Hello Everyone,
I am trying to simulate a high speed train (2D) , in fluent.
I have used sliding mesh before , so I setup my problem accordingly, creating two separate zones, one for solid and one for fluid, and made an interface at the boundary of solid and fluid.
As per sliding mesh technique, I assigned mesh motion to the solid zone and used transient, density based solver.
But I am not getting convergence, energy residual is of the order of 1e2 and results are far from being accurate.
Should I use, dynamic mesh for this problem?
Is this issue because of very high speed of train (100 m/s).
Thanks
Regards
Shashank Mishra
Sliding mesh is as accurate ad dynamic mesh. Be sure that the transition between two elements at the sliding mesh interface is performed at at least 2 time steps.
Franco
Sir, in reality if we wish to study such case (in wind tunnel), we will built a model and fix it, while changing the speed of air. I think this will be easier and more accurate. In addition, this will help you to keep y+ near from 1 and to add enough number of layers (15 to 20) in the boundary layer region.
Hello everyone, thanks for your suggestions.
My boundary condition is pressure inlet at the inlet. Since I am simulating in a big tunnel where pressure downstream of the train, at sufficient distance remains the same so I have an extended outlet with same pressure as of inlet.
Now, specifying wind velocity is not a possibility since I need to impose a given static pressure at the inlet.
Also, energy equation is coupled with the momentum equations.
But I am getting results which are absurd.
There is no effect of train motion on the flow field
I ran a steady state with no train motion for intial condition and then started sliding mesh simulation.
the mesh motion is not having any effect.
also downstream of the train, the flow should have air but it seems like that region is always empty.
Try this, For inlet, specify total pressure(velocity head + static pressure) with reference-pressure-location as out let. Target mass flow may be required.
Or check with pressure-far-field boundary conditions where velocity is required to be specified
Neither. You should use a static mesh and a coordinate system that is fixed to the train. In this coordinate system the train and mesh are fixed and the air moves towards the train. Navier-Stokes works perfectly in every coordinate system that moves at a constant speed (due to the fact that it just reflects classical Newtonian physics).
This removes the mesh motion (which is difficult to use even for experts) from your problem. Adding mesh motion is like flying a jet before you have figured out how to drive a car. And you have absolutely zero reason to use mesh motion for this problem.
So don't. Use neither sliding nor dynamic mesh motion for this problem.
Dr. Perot has rightly pointed to modelling error. Please look into the the simulation of automobiles and your problem at hand requires only choice of suitable turbulence model at high mach number and corresponding y+ requirement. Regarding BCs, it is already been answered
This the the mesh motion after using sliding mesh
So does it mean sliding mesh cannot be used for translation motion.
The issue with simulating with changing air inlet velocity will distort the boundary conditions.
I have to impose 100 pascal of total pressure. Now if I calculate 0.5*rho*v*v and add it the this, then the total pressure is changed.
If a part of flow domain is moving relatively to rest of the domain then only use these sliding mesh or dynamics mesh like in turbomachines. Otherwise use static mesh with appropriate BCs
The total pressure is indeed changed in a moving coordinate system. Because it is NOT a Galilean invariant quantity (it varies with each coordinate system). So change it., do the calculation. And then change it back after your calculation is done (usually in the post processor, like Paraview). Stop making excuses. Stop being stubborn. You have been told how to easily do this problem. Make a mountain if you like. But this is a molehill.
- Badges
- Science topic