From nek5000-users at lists.mcs.anl.gov Thu Jun 1 03:12:02 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 1 Jun 2017 10:12:02 +0200 Subject: [Nek5000-users] Is turbulent model RANS available in Nek5000? In-Reply-To: References: Message-ID: At some point there was a k-eps implementation in nek. However I don't think it works/exists anymore.? On 1 Jun 2017, at 04:42, "nek5000-users at lists.mcs.anl.gov " > wrote: Hi all, I found several settings about RANS in Nek5000, but no instruction about it. I checked the output settings in subroutine rdout, and found only "X","U","P","T" can be exported. I thought the turbulence kinetic energy and dissipation ratio may be contained in passive scalars,?then I added?two passive scalars to export ,and found the values are all zero.? I really?waste lots of time to read the programs.Now I?am confused whether RANS is available in Nek5000? If anybody knows, please tell me is turbulent model RANS available in Nek5000? If not, then I can pay attention to other works. Thanks a lot!! Hu ? _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Jun 1 07:27:00 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 1 Jun 2017 17:57:00 +0530 Subject: [Nek5000-users] Corner elements giving large values Message-ID: Dear Nek forum, I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. I am getting spurious large values of velocity at the corner regions of the annulus i.e. the edges where the inner cylinder joins the top and bottom planes. The boundary condition is no-slip and no penetration for velocity at these edges. Can someone point out a plausible reason. Thank you Swarandeep From nek5000-users at lists.mcs.anl.gov Thu Jun 1 09:08:25 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 1 Jun 2017 14:08:25 +0000 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: Message-ID: Hi Swarandeep, How large is your Rayleigh number and what is your resolution? What are your thermal bcs? What are your ICs? There should be no particular issue with your configuration - but you do need to control spatial and temporal resolution. For RB, you often have to start at low Ra and/or small dt because the nonlinear response can often drive the velocities to be quite high very rapidly before the CFL constraint kicks in. RB flows are one of the rare instances where I use variable time step size (param 12 > 0), but even then you have to keep the target CFL quite small until the flow is established. hth, Paul ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Thursday, June 01, 2017 7:27 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Corner elements giving large values Dear Nek forum, I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. I am getting spurious large values of velocity at the corner regions of the annulus i.e. the edges where the inner cylinder joins the top and bottom planes. The boundary condition is no-slip and no penetration for velocity at these edges. Can someone point out a plausible reason. Thank you Swarandeep _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Jun 1 12:53:32 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 1 Jun 2017 23:23:32 +0530 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: Message-ID: Hi Nek Forum, Paul I am having temp = 1 and temp = 0 at the inner and outer boundaries. Gravity is pointing in the -ve z-direction along the axis of the annulus. The top and bottom planes have zero heat flux conditions. I have Coriolis force added as a body force. The Rayleigh number is around 10000. The resolution is 8, 25, 24 elements in the r, phi and z directions respectively with order 9. My initial condition is very low arbitrary velocity perturbation (~10^(-8)). The time resolution is set to 10^(-5) for param12. The CFL is set to 0.5. Please indicate how to control it. As you have rightly mentioned, I start the simulation with a very low value of the Rayleigh number and gradually increase it. Yet, the corner values only get large. Other regions of the domain are fine. Thank you Swarandeep On Thu, Jun 1, 2017 at 7:38 PM, wrote: > > Hi Swarandeep, > > How large is your Rayleigh number and what is your resolution? > What are your thermal bcs? What are your ICs? > > There should be no particular issue with your configuration - but you do need to > control spatial and temporal resolution. > > > For RB, you often have to start at low Ra and/or small dt because the nonlinear > response can often drive the velocities to be quite high very rapidly before the > CFL constraint kicks in. RB flows are one of the rare instances where I use > variable time step size (param 12 > 0), but even then you have to keep the target > CFL quite small until the flow is established. > > hth, > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Thursday, June 01, 2017 7:27 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] Corner elements giving large values > > Dear Nek forum, > > I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. > I am getting spurious large values of velocity at the corner regions > of the annulus i.e. the edges where the inner cylinder joins the > top and bottom > planes. The boundary condition