1D Navier Stokes

[s.j...@usnc.com]

Hi, new moose user here. I am trying to create a 1D pipe flow model using the incompressible navier stokes package, but having issues with convergence. I built off of the "pressure_channel" test case in navier stokes module, with the only changes being the removal of one dimension from the original input file and element type change from QUAD9 to EDGE3. Have attached moose output showing nonconvergence, and also input file used. Am confused because the first set of linear iterations seem to converge and then on the start of the second nonlinear iteration it aborts. Would appreciate any help/advice. 

Thanks!

Sam 

[Peterson, JW]

On Thu, Oct 19, 2017 at 12:20 PM, <s.j...@usnc.com> wrote:

Hi, new moose user here. I am trying to create a 1D pipe flow model using the incompressible navier stokes package, but having issues with convergence. I built off of the "pressure_channel" test case in navier stokes module, with the only changes being the removal of one dimension from the original input file and element type change from QUAD9 to EDGE3. Have attached moose output showing nonconvergence, and also input file used. Am confused because the first set of linear iterations seem to converge and then on the start of the second nonlinear iteration it aborts. Would appreciate any help/advice. 

Hi,

We just had an interesting conversation about this problem here, which I think is ill-posed.

In 1D, the continuity equation reduces to du/dx = 0, where u is the x-component (aka the only component) of the velocity. This equation has only the solution u=const. Substituting u=const into the 1D x-momentum equation yields dp/dx=0, i.e. p=const. Thus the boundary conditions of p=1 and p=0 that you are trying to impose are not valid, hence the solver fails.

--

John

[Berry, Ray A]

Probably not ill-posed but could be unrealizable from a physical viewpoint.  For dp/dx=0 there would have also to be no wall friction for and no body force either.  Then there would be a continuous uniform acceleration, with a steadily increasing uniform velocity.

--
You received this message because you are subscribed to the Google Groups "moose-users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to moose-users+unsubscribe@googlegroups.com.
Visit this group at https://groups.google.com/group/moose-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/CANj%2BmwBjsKouGaQarhp7h8pD8yeQxORdUEsOLXvkefoQhGZS5A%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

--

Ray A. Berry, Ph.D.

Idaho National Laboratory

Idaho Falls, Idaho  U.S.A

[s.j...@usnc.com]

Thanks all for the comments, we agree, does seem ill-posed. Definitely missed that, thanks for the help!

On Friday, October 20, 2017 at 2:31:27 PM UTC-7, Ray Berry wrote:

Probably not ill-posed but could be unrealizable from a physical viewpoint.  For dp/dx=0 there would have also to be no wall friction for and no body force either.  Then there would be a continuous uniform acceleration, with a steadily increasing uniform velocity.

On Thu, Oct 19, 2017 at 1:51 PM, Peterson, JW <jw.pe...@inl.gov> wrote:

On Thu, Oct 19, 2017 at 12:20 PM, <s.j...@usnc.com> wrote:

Hi, new moose user here. I am trying to create a 1D pipe flow model using the incompressible navier stokes package, but having issues with convergence. I built off of the "pressure_channel" test case in navier stokes module, with the only changes being the removal of one dimension from the original input file and element type change from QUAD9 to EDGE3. Have attached moose output showing nonconvergence, and also input file used. Am confused because the first set of linear iterations seem to converge and then on the start of the second nonlinear iteration it aborts. Would appreciate any help/advice. 

Hi,

We just had an interesting conversation about this problem here, which I think is ill-posed.

In 1D, the continuity equation reduces to du/dx = 0, where u is the x-component (aka the only component) of the velocity. This equation has only the solution u=const. Substituting u=const into the 1D x-momentum equation yields dp/dx=0, i.e. p=const. Thus the boundary conditions of p=1 and p=0 that you are trying to impose are not valid, hence the solver fails.

--

John

--
You received this message because you are subscribed to the Google Groups "moose-users" group.

To unsubscribe from this group and stop receiving emails from it, send an email to moose-users...@googlegroups.com.

Visit this group at https://groups.google.com/group/moose-users.
To view this discussion on the web visit https://groups.google.com/d/msgid/moose-users/CANj%2BmwBjsKouGaQarhp7h8pD8yeQxORdUEsOLXvkefoQhGZS5A%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.