Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary? cheers ---------------------------------------------- Zhang Wei [email protected]
Hi Zhang, You can use periodic boundary conditions in the z direction, provided you have at least 3 elements in z. Prescribed boundary conditions are given in Cartesian coordinates. If you have the "mv " bc on your moving surface (i.e., the cylinder surface), along with the following things lx2=lx1-2 ly2=ly1-2 lz2=lz1-2 parameter (lx1m=lx1,ly1m=ly1,lz1m=lz1) in SIZE, and T ifstrs, T ifmvbd in the .rea file, then the mesh will move with the fluid velocity (which you prescribe as Cartesian velocity vectors, u,v,w). Note that nek uses an ALE formulation and the mesh motion must relatively limited or the mesh becomes too distorted. For an oscillating cylinder, this should be ok (see www.mcs.anl.gov/~fischer/ocyl.gif). I've just added a 2D oscillating cylinder case that illustrates the basic procedure to the examples suite. Paul ________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 8:40 AM To: [email protected] Subject: [Nek5000-users] quasi 3D and boundary condition Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary? cheers ---------------------------------------------- Zhang Wei [email protected] _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
Hi Paul Thanks for replying. for quasi-2D, I mean I wish to define the velocity u(x,y,z) as sum(u(x,y,k)e^kz) to reduce the computational cost for really 3D case and also to avoid the long wave (which may longer then the geometry size in the z direction) in the axial direction.does nek currently support such feature? for prescribe the bc on curl surface, I mean something like non-immersed bc and and have velocity component in the tangential direction, like in the potential theory. cheers ---------------------------------------------- Zhang Wei [email protected] On 8 Jun, 2014, at 10:44 pm, [email protected] wrote:
Hi Zhang,
You can use periodic boundary conditions in the z direction, provided you have at least 3 elements in z.
Prescribed boundary conditions are given in Cartesian coordinates. If you have the "mv " bc on your moving surface (i.e., the cylinder surface), along with the following things
lx2=lx1-2 ly2=ly1-2 lz2=lz1-2
parameter (lx1m=lx1,ly1m=ly1,lz1m=lz1)
in SIZE, and T ifstrs, T ifmvbd in the .rea file, then the mesh will move with the fluid velocity (which you prescribe as Cartesian velocity vectors, u,v,w). Note that nek uses an ALE formulation and the mesh motion must relatively limited or the mesh becomes too distorted. For an oscillating cylinder, this should be ok (see www.mcs.anl.gov/~fischer/ocyl.gif). I've just added a 2D oscillating cylinder case that illustrates the basic procedure to the examples suite.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 8:40 AM To: [email protected] Subject: [Nek5000-users] quasi 3D and boundary condition
Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary?
cheers ---------------------------------------------- Zhang Wei [email protected]
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
Hi Zhang, There is no support for 2.5D with Fourier in z and currently no plans to move in that direction. If you want a slip velocity on the curved surface you should use the stress formulation (lx2=lx1-2, etc., with ifstrs T. Then the "SYM" bc should give you the desired bc, i.e., U.nhat = 0, d/dn (U.t_hat) = 0. Paul _______________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 10:03 AM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition Hi Paul Thanks for replying. for quasi-2D, I mean I wish to define the velocity u(x,y,z) as sum(u(x,y,k)e^kz) to reduce the computational cost for really 3D case and also to avoid the long wave (which may longer then the geometry size in the z direction) in the axial direction.does nek currently support such feature? for prescribe the bc on curl surface, I mean something like non-immersed bc and and have velocity component in the tangential direction, like in the potential theory. cheers ---------------------------------------------- Zhang Wei [email protected] On 8 Jun, 2014, at 10:44 pm, [email protected] wrote:
Hi Zhang,
You can use periodic boundary conditions in the z direction, provided you have at least 3 elements in z.