is no-slip and no penetration for velocity > at these edges. Can someone point out a plausible reason. > > Thank you > Swarandeep > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Jun 1 12:58:33 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 1 Jun 2017 17:58:33 +0000 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: , Message-ID: Hi Swarandeep, Is 0/1 a realistic BC for T ? Best, Paul ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Thursday, June 01, 2017 12:53 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Corner elements giving large values Hi Nek Forum, Paul I am having temp = 1 and temp = 0 at the inner and outer boundaries. Gravity is pointing in the -ve z-direction along the axis of the annulus. The top and bottom planes have zero heat flux conditions. I have Coriolis force added as a body force. The Rayleigh number is around 10000. The resolution is 8, 25, 24 elements in the r, phi and z directions respectively with order 9. My initial condition is very low arbitrary velocity perturbation (~10^(-8)). The time resolution is set to 10^(-5) for param12. The CFL is set to 0.5. Please indicate how to control it. As you have rightly mentioned, I start the simulation with a very low value of the Rayleigh number and gradually increase it. Yet, the corner values only get large. Other regions of the domain are fine. Thank you Swarandeep On Thu, Jun 1, 2017 at 7:38 PM, wrote: > > Hi Swarandeep, > > How large is your Rayleigh number and what is your resolution? > What are your thermal bcs? What are your ICs? > > There should be no particular issue with your configuration - but you do need to > control spatial and temporal resolution. > > > For RB, you often have to start at low Ra and/or small dt because the nonlinear > response can often drive the velocities to be quite high very rapidly before the > CFL constraint kicks in. RB flows are one of the rare instances where I use > variable time step size (param 12 > 0), but even then you have to keep the target > CFL quite small until the flow is established. > > hth, > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Thursday, June 01, 2017 7:27 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] Corner elements giving large values > > Dear Nek forum, > > I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. > I am getting spurious large values of velocity at the corner regions > of the annulus i.e. the edges where the inner cylinder joins the > top and bottom > planes. The boundary condition is no-slip and no penetration for velocity > at these edges. Can someone point out a plausible reason. > > Thank you > Swarandeep > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Jun 1 13:57:30 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 2 Jun 2017 00:27:30 +0530 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: Message-ID: Hi NekForum, Paul I intent to have isothermal conditions at the inner and outer cylinders. The differential heating is achieved by imposing a temperature difference across the two cylinders. I have used T = 1 and 0 values for non-dimensional temperature. I think it should be fine. Do you see any problem here. Thank you Swarandeep On Thu, Jun 1, 2017 at 11:28 PM, wrote: > > Hi Swarandeep, > > Is 0/1 a realistic BC for T ? > > Best, > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Thursday, June 01, 2017 12:53 PM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > > Hi Nek Forum, Paul > > I am having temp = 1 and temp = 0 at the inner and outer > boundaries. Gravity is pointing in the -ve z-direction along the axis > of the > annulus. The top and bottom planes have zero heat flux conditions. > I have Coriolis force added as a body force. > The Rayleigh number is around 10000. The resolution is 8, 25, > 24 elements in the r, phi and z directions respectively with order 9. > My initial condition is very low arbitrary velocity > perturbation (~10^(-8)). The time resolution is set to 10^(-5) for > param12. The CFL is > set to 0.5. Please indicate how to control it. > > As you have rightly mentioned, I start the simulation with a > very low value of the Rayleigh number and gradually increase it. Yet, > the corner values only get large. Other regions of the domain are > fine. > > Thank you > Swarandeep > > On Thu, Jun 1, 2017 at 7:38 PM, wrote: >> >> Hi Swarandeep, >> >> How large is your Rayleigh number and what is your resolution? >> What are your thermal bcs? What are your ICs? >> >> There should be no particular issue with your configuration - but you do need to >> control spatial and temporal resolution. >> >> >> For RB, you often have to start at low Ra and/or small dt because the nonlinear >> response can often drive the velocities to be quite high very rapidly before the >> CFL constraint kicks in. RB flows are one of the rare instances where I use >> variable time step size (param 12 > 0), but even then you have to keep the target >> CFL quite small until the flow is established. >> >> hth, >> >> Paul >> >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Thursday, June 01, 2017 7:27 AM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: [Nek5000-users] Corner elements giving large values >> >> Dear Nek forum, >> >> I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. >> I am getting spurious large values of velocity at the corner regions >> of the annulus i.e. the edges where the inner cylinder joins the >> top and bottom >> planes. The boundary condition is no-slip and no penetration for velocity >> at these edges. Can someone point out a plausible reason. >> >> Thank you >> Swarandeep >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Jun 1 14:20:36 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 1 Jun 2017 19:20:36 +0000 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: , Message-ID: Hi Swarandeep, Sorry - I misunderstood. I guess my question was more about the temperature bcs at the bottom and top of the domain? In general, however, there should be no difficulty. What happens when you reduce your dt ? You need to keep the CFL < ~0.5 in general. CFL is printed in the logfile, grep tep logfile and you'll see it "C= ....." Paul ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Thursday, June 01, 2017 1:57 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Corner elements giving large values Hi NekForum, Paul I intent to have isothermal conditions at the inner and outer cylinders. The differential heating is achieved by imposing a temperature difference across the two cylinders. I have used T = 1 and 0 values for non-dimensional temperature. I think it should be fine. Do you see any problem here. Thank you Swarandeep On Thu, Jun 1, 2017 at 11:28 PM, wrote: > > Hi Swarandeep, > > Is 0/1 a realistic BC for T ? > > Best, > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Thursday, June 01, 2017 12:53 PM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > > Hi Nek Forum, Paul > > I am having temp = 1 and temp = 0 at the inner and outer > boundaries. Gravity is pointing in the -ve z-direction along the axis > of the > annulus. The top and bottom planes have zero heat flux conditions. > I have Coriolis force added as a body force. > The Rayleigh number is around 10000. The resolution is 8, 25, > 24 elements in the r, phi and z directions respectively with order 9. > My initial condition is very low arbitrary velocity > perturbation (~10^(-8)). The time resolution is set to 10^(-5) for > param12. The CFL is > set to 0.5. Please indicate how to control it. > > As you have rightly mentioned, I start the simulation with a > very low value of the Rayleigh number and gradually increase it. Yet, > the corner values only get large. Other regions of the domain are > fine. > > Thank you > Swarandeep > > On Thu, Jun 1, 2017 at 7:38 PM, wrote: >> >> Hi Swarandeep, >> >> How large is your Rayleigh number and what is your resolution? >> What are your thermal bcs? What are your ICs? >> >> There should be no particular issue with your configuration - but you do need to >> control spatial and temporal resolution. >> >> >> For RB, you often have to start at low Ra and/or small dt because the nonlinear >> response can often drive the velocities to be quite high very rapidly before the >> CFL constraint kicks in. RB flows are one of the rare instances where I use >> variable time step size (param 12 > 0), but even then you have to keep the target >> CFL quite small until the flow is established. >> >> hth, >> >> Paul >> >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Thursday, June 01, 2017 7:27 AM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: [Nek5000-users] Corner elements giving large values >> >> Dear Nek forum, >> >> I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. >> I am getting spurious large values of velocity at the corner regions >> of the annulus i.e. the edges where the inner cylinder joins the >> top and bottom >> planes. The boundary condition is no-slip and no penetration for velocity >> at these edges. Can someone point out a plausible reason. >> >> Thank you >> Swarandeep >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Jun 1 19:28:28 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 2 Jun 2017 08:28:28 +0800 (GMT+08:00) Subject: [Nek5000-users] Is turbulent model RANS available in Nek5000? In-Reply-To: References: Message-ID: Thank you for you reply! Best wishes! Hu -----Original Messages----- From: nek5000-users at lists.mcs.anl.gov Sent Time: Thursday, June 1, 2017 To: "nek5000-users at lists.mcs.anl.gov" Cc: Subject: Re: [Nek5000-users] Is turbulent model RANS available in Nek5000? At some point there was a k-eps implementation in nek. However I don't think it works/exists anymore. On 1 Jun 2017, at 04:42, "nek5000-users at lists.mcs.anl.gov" wrote: Hi all, I found several settings about RANS in Nek5000, but no instruction about it. I checked the output settings in subroutine rdout, and found only "X","U","P","T" can be exported. I thought the turbulence kinetic energy and dissipation ratio may be contained in passive scalars, then I added two passive scalars to export ,and found the values are all zero. I really waste lots of time to read the programs.Now I am confused whether RANS is available in Nek5000? If anybody knows, please tell me is