[Roland R]

Hello,

I complated an interesting test. I calculate the flow around a simply 3D body (like a brick). The body is located in a large wind tunnel. A fix velocity was defined for inlet and static pressure for outlet. The simulation is steady state of course.

In second step, I prepared this same geometry but I defined a moving for the body (with deforming mesh method). The speed of displacement is same to velocity inlet of previous simulation. In this case there are not inlet and outlet, the body moved in a standing air field (v=0)

In other words I wanted to compare the wind tunnel investigation with the real moving of body.

Based on my comparison, the velocity distribution, the separated zones around the body are same but the pressure distribution at front face is different. The front face of moving body has higher average static pressure with 20%, as a result the acting drag force also is higher at the full body in the case moving body.

To tell the truth, I dont understand what is the reason of this difference.
What do you think about it? Does this difference come from simulation error? Or can the reality show similar differences?

Thanks
Roland

[singer1812]

How big is your domain? I would tend to think your moving body simluation is not at steady state.
How fast is it moving? I would think that you couldnt make your domain large enough to accomadate a moving body simulation as you describe and get it to a "steady state" state

[ghorrocks]

Unless you have done the necessary work to show that both the stationary and moving models are accurate then you are just comparing one random number with another random number so the comparison is meaningless.

So I think this question is effectively this FAQ: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F

[khaled ali]

I want to help i have H shape wind turbine 2 blades and 3D I want to simulation with CFX how write expressions and applications it

[Bodo1993]

Originally Posted by Roland R

Hello,

I complated an interesting test. I calculate the flow around a simply 3D body (like a brick). The body is located in a large wind tunnel. A fix velocity was defined for inlet and static pressure for outlet. The simulation is steady state of course.

In second step, I prepared this same geometry but I defined a moving for the body (with deforming mesh method). The speed of displacement is same to velocity inlet of previous simulation. In this case there are not inlet and outlet, the body moved in a standing air field (v=0)

In other words I wanted to compare the wind tunnel investigation with the real moving of body.

Based on my comparison, the velocity distribution, the separated zones around the body are same but the pressure distribution at front face is different. The front face of moving body has higher average static pressure with 20%, as a result the acting drag force also is higher at the full body in the case moving body.

To tell the truth, I dont understand what is the reason of this difference.
What do you think about it? Does this difference come from simulation error? Or can the reality show similar differences?

Thanks
Roland

Hi, I am also interested to know the difference. I am wondering if you were able to make some conclusions from your comparison. Thanks.

[ghorrocks]

Assuming the wind tunnel is big enough that the blockage factor is insignificant, then they are the same. This is because in an inertial frame of reference, any non-accelerating frame of reference is valid. So the frame of reference travelling at a constant speed equal to the body will give the same result as a frame of reference fixed to some external frame of reference.

See here: https://en.wikipedia.org/wiki/Inerti...e_of_reference
Particularly the statement "Measurements in one inertial frame can be converted to measurements in another by a simple transformation"

Note this does not apply if the frame has any form of acceleration. This could be linear acceleration (eg an accelerating car), rotation or other more complex movement.

But refer back to my post #3. Unless you have carefully validated and verified both models then you are comparing one random number to another random number and the comparison is meaningless. Unless both simulations are accurate you cannot compare them. This is especially so for the moving body in fixed frame of reference simulation - getting the flow near the body and in the boundary layers accurate will be extremely challenging for this model. The wind tunnel model is far easier to get accurate.

[Bodo1993]

Originally Posted by ghorrocks

Assuming the wind tunnel is big enough that the blockage factor is insignificant, then they are the same. This is because in an inertial frame of reference, any non-accelerating frame of reference is valid. So the frame of reference travelling at a constant speed equal to the body will give the same result as a frame of reference fixed to some external frame of reference.

See here: https://en.wikipedia.org/wiki/Inerti...e_of_reference
Particularly the statement "Measurements in one inertial frame can be converted to measurements in another by a simple transformation"

Note this does not apply if the frame has any form of acceleration. This could be linear acceleration (eg an accelerating car), rotation or other more complex movement.

But refer back to my post #3. Unless you have carefully validated and verified both models then you are comparing one random number to another random number and the comparison is meaningless. Unless both simulations are accurate you cannot compare them. This is especially so for the moving body in fixed frame of reference simulation - getting the flow near the body and in the boundary layers accurate will be extremely challenging for this model. The wind tunnel model is far easier to get accurate.

Thanks ghorrocks. I relate the above to simplify a setup for a CFD simulation, where I am interested to pull a solid wall out of a 2D fluid domain at a certain velocity. Typically, dynamic meshing is used, however, I do not want to use it due to many complications. Instead, I would like to set the case up such that the solid wall is fixed and the fluid moves (change the reference frame). I am wondering about the proper setup for the initial condition and boundary conditions (perhaps time dependent?) for this case.

[ghorrocks]

Can you show an image of what you are trying to do? I don't understand what you are asking.

[Bodo1993]

Thanks for the reply. For the dam break problem below, the gate should move upwards at a certain velocity.
Dam-break with a vertical-lifting gate (without dynamic mesh technique?)
We typically use dynamic meshing to do the job. However, what would be an alternative setup for the problem if dynamic meshing is not an option? I was thinking of changing the reference frame, where I assume it moving with the gate. In this case from an observer on that moving reference frame, the gate is stationary and everything else is moving at the same velocity magnitude and opposite direction, vertically downwards (Similar to the wind tunnel concept). What I am not sure about is: what should be the correct initial and boundary conditions in this case such that the two problems (with dynamic meshing and without) are equivalent.

[ghorrocks]

While the gate does not move relative to that reference frame, the ground does. So you are either modelling in a stationary frame of reference with a moving gate or modelling in a frame of reference following the gate and then you have a moving ground. I would think the stationary frame of reference would be a simpler model in most cases.

Note that moving mesh is not your only option. You can model the gate with immersed solids - in fact I would recommend this approach as it is MUCH easier than moving mesh, as long as the immersed solids restrictions are not a problem for you.

You can also model it using source terms (which is in effect the same as immersed solids) or dynamic remeshing (which is even harder than moving mesh )

[Jiandong]

Hi,
I would like put this error to the different setting of interface type,if you choose different types betweent your two models.
And, Verfication is very important.

Originally Posted by ghorrocks

Unless you have done the necessary work to show that both the stationary and moving models are accurate then you are just comparing one random number with another random number so the comparison is meaningless.

So I think this question is effectively this FAQ: http://www.cfd-online.com/Wiki/Ansys..._inaccurate.3F