Prescribed boundary conditions are given in Cartesian coordinates. If you have the "mv " bc on your moving surface (i.e., the cylinder surface), along with the following things
lx2=lx1-2 ly2=ly1-2 lz2=lz1-2
parameter (lx1m=lx1,ly1m=ly1,lz1m=lz1)
in SIZE, and T ifstrs, T ifmvbd in the .rea file, then the mesh will move with the fluid velocity (which you prescribe as Cartesian velocity vectors, u,v,w). Note that nek uses an ALE formulation and the mesh motion must relatively limited or the mesh becomes too distorted. For an oscillating cylinder, this should be ok (see www.mcs.anl.gov/~fischer/ocyl.gif). I've just added a 2D oscillating cylinder case that illustrates the basic procedure to the examples suite.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 8:40 AM To: [email protected] Subject: [Nek5000-users] quasi 3D and boundary condition
Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary?
cheers ---------------------------------------------- Zhang Wei [email protected]
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
How should I impose time-dependant BC for velocity on the left vertical boundaty for a 2D box geometry. The velocity profile should be Vx=cos(pi*y)*cos(2*pi*t) Vy=0 To illustrate the idea here are the pictures. Instead of constant velocity Vx=c on the left (inflow) boundary of a channel V __________________ | | | |__________________ I would impose Vx=cos(pi*y)*cos(2*pi*t) so for t = 0 velocity profile is ____________________ \ \ \ \_________________ for t=1 velocity profile is _________________ / / / /____________________ and so on. I know it should be possible in nek5000, but still not very familiar with syntax and fortran. Thank you in advance.
In userbc: ux = cos(pi*y)*cos(2.*pi*time) uy = 0. Paul ________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Monday, June 09, 2014 2:30 AM To: [email protected] Subject: [Nek5000-users] Time-dependant boundary condition How should I impose time-dependant BC for velocity on the left vertical boundaty for a 2D box geometry. The velocity profile should be Vx=cos(pi*y)*cos(2*pi*t) Vy=0 To illustrate the idea here are the pictures. Instead of constant velocity Vx=c on the left (inflow) boundary of a channel V __________________ | | | |__________________ I would impose Vx=cos(pi*y)*cos(2*pi*t) so for t = 0 velocity profile is ____________________ \ \ \ \_________________ for t=1 velocity profile is _________________ / / / /____________________ and so on. I know it should be possible in nek5000, but still not very familiar with syntax and fortran. Thank you in advance. _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
Thanks, I will try Van
09.06.2014, 16:06:05 пользователь ([email protected]) написал:
In userbc:
ux = cos(pi*y)*cos(2.*pi*time) uy = 0.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Monday, June 09, 2014 2:30 AM To: [email protected] Subject: [Nek5000-users] Time-dependant boundary condition
How should I impose time-dependant BC for velocity on the left vertical boundaty for a 2D box geometry. The velocity profile should be Vx=cos(pi*y)*cos(2*pi*t) Vy=0
To illustrate the idea here are the pictures. Instead of constant velocity Vx=c on the left (inflow) boundary of a channel V __________________ | | | |__________________
I would impose Vx=cos(pi*y)*cos(2*pi*t) so for t = 0 velocity profile is ____________________ \ \ \ \_________________ for t=1 velocity profile is _________________ / / / /____________________
and so on.
I know it should be possible in nek5000, but still not very familiar with syntax and fortran.
Thank you in advance. _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
Works greate! Van
6/9/2014 4:06:05 PM пользователь ([email protected]) написал:
In userbc:
ux = cos(pi*y)*cos(2.*pi*time) uy = 0.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Monday, June 09, 2014 2:30 AM To: [email protected] Subject: [Nek5000-users] Time-dependant boundary condition
How should I impose time-dependant BC for velocity on the left vertical boundaty for a 2D box geometry. The velocity profile should be Vx=cos(pi*y)*cos(2*pi*t) Vy=0
To illustrate the idea here are the pictures. Instead of constant velocity Vx=c on the left (inflow) boundary of a channel V __________________ | | | |__________________
I would impose Vx=cos(pi*y)*cos(2*pi*t) so for t = 0 velocity profile is ____________________ \ \ \ \_________________ for t=1 velocity profile is _________________ / / / /____________________
and so on.
I know it should be possible in nek5000, but still not very familiar with syntax and fortran.
Thank you in advance. _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
Hi Paul what is implementation of SYM BC in NEK5000? I mean the variation formula. cheers ---------------------------------------------- Zhang Wei [email protected] On 9 Jun, 2014, at 4:15 am, [email protected] wrote:
Hi Zhang,
There is no support for 2.5D with Fourier in z and currently no plans to move in that direction.