turbulent model RANS available in Nek5000? If not, then I can pay attention to other works. Thanks a lot!! Hu _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Jun 2 01:04:08 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 2 Jun 2017 11:34:08 +0530 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: Message-ID: Hi Nek Forum, Paul The top and bottom BC are zero heat flux. Please let me n=know how to control the CFL. It is 0.5 in the logfile. I am having no particular difficulty except these corner elements at the junction of inner cylinder and the top/bottom planes. Is there a better way to treat edge boundary elements. Also is it possible to have different order of elements at different locations. Thank you Swarandeep On Fri, Jun 2, 2017 at 12:50 AM, wrote: > > Hi Swarandeep, > > Sorry - I misunderstood. I guess my question was more about the > temperature bcs at the bottom and top of the domain? > > In general, however, there should be no difficulty. What happens > when you reduce your dt ? > > You need to keep the CFL < ~0.5 in general. CFL is printed in the logfile, > > grep tep logfile > > and you'll see it "C= ....." > > Paul > > > > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Thursday, June 01, 2017 1:57 PM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > > Hi NekForum, Paul > > I intent to have isothermal conditions at the inner and outer > cylinders. The differential heating is achieved by imposing a > temperature difference across the two cylinders. I have used T = 1 and > 0 values for non-dimensional temperature. I think it should be fine. > Do you see any problem here. > > Thank you > Swarandeep > > On Thu, Jun 1, 2017 at 11:28 PM, wrote: >> >> Hi Swarandeep, >> >> Is 0/1 a realistic BC for T ? >> >> Best, >> >> Paul >> >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Thursday, June 01, 2017 12:53 PM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Corner elements giving large values >> >> Hi Nek Forum, Paul >> >> I am having temp = 1 and temp = 0 at the inner and outer >> boundaries. Gravity is pointing in the -ve z-direction along the axis >> of the >> annulus. The top and bottom planes have zero heat flux conditions. >> I have Coriolis force added as a body force. >> The Rayleigh number is around 10000. The resolution is 8, 25, >> 24 elements in the r, phi and z directions respectively with order 9. >> My initial condition is very low arbitrary velocity >> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for >> param12. The CFL is >> set to 0.5. Please indicate how to control it. >> >> As you have rightly mentioned, I start the simulation with a >> very low value of the Rayleigh number and gradually increase it. Yet, >> the corner values only get large. Other regions of the domain are >> fine. >> >> Thank you >> Swarandeep >> >> On Thu, Jun 1, 2017 at 7:38 PM, wrote: >>> >>> Hi Swarandeep, >>> >>> How large is your Rayleigh number and what is your resolution? >>> What are your thermal bcs? What are your ICs? >>> >>> There should be no particular issue with your configuration - but you do need to >>> control spatial and temporal resolution. >>> >>> >>> For RB, you often have to start at low Ra and/or small dt because the nonlinear >>> response can often drive the velocities to be quite high very rapidly before the >>> CFL constraint kicks in. RB flows are one of the rare instances where I use >>> variable time step size (param 12 > 0), but even then you have to keep the target >>> CFL quite small until the flow is established. >>> >>> hth, >>> >>> Paul >>> >>> ________________________________________ >>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>> Sent: Thursday, June 01, 2017 7:27 AM >>> To: nek5000-users at lists.mcs.anl.gov >>> Subject: [Nek5000-users] Corner elements giving large values >>> >>> Dear Nek forum, >>> >>> I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. >>> I am getting spurious large values of velocity at the corner regions >>> of the annulus i.e. the edges where the inner cylinder joins the >>> top and bottom >>> planes. The boundary condition is no-slip and no penetration for velocity >>> at these edges. Can someone point out a plausible reason. >>> >>> Thank you >>> Swarandeep >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Fri Jun 2 09:30:27 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 2 Jun 2017 14:30:27 +0000 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: , Message-ID: Dear Swarandeep, Did you refine your element size at the top and bottom and near the side walls? Please set your CFL to 0.05 in order to get the flow started. Best, Paul ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Friday, June 02, 2017 1:04 AM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Corner elements giving large values Hi Nek Forum, Paul The top and bottom BC are zero heat flux. Please let me n=know how to control the CFL. It is 0.5 in the logfile. I am having no particular difficulty except these corner elements at the junction of inner cylinder and the top/bottom planes. Is there a better way to treat edge boundary elements. Also is it possible to have different order