with y+ values to obtain correct features of the flow. Select your turbulence models accordingly.
Hope this helps!
Hi Omer,
Thanks for your help.
I followed exactly same procedure when I used Sliding mesh technique for 2D VAWT analysis. First I ran the steady state solution using MRF and then coupled it with transient solution using Sliding mesh.
The reason why I want to use dynamic mesh is whether you use MRF or Sliding Mesh, one has to provide rpm as an input. But I want to rotate the turbine by calculating torque due to fluid blade interaction . I wrote the UDF and hooked it in dynamic mesh zone. When I make the cell adjacenet to blade as deforming, I get negative volume error. I also tried running simulation by selecting 6dof passive mode on for adjacent cell, then the blades didnt rotate at all.
And what is y+ values?
Shashank,
I am assuming that you are using a turbulence model for your case. In order to comply with the models you select as your turbulence model, you have to set up your mesh accordingly.
If you are still getting errors, define another zone inside the fluid domain that covers the blades. Define your new zone as rigid and the rest of the fluid as deforming. This way, the cells that are very close to the blades will not be deformed but the cells which are bigger in size (which are in the outer domain) will be deformed instead.
You may as well set a minimum limit to your cell size for your remeshing. I think the default value for this is 0 so if you do not change that, dynamic meshing does not do remeshing until these cells are really small. After it attempts to do remeshing it is too late because your cell size is already down below zero, which may be the reason of your case obtaining "negative volume".
Hi Omer,
I am using SST (4 equation ) model for this simulation because my flow field have turbulent as well laminar flows.
Do you know if lets say I define one udf for time dependent omega and hook it to blade region, and in another udf, I define 6DOF properties like mass, Ixx and motion constraints ( because my turbine will only have on DOF that is rotation in Z axis ), can I hook both these udf to one thread that is blade zone?
And I really appreciate your help .
thanks
I do not know how your UDF works but I think you may specify all these in just one file. However if you have negative volume problems, the first place to look at should be your dynamic mesh settings; not your UDF. Depending on the software you use, you may want to take a look at the manuals for the dynamic mesh setup. If you are using Fluent, the manual is quite good to explain how dynamic meshing works.
Hi Shashank,
The dynamic mesh can covers both fine and coarse mesh elements, however the time step should be chosen correctly. Wrong time step gives you negative cell. Consequently, the run of simulation will stop when the first negative cell is generated. Meanwhile, another effective feature in the selection of time step is the type of dynamic mesh, because the movement of mesh is different depends on the chosen type of dynamic mesh.
For coarse mesh, you can select larger time step and for the finer one the time step can be smaller. However, for both of them the time step should be selected wisely.
Good luck.
- Badges
- Science topic