If you want a slip velocity on the curved surface you should use the stress formulation (lx2=lx1-2, etc., with ifstrs T. Then the "SYM" bc should give you the desired bc, i.e., U.nhat = 0, d/dn (U.t_hat) = 0.
Paul
_______________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 10:03 AM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul Thanks for replying. for quasi-2D, I mean I wish to define the velocity u(x,y,z) as sum(u(x,y,k)e^kz) to reduce the computational cost for really 3D case and also to avoid the long wave (which may longer then the geometry size in the z direction) in the axial direction.does nek currently support such feature?
for prescribe the bc on curl surface, I mean something like non-immersed bc and and have velocity component in the tangential direction, like in the potential theory.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 8 Jun, 2014, at 10:44 pm, [email protected] wrote:
Hi Zhang,
You can use periodic boundary conditions in the z direction, provided you have at least 3 elements in z.
Prescribed boundary conditions are given in Cartesian coordinates. If you have the "mv " bc on your moving surface (i.e., the cylinder surface), along with the following things
lx2=lx1-2 ly2=ly1-2 lz2=lz1-2
parameter (lx1m=lx1,ly1m=ly1,lz1m=lz1)
in SIZE, and T ifstrs, T ifmvbd in the .rea file, then the mesh will move with the fluid velocity (which you prescribe as Cartesian velocity vectors, u,v,w). Note that nek uses an ALE formulation and the mesh motion must relatively limited or the mesh becomes too distorted. For an oscillating cylinder, this should be ok (see www.mcs.anl.gov/~fischer/ocyl.gif). I've just added a 2D oscillating cylinder case that illustrates the basic procedure to the examples suite.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 8:40 AM To: [email protected] Subject: [Nek5000-users] quasi 3D and boundary condition
Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary?
cheers ---------------------------------------------- Zhang Wei [email protected]
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
Assuming that the surface normal is x, y, or z, then it is simply U.nhat = 0 and Neumann on the other two boundaries. Paul ________________________ ________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 15, 2014 9:15 PM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition Hi Paul what is implementation of SYM BC in NEK5000? I mean the variation formula. cheers ---------------------------------------------- Zhang Wei [email protected] On 9 Jun, 2014, at 4:15 am, [email protected] wrote:
Hi Zhang,
There is no support for 2.5D with Fourier in z and currently no plans to move in that direction.
If you want a slip velocity on the curved surface you should use the stress formulation (lx2=lx1-2, etc., with ifstrs T. Then the "SYM" bc should give you the desired bc, i.e., U.nhat = 0, d/dn (U.t_hat) = 0.
Paul
_______________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 10:03 AM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul Thanks for replying. for quasi-2D, I mean I wish to define the velocity u(x,y,z) as sum(u(x,y,k)e^kz) to reduce the computational cost for really 3D case and also to avoid the long wave (which may longer then the geometry size in the z direction) in the axial direction.does nek currently support such feature?
for prescribe the bc on curl surface, I mean something like non-immersed bc and and have velocity component in the tangential direction, like in the potential theory.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 8 Jun, 2014, at 10:44 pm, [email protected] wrote:
Hi Zhang,
You can use periodic boundary conditions in the z direction, provided you have at least 3 elements in z.
Prescribed boundary conditions are given in Cartesian coordinates. If you have the "mv " bc on your moving surface (i.e., the cylinder surface), along with the following things
lx2=lx1-2 ly2=ly1-2 lz2=lz1-2
parameter (lx1m=lx1,ly1m=ly1,lz1m=lz1)
in SIZE, and T ifstrs, T ifmvbd in the .rea file, then the mesh will move with the fluid velocity (which you prescribe as Cartesian velocity vectors, u,v,w). Note that nek uses an ALE formulation and the mesh motion must relatively limited or the mesh becomes too distorted. For an oscillating cylinder, this should be ok (see www.mcs.anl.gov/~fischer/ocyl.gif). I've just added a 2D oscillating cylinder case that illustrates the basic procedure to the examples suite.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 8:40 AM To: [email protected] Subject: [Nek5000-users] quasi 3D and boundary condition
Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary?
cheers ---------------------------------------------- Zhang Wei [email protected]
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
Hi Paul but for curved boundary, how you do so? I want to know the formula for what you mentioned as stress formula. cheers ---------------------------------------------- Zhang Wei [email protected] On 16 Jun, 2014, at 11:09 am, [email protected] wrote:
Assuming that the surface normal is x, y, or z, then it is simply
U.nhat = 0
and Neumann on the other two boundaries.
Paul
________________________ ________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 15, 2014 9:15 PM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul what is implementation of SYM BC in NEK5000? I mean the variation formula.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 9 Jun, 2014, at 4:15 am, [email protected] wrote:
Hi Zhang,
There is no support for 2.5D with Fourier in z and currently no plans to move in that direction.
If you want a slip velocity on the curved surface you should use the stress formulation (lx2=lx1-2, etc., with ifstrs T. Then the "SYM" bc should give you the desired bc, i.e., U.nhat = 0, d/dn (U.t_hat) = 0.
Paul
_______________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 10:03 AM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul Thanks for replying. for quasi-2D, I mean I wish to define the velocity u(x,y,z) as sum(u(x,y,k)e^kz) to reduce the computational cost for really 3D case and also to avoid the long wave (which may longer then the geometry size in the z direction) in the axial direction.does nek currently support such feature?
for prescribe the bc on curl surface, I mean something like non-immersed bc and and have velocity component in the tangential direction, like in the potential theory.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 8 Jun, 2014, at 10:44 pm, [email protected] wrote:
Hi Zhang,
You can use periodic boundary conditions in the z direction, provided you have at least 3 elements in z.
Prescribed boundary conditions are given in Cartesian coordinates. If you have the "mv " bc on your moving surface (i.e., the cylinder surface), along with the following things
lx2=lx1-2 ly2=ly1-2 lz2=lz1-2
parameter (lx1m=lx1,ly1m=ly1,lz1m=lz1)
in SIZE, and T ifstrs, T ifmvbd in the .rea file, then the mesh will move with the fluid velocity (which you prescribe as Cartesian velocity vectors, u,v,w). Note that nek uses an ALE formulation and the mesh motion must relatively limited or the mesh becomes too distorted. For an oscillating cylinder, this should be ok (see www.mcs.anl.gov/~fischer/ocyl.gif). I've just added a 2D oscillating cylinder case that illustrates the basic procedure to the examples suite.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 8:40 AM To: [email protected] Subject: [Nek5000-users] quasi 3D and boundary condition
Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary?
cheers ---------------------------------------------- Zhang Wei [email protected]
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
Hi, In that case, you need to use the stress formulation and the net result is U.n_hat = 0 and d/dn (U.t_hat) = 0, where t_hat constitutes the two tangent vectors. I've not tried this myself but I assume it works - it was written by Lee W. Ho when we were graduate students and I believe that others have used it. If I were you I would do a small test case in 2D first. -- I just tried it and it worked ok... you need to set ifstrs T in the .rea file and lx1m=lx1, etc. in the SIZE file. Paul ________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 15, 2014 10:13 PM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition Hi Paul but for curved boundary, how you do so? I want to know the formula for what you mentioned as stress formula. cheers ---------------------------------------------- Zhang Wei [email protected] On 16 Jun, 2014, at 11:09 am, [email protected] wrote:
Assuming that the surface normal is x, y, or z, then it is simply
U.nhat = 0
and Neumann on the other two boundaries.
Paul
________________________ ________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 15, 2014 9:15 PM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul what is implementation of SYM BC in NEK5000? I mean the variation formula.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 9 Jun, 2014, at 4:15 am, [email protected] wrote:
Hi Zhang,
There is no support for 2.5D with Fourier in z and currently no plans to move in that direction.
If you want a slip velocity on the curved surface you should use the stress formulation (lx2=lx1-2, etc., with ifstrs T. Then the "SYM" bc should give you the desired bc, i.e., U.nhat = 0, d/dn (U.t_hat) = 0.
Paul
_______________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 10:03 AM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul Thanks for replying. for quasi-2D, I mean I wish to define the velocity u(x,y,z) as sum(u(x,y,k)e^kz) to reduce the computational cost for really 3D case and also to avoid the long wave (which may longer then the geometry size in the z direction) in the axial direction.does nek currently support such feature?
for prescribe the bc on curl surface, I mean something like non-immersed bc and and have velocity component in the tangential direction, like in the potential theory.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 8 Jun, 2014, at 10:44 pm, [email protected] wrote:
Hi Zhang,
You can use periodic boundary conditions in the z direction, provided you have at least 3 elements in z.
Prescribed boundary conditions are given in Cartesian coordinates. If you have the "mv " bc on your moving surface (i.e., the cylinder surface), along with the following things
lx2=lx1-2 ly2=ly1-2 lz2=lz1-2
parameter (lx1m=lx1,ly1m=ly1,lz1m=lz1)
in SIZE, and T ifstrs, T ifmvbd in the .rea file, then the mesh will move with the fluid velocity (which you prescribe as Cartesian velocity vectors, u,v,w). Note that nek uses an ALE formulation and the mesh motion must relatively limited or the mesh becomes too distorted. For an oscillating cylinder, this should be ok (see www.mcs.anl.gov/~fischer/ocyl.gif). I've just added a 2D oscillating cylinder case that illustrates the basic procedure to the examples suite.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 8:40 AM To: [email protected] Subject: [Nek5000-users] quasi 3D and boundary condition
Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary?
cheers ---------------------------------------------- Zhang Wei [email protected]
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
Hi Paul. is there any paper of about that? I would know more math work of this implementation. cheers ---------------------------------------------- Zhang Wei [email protected] On 16 Jun, 2014, at 11:34 am, [email protected] wrote:
Hi,
In that case, you need to use the stress formulation and the net result is U.n_hat = 0 and d/dn (U.t_hat) = 0, where t_hat constitutes the two tangent vectors.
I've not tried this myself but I assume it works - it was written by Lee W. Ho when we were graduate students and I believe that others have used it.
If I were you I would do a small test case in 2D first. -- I just tried it and it worked ok... you need to set ifstrs T in the .rea file and lx1m=lx1, etc. in the SIZE file.
Paul ________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 15, 2014 10:13 PM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul
but for curved boundary, how you do so? I want to know the formula for what you mentioned as stress formula.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 16 Jun, 2014, at 11:09 am, [email protected] wrote:
Assuming that the surface normal is x, y, or z, then it is simply
U.nhat = 0
and Neumann on the other two boundaries.
Paul
________________________ ________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 15, 2014 9:15 PM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul what is implementation of SYM BC in NEK5000? I mean the variation formula.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 9 Jun, 2014, at 4:15 am, [email protected] wrote:
Hi Zhang,
There is no support for 2.5D with Fourier in z and currently no plans to move in that direction.
If you want a slip velocity on the curved surface you should use the stress formulation (lx2=lx1-2, etc., with ifstrs T. Then the "SYM" bc should give you the desired bc, i.e., U.nhat = 0, d/dn (U.t_hat) = 0.
Paul
_______________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 10:03 AM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul Thanks for replying. for quasi-2D, I mean I wish to define the velocity u(x,y,z) as sum(u(x,y,k)e^kz) to reduce the computational cost for really 3D case and also to avoid the long wave (which may longer then the geometry size in the z direction) in the axial direction.does nek currently support such feature?
for prescribe the bc on curl surface, I mean something like non-immersed bc and and have velocity component in the tangential direction, like in the potential theory.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 8 Jun, 2014, at 10:44 pm, [email protected] wrote:
Hi Zhang,
You can use periodic boundary conditions in the z direction, provided you have at least 3 elements in z.
Prescribed boundary conditions are given in Cartesian coordinates. If you have the "mv " bc on your moving surface (i.e., the cylinder surface), along with the following things
lx2=lx1-2 ly2=ly1-2 lz2=lz1-2
parameter (lx1m=lx1,ly1m=ly1,lz1m=lz1)
in SIZE, and T ifstrs, T ifmvbd in the .rea file, then the mesh will move with the fluid velocity (which you prescribe as Cartesian velocity vectors, u,v,w). Note that nek uses an ALE formulation and the mesh motion must relatively limited or the mesh becomes too distorted. For an oscillating cylinder, this should be ok (see www.mcs.anl.gov/~fischer/ocyl.gif). I've just added a 2D oscillating cylinder case that illustrates the basic procedure to the examples suite.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 8:40 AM To: [email protected] Subject: [Nek5000-users] quasi 3D and boundary condition
Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary?
