[louisdub11]

Hello everyone !

I am currently doing my master thesis, and the subject is doing a simulation of a scroll expander. For that I had to generate a mesh in ICEM CFD, and I have to make the simulation with CFX software.

So far, I have generated a tetra mesh on ICEM, imported this mesh in CFX-Pre, and I have done a "static" simulation in CFX Solver and CFX-Post (You can see some of the results in the pictures below).

My question is: how could I do the dynamic mesh and the movement of the orbiting scroll in CFX-Pre?

I'm trying to put the movement equations in the moving wall domain, but I'm not sure if I have the correct approach, and haven't found relevant answers on the internet (And I have 1 month left for my thesis so I'm getting a bit nervous ).

Mesh.jpg

Convergence.PNG

Velocity profile.jpg

Capture.PNG

Here you can see the equations I'm trying to enter:

moving scroll1.PNG

When I enter these equations, I get the error:

Bad expression value '-2*cos(t*2*pi*5)*2*pi*5' detected in parameter 'Displacement Y Component' in object '/FLOW:Flow Analysis 1/DOMAIN:Scroll Flow/BOUNDARY:moving scroll/BOUNDARY CONDITIONS/MESH MOTION/DISPLACEMENT'.
CEL error:
The dimensions of the argument to function 'cos' are inconsistent with expected dimensions.
Argument dimensions resolve to 's'; expected 'rad'.

If someone has a clue about how I should do, I would be very grateful !

Don't hesitate to ask me if you need more details !

Thanks,

Louis

[ghorrocks]

You won't be able to do a motion like this using moving mesh. The only way you are going to be able to model this motion is using dynamic remeshing which links ICEM into CFX and remeshes the geometry to handle the large deformations. For a tutorial on how to do dynamic remeshing have a look at the ANSYS customer webpage.

Also - the simulation is now transient. You will have to do this as a transient model.

[louisdub11]

Thanks for your answer, I appreciate the help ! :-)

I will try to find how to do the dynamic remeshing

[louisdub11]

Hi again,

I have no access to the customer portal, so I asked my supervisor if she can give me a user name for accessing it.

In the meantime, I think I might have found useful information on the CFX help guide. If I'm not mistaken, is it the technique you're referring to for the link between ICEM and CFX (see picture below) ?

Thanks a lot !

remesh.PNG

[ghorrocks]

Yes, that is it.

There is also a student/academic ANSYS website, but it only contains simple tutorials. You should look there as well (you will be entitled to a login on that site), but be aware that as your simulation is complex it might not have information on it.

[louisdub11]

Hello again

My supervisor doesn't think she can give me a username for the customer portal because they only have academic licences.

Does anyone have a PDF with the tutorials or a username available?

Another question: I saw that the mesh quality should be good with angles above 18 degrees, and determinant above 0.3. Is there an acceptable percentage of lower values? I have a mesh with only 1 element below 18 degrees, which represents in this case 0.003%. Would it create any errors in the simulations?

Thanks a lot !

[ghorrocks]

Of course your supervisor cannot give you a login. He/she does not work for ANSYS. You have to ask ANSYS for a login. If you don't have a commercial or research license then you are probably restricted to the academic ANSYS page - but that is still worthwhile, so get a login for that.

Mesh quality - the actual requirement for mesh quality depends on what you are modelling. If you are doing incompressible fluids at moderate Reynolds numbers it should be quite tolerant of poor mesh quality.

[louisdub11]

Hi !

After having looked for how to do the dynamic remeshing with ICEM Replay, I have saved a replay file with all the steps in generating the mesh in ICEM. I also saved a reference geometry in order to do the dynamic remeshing.

I thought of doing an expression for the motion of the moving wall. Then I would create an expression for mesh quality check (the remesh would occur every time the quality doesn't meet the requirements). Then, with the other parameters, I should be able to do the simulation.

My question is: Is my approach correct? And for the motion expression, how can I enter equations like this:

x = R sin(t*2*pi*f) * 2*pi*f
y = -R cos(t*2*pi*F) * 2*pi*f

I don't know if it is possible to do a parametric equation in the expressions and I'm a bit lost in the methodology I should follow for the simulation

Thanks for answering my questions

[ghorrocks]

Yes, you can do parametric equations in CFX, as long as you define a function for "f" it will work. Note that you have an upper case "F" in the y equation, that is probably meant to be a lower case.

[louisdub11]

Yes, the F was meant to be an "f"

So, if I understood correctly, I should do an expression for x, then another expression for y, and for f it would be for example "2 [rad/s]" ?

Or is it possible to make a unique entire expression with x and y motions?

thanks

[ghorrocks]

I cannot second guess your equations. If you want us to check your equations you have to tell us what you are trying to do, what is the value of key input variables and what output you are looking for.

[louisdub11]

Ok

So, for my simulation, I need to do a dynamic simulation of the scroll expander.
For that, I have created the following geometry:

scroll.PNG

I have assigned the name "wall 1" to the shell and fixed scroll, and the name "wall 2" to the orbiting scroll.

This orbiting scroll should follow the motion of my equations. As they are parametric, can I just enter them like that in an expression?

motion expression.PNG

Here are the different parameters:

R= (pi - pi/3)
t= time
f= rotation frequency of the rotor

The thing I try to achieve is to get the different pressures and velocities in post-processing and have a dynamic animation like this:

https://www.youtube.com/watch?v=iU5P...045gYk&index=8

Is it clear enough to understand what I am trying to do?

Thanks

[louisdub11]

I thought of doing like this, if it can give you some more details..

expression.PNG

But it gives me this error:

error.PNG

Any idea why I get this error?

[Gert-Jan]

You should remove the 'x=' and 'y='.
Btw, are you sure about the unit of the frequency?

[louisdub11]

Thanks for your answer !

Without the 'x=' and 'y=', it seems to work.

However, there is a problem with the units. Indeed, I am not sure about the units of the frequency...

I should have a displacement in meters, but I have this error:

error2.PNG

I tried with a frequency in [Hz] and in [rad/s], but I get the same error, and I don't see what else I could try...

[Gert-Jan]

- You named the expressions xmotion and ymotion. That suggests that you want something like a velocity in x an y. This would imply something like [m/s]. Is that what you want?
- Or do you mean xdisplacement and y displacement? Then you indeed need something like [m]. But you have a dimensionless constant 4, a dimensionless sin-function, and multiplied by frequency. This indeed resolves to [1/s]
If you want [m], then you could give your constant 4 the dimension [m*s]. But I have no idea if this is realistic since I don't know your problem. But then at least you can get CFX to accept your expression.

[louisdub11]

Yes, I need displacements in [m].

The scroll expander motion is an orbiting motion, represented by these equations.

The constant 4 represents the orbiting radius... I found these equations in a scientific article, but they don't explain the different variables :/

In this article I am referring to, they talk about 'macro defined function'. Is it of any help for you to understand?

[Gert-Jan]

You just have to try in a spread sheet to see if the numbers make sense. You can also make a plot in Pre, so you can find out if the conversion from spreadsheet to pre went ok. Sometimes Pre results in different numbers because of difference in treatment of the radians.

You can also ask ANSYS. They have an office in you back garden in Brussels (Wavre).

[louisdub11]

Ok, I found that in the article, they define the frequency of the rotation as [rad] and not [rad/s]

I am currently in Italy to do my Master Thesis, so it will be difficult to go to Wavre at the moment

I will try to put the Freq in [rad]

[louisdub11]

Maybe I found something.

I think there might have been a mistake in the article, because the equations are not possible in the Grapher tool on my computer.

However, when I delete the final term "*2*pi*f", it gives me an orbiting radius like this:

Screen Shot 2018-05-17 at 14.14.05.png