Discussion Groups

Simulation Mechanical and Multiphysics

Reply
Contributor
alok
Posts: 24
Registered: ‎11-02-2011

Flow over a bar

323 Views, 12 Replies
04-27-2012 07:42 AM

I am trying to simulate a fluid flow over a bar (see attached) using Multiphysics 2013. I used to work on Multiphysics 2012 before my company decided to upgrade to 2013 a couple of weeks back. 

How do I generate the external fluid region after exporting the model to Fusion? To capture the 'cap surfaces' do I need to sketch them at the inlet and outlet surfaces?

 

Any help would be appreciated.

 

Thanks!

Alok

Please use plain text.
Employee
Joey.X
Posts: 259
Registered: ‎06-14-2010

Re: Flow over a bar

04-30-2012 05:39 AM in reply to: alok

Using CAD's boolean operation to make fluid part(s), for the external fluid region, do "box - your part".

Jianhui Xie, Ph.D
Principal Engineer
MFG-Digital Simulation
Please use plain text.
Contributor
alok
Posts: 24
Registered: ‎11-02-2011

Re: Flow over a bar

05-29-2012 10:55 AM in reply to: Joey.X

Thanks for your response Joey. I didn't really get a chance to continue working on this until now. So after exporting the model to fusion, I did the external fluid region using the box (see the attached picture).

I need the pressure drop across the bar with the working fluid being water. When I import the model back to multiphysics, the new component (box) is created. Are the boundary conditions to be specified on the surfaces of the box?

I tried running the steady state flow analysis and got some unrealistic numbers. Don't know what exactly I'm doing wrong.

 

Thanks!

Alok

Please use plain text.
Employee
Joey.X
Posts: 259
Registered: ‎06-14-2010

Re: Flow over a bar

05-29-2012 11:42 AM in reply to: alok

Yes, the BCs(velocity, pressure etc..) are to be set on the box surfaces, the do-nothing surfaces will be automatically set as wall, i.e, no slip BCs.

BTW, it is suggested to extend the outflow surface of the box to centain distance to avoid the unphysical BC on the outlet surface(this will bring convergence and accurancy issues), note the pressure on the outlet suface is far away from constant if the box end surface is too close to your object. 

 

Jianhui Xie, Ph.D
Principal Engineer
MFG-Digital Simulation
Please use plain text.
Contributor
alok
Posts: 24
Registered: ‎11-02-2011

Re: Flow over a bar

05-30-2012 06:21 AM in reply to: alok

Hi Joey,

 

Thanks for the quick response.

I got everything set up, but when I run the analysis I get a warning "Converge with stagnation due to oscillation" and it gets stuck there. This is the first time I have had this warning.

I have attached the log file herewith.

 

Thanks!

Alok

Please use plain text.
Employee
Joey.X
Posts: 259
Registered: ‎06-14-2010

Re: Flow over a bar

05-30-2012 06:35 AM in reply to: alok

Hi, Alok,

The warning "Converge with stagnation due to oscillation"  is usually okay to accept, but user may need to justify the result by themself to avoid false convergence.

Think about the the example of "Vortex shedding ", i.e flow passing through a cyclinder, the flow after the cyclinder keeps oscillation, and the specified norm can not always satisfied and result norm oscillates at relative higher level, this is usually acceptable.  

Some CFD package uses number of iterations to move forward to next time step (transient analysis) or finish once the nonlinear iteration reaches the specified number. That is another way to take care of the oscillation, just rember the  fluid oscillation always exists physically and numerically.

Thanks, 

 

Jianhui Xie, Ph.D
Principal Engineer
MFG-Digital Simulation
Please use plain text.
Contributor
alok
Posts: 24
Registered: ‎11-02-2011

Re: Flow over a bar

05-30-2012 08:44 AM in reply to: alok

Hi Joey,

 

Do I need to specify any B.C.s / nodes on the part itself for the flow to go over it?

The flow passes right through from inlet to outlet instead of going over the object (see attached). I saw one of your models for "2D steady planar flow through submarine" where you have specified nodes on the submarine surface.

 

Thanks!

Alok

Please use plain text.
Employee
Joey.X
Posts: 259
Registered: ‎06-14-2010

Re: Flow over a bar

05-30-2012 02:56 PM in reply to: alok

Hi, alok,

I assume that you want simulate flow passing through an object, then the part is stationary so that all the nodes on the part surfaces have zero velocity but leave node pressure as free degree of freedom(DOF),  i.e, no-slip boundary condition. 

Back to your question, you don't need specify any BCs on the part surfaces, the software will set  no-slip boundary condition on all wall nodes, it refered as automatic wall boundary condition to save user's efforts.

 

Regards, 

Jianhui Xie, Ph.D
Principal Engineer
MFG-Digital Simulation
Please use plain text.
Contributor
alok
Posts: 24
Registered: ‎11-02-2011

Re: Flow over a bar

05-31-2012 06:34 AM in reply to: alok

Hi Joey,

 

Thanks for your response.

I want to simulate the flow OVER the object and not THROUGH the object (external flow). In my model, the flow is just passing through the box (fluid region created in fusion); so I was wondering if I need to specify any nodes on the part itself. I went through the user manual for Multiphysics and it says when modeling an external flow over an object, the object should be excluded from the mesh (like a construction object) and the mesh should only be the fluid part. How do we achieve this?

 

Thanks!

Alok

Please use plain text.
Employee
Joey.X
Posts: 259
Registered: ‎06-14-2010

Re: Flow over a bar

05-31-2012 06:46 AM in reply to: alok

Let's clarify this, from your posted picture (mesaage 3 in this thread),  it appears you put the object inside a box, i.e, your intention is simulate the flow passing the object(not inside the object), right? This is like setting a car in wind tunnel, the simulation domain is the volume between the object and box. 

If this is your intention,  you don't need solid (volume) mesh in the object, the simulation domain is the volume between the object and box, that's the domain to be meshed and simulated. And you don't need put any BCs on the object, all nodes on the object surfaces are no-slip BC.

Jianhui Xie, Ph.D
Principal Engineer
MFG-Digital Simulation
Please use plain text.