cheers ---------------------------------------------- Zhang Wei [email protected]
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
You can find Lee Ho's MIT Ph.D. thesis online - 1989. It's really just the variational formulation for the Stokes operator. Paul ________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 15, 2014 10:36 PM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition Hi Paul. is there any paper of about that? I would know more math work of this implementation. cheers ---------------------------------------------- Zhang Wei [email protected] On 16 Jun, 2014, at 11:34 am, [email protected] wrote:
Hi,
In that case, you need to use the stress formulation and the net result is U.n_hat = 0 and d/dn (U.t_hat) = 0, where t_hat constitutes the two tangent vectors.
I've not tried this myself but I assume it works - it was written by Lee W. Ho when we were graduate students and I believe that others have used it.
If I were you I would do a small test case in 2D first. -- I just tried it and it worked ok... you need to set ifstrs T in the .rea file and lx1m=lx1, etc. in the SIZE file.
Paul ________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 15, 2014 10:13 PM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul
but for curved boundary, how you do so? I want to know the formula for what you mentioned as stress formula.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 16 Jun, 2014, at 11:09 am, [email protected] wrote:
Assuming that the surface normal is x, y, or z, then it is simply
U.nhat = 0
and Neumann on the other two boundaries.
Paul
________________________ ________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 15, 2014 9:15 PM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul what is implementation of SYM BC in NEK5000? I mean the variation formula.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 9 Jun, 2014, at 4:15 am, [email protected] wrote:
Hi Zhang,
There is no support for 2.5D with Fourier in z and currently no plans to move in that direction.
If you want a slip velocity on the curved surface you should use the stress formulation (lx2=lx1-2, etc., with ifstrs T. Then the "SYM" bc should give you the desired bc, i.e., U.nhat = 0, d/dn (U.t_hat) = 0.
Paul
_______________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 10:03 AM To: [email protected] Subject: Re: [Nek5000-users] quasi 3D and boundary condition
Hi Paul Thanks for replying. for quasi-2D, I mean I wish to define the velocity u(x,y,z) as sum(u(x,y,k)e^kz) to reduce the computational cost for really 3D case and also to avoid the long wave (which may longer then the geometry size in the z direction) in the axial direction.does nek currently support such feature?
for prescribe the bc on curl surface, I mean something like non-immersed bc and and have velocity component in the tangential direction, like in the potential theory.
cheers ---------------------------------------------- Zhang Wei [email protected]
On 8 Jun, 2014, at 10:44 pm, [email protected] wrote:
Hi Zhang,
You can use periodic boundary conditions in the z direction, provided you have at least 3 elements in z.
Prescribed boundary conditions are given in Cartesian coordinates. If you have the "mv " bc on your moving surface (i.e., the cylinder surface), along with the following things
lx2=lx1-2 ly2=ly1-2 lz2=lz1-2
parameter (lx1m=lx1,ly1m=ly1,lz1m=lz1)
in SIZE, and T ifstrs, T ifmvbd in the .rea file, then the mesh will move with the fluid velocity (which you prescribe as Cartesian velocity vectors, u,v,w). Note that nek uses an ALE formulation and the mesh motion must relatively limited or the mesh becomes too distorted. For an oscillating cylinder, this should be ok (see www.mcs.anl.gov/~fischer/ocyl.gif). I've just added a 2D oscillating cylinder case that illustrates the basic procedure to the examples suite.
Paul
________________________________________ From: [email protected] [[email protected]] on behalf of [email protected] [[email protected]] Sent: Sunday, June 08, 2014 8:40 AM To: [email protected] Subject: [Nek5000-users] quasi 3D and boundary condition
Hi Everyone. I am trying to run the flow past a cylinder case. I am going to move to the 3D simulation for my study. I have two questions: 1) how to run quasi 3D simulation in nek5000 ? if there is not existing code, andy suggestion to how to implement so? 2) how to specify the normal and tangential boundary condition for a curvature boundary?
cheers ---------------------------------------------- Zhang Wei [email protected]
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
participants (1)
-
nek5000-users＠lists.mcs.anl.gov