of elements at different locations. Thank you Swarandeep On Fri, Jun 2, 2017 at 12:50 AM, wrote: > > Hi Swarandeep, > > Sorry - I misunderstood. I guess my question was more about the > temperature bcs at the bottom and top of the domain? > > In general, however, there should be no difficulty. What happens > when you reduce your dt ? > > You need to keep the CFL < ~0.5 in general. CFL is printed in the logfile, > > grep tep logfile > > and you'll see it "C= ....." > > Paul > > > > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Thursday, June 01, 2017 1:57 PM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > > Hi NekForum, Paul > > I intent to have isothermal conditions at the inner and outer > cylinders. The differential heating is achieved by imposing a > temperature difference across the two cylinders. I have used T = 1 and > 0 values for non-dimensional temperature. I think it should be fine. > Do you see any problem here. > > Thank you > Swarandeep > > On Thu, Jun 1, 2017 at 11:28 PM, wrote: >> >> Hi Swarandeep, >> >> Is 0/1 a realistic BC for T ? >> >> Best, >> >> Paul >> >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Thursday, June 01, 2017 12:53 PM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Corner elements giving large values >> >> Hi Nek Forum, Paul >> >> I am having temp = 1 and temp = 0 at the inner and outer >> boundaries. Gravity is pointing in the -ve z-direction along the axis >> of the >> annulus. The top and bottom planes have zero heat flux conditions. >> I have Coriolis force added as a body force. >> The Rayleigh number is around 10000. The resolution is 8, 25, >> 24 elements in the r, phi and z directions respectively with order 9. >> My initial condition is very low arbitrary velocity >> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for >> param12. The CFL is >> set to 0.5. Please indicate how to control it. >> >> As you have rightly mentioned, I start the simulation with a >> very low value of the Rayleigh number and gradually increase it. Yet, >> the corner values only get large. Other regions of the domain are >> fine. >> >> Thank you >> Swarandeep >> >> On Thu, Jun 1, 2017 at 7:38 PM, wrote: >>> >>> Hi Swarandeep, >>> >>> How large is your Rayleigh number and what is your resolution? >>> What are your thermal bcs? What are your ICs? >>> >>> There should be no particular issue with your configuration - but you do need to >>> control spatial and temporal resolution. >>> >>> >>> For RB, you often have to start at low Ra and/or small dt because the nonlinear >>> response can often drive the velocities to be quite high very rapidly before the >>> CFL constraint kicks in. RB flows are one of the rare instances where I use >>> variable time step size (param 12 > 0), but even then you have to keep the target >>> CFL quite small until the flow is established. >>> >>> hth, >>> >>> Paul >>> >>> ________________________________________ >>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>> Sent: Thursday, June 01, 2017 7:27 AM >>> To: nek5000-users at lists.mcs.anl.gov >>> Subject: [Nek5000-users] Corner elements giving large values >>> >>> Dear Nek forum, >>> >>> I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. >>> I am getting spurious large values of velocity at the corner regions >>> of the annulus i.e. the edges where the inner cylinder joins the >>> top and bottom >>> planes. The boundary condition is no-slip and no penetration for velocity >>> at these edges. Can someone point out a plausible reason. >>> >>> Thank you >>> Swarandeep >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Fri Jun 2 10:31:22 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 2 Jun 2017 21:01:22 +0530 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: Message-ID: Dear Nek Forum, Paul Yes, I have user defined grid such that the locations of the elements are distributed to lie on the chebyshev grid at the top and bottom and near the side walls. Also please tell which param sets the CFL. Thank you Swarandeep On Fri, Jun 2, 2017 at 8:00 PM, wrote: > > Dear Swarandeep, > > Did you refine your element size at the top and bottom and near the side walls? > > Please set your CFL to 0.05 in order to get the flow started. > > Best, > > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Friday, June 02, 2017 1:04 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > > Hi Nek Forum, Paul > > The top and bottom BC are zero heat flux. Please let me n=know > how to control the CFL. It is 0.5 in the logfile. I am having no > particular difficulty except these corner elements at the junction of > inner cylinder and the top/bottom planes. Is there a better way to > treat edge boundary elements. Also is it possible to have different > order of elements at different locations. > > Thank you > Swarandeep > > On Fri, Jun 2, 2017 at 12:50 AM, wrote: >> >> Hi Swarandeep, >> >> Sorry - I misunderstood. I guess my question was more about the >> temperature bcs at the bottom and top of the domain? >> >> In general, however, there should be no difficulty. What happens >> when you reduce your dt ? >> >> You need to keep the CFL < ~0.5 in general. CFL is printed in the logfile, >> >> grep tep logfile >> >> and you'll see it "C= ....." >> >> Paul >> >> >> >> >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Thursday, June 01, 2017 1:57 PM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Corner elements giving large values >> >> Hi NekForum, Paul >> >> I intent to have isothermal conditions at the inner and outer >> cylinders. The differential heating is achieved by imposing a >> temperature difference across the two cylinders. I have used T = 1 and >> 0 values for non-dimensional temperature. I think it should be fine. >> Do you see any problem here. >> >> Thank you >> Swarandeep >> >> On Thu, Jun 1, 2017 at 11:28 PM, wrote: >>> >>> Hi Swarandeep, >>> >>> Is 0/1 a realistic BC for T ? >>> >>> Best, >>> >>> Paul >>> >>> ________________________________________ >>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>> Sent: Thursday, June 01, 2017 12:53 PM >>> To: nek5000-users at lists.mcs.anl.gov >>> Subject: Re: [Nek5000-users] Corner elements giving large values >>> >>> Hi Nek Forum, Paul >>> >>> I am having temp = 1 and temp = 0 at the inner and outer >>> boundaries. Gravity is pointing in the -ve z-direction along the axis >>> of the >>> annulus. The top and bottom planes have zero heat flux conditions. >>> I have Coriolis force added as a body force. >>> The Rayleigh number is around 10000. The resolution is 8, 25, >>> 24 elements in the r, phi and z directions respectively with order 9. >>> My initial condition is very low arbitrary velocity >>> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for >>> param12. The CFL is >>> set to 0.5. Please indicate how to control it. >>> >>> As you have rightly mentioned, I start the simulation with a >>> very low value of the Rayleigh number and gradually increase it. Yet, >>> the corner values only get large. Other regions of the domain are >>> fine. >>> >>> Thank you >>> Swarandeep >>> >>> On Thu, Jun 1, 2017 at 7:38 PM, wrote: >>>> >>>> Hi Swarandeep, >>>> >>>> How large is your Rayleigh number and what is your resolution? >>>> What are your thermal bcs? What are your ICs? >>>> >>>> There should be no particular issue with your configuration - but you do need to >>>> control spatial and temporal resolution. >>>> >>>> >>>> For RB, you often have to start at low Ra and/or small dt because the nonlinear >>>> response can often drive the velocities to be quite high very rapidly before the >>>> CFL constraint kicks in. RB flows are one of the rare instances where I use >>>> variable time step size (param 12 > 0), but even then you have to keep the target >>>> CFL quite small until the flow is established. >>>> >>>> hth, >>>> >>>> Paul >>>> >>>> ________________________________________ >>>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>>> Sent: Thursday, June 01, 2017 7:27 AM >>>> To: nek5000-users at lists.mcs.anl.gov >>>> Subject: [Nek5000-users] Corner elements giving large values >>>> >>>> Dear Nek forum, >>>> >>>> I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. >>>> I am getting spurious large values of velocity at the corner regions >>>> of the annulus i.e. the edges where the inner cylinder joins the >>>> top and bottom >>>> planes. The boundary condition is no-slip and no penetration for velocity >>>> at these edges. Can someone point out a plausible reason. >>>> >>>> Thank you >>>> Swarandeep >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Fri Jun 2 11:58:06 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 2 Jun 2017 16:58:06 +0000 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: , Message-ID: Hi Swarandeep, Set p26 in the .rea file to be 0.1 and make certain IFCHAR is F (a few lines below); Make certain your dt is positive (param 12) Paul ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Friday, June 02, 2017 10:31 AM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Corner elements giving large values Dear Nek Forum, Paul Yes, I have user defined grid such that the locations of the elements are distributed to lie on the chebyshev grid at the top and bottom and near the side walls. Also please tell which param sets the CFL. Thank you Swarandeep On Fri, Jun 2, 2017 at 8:00 PM, wrote: > > Dear Swarandeep, > > Did you refine your element size at the top and bottom and near the side walls? > > Please set your CFL to 0.05 in order to get the flow started. > > Best, > > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Friday, June 02, 2017 1:04 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > > Hi Nek Forum, Paul > > The top and bottom BC are zero heat flux. Please let me n=know > how to control the CFL. It is 0.5 in the logfile. I am having no > particular difficulty except these corner elements at the junction of > inner cylinder and the top/bottom planes. Is there a better way to > treat edge boundary elements. Also is it possible to have different > order of elements at different locations. > > Thank you > Swarandeep > > On Fri, Jun 2, 2017 at 12:50 AM, wrote: >> >> Hi Swarandeep, >> >> Sorry - I misunderstood. I guess my question was more about the >> temperature bcs at the bottom and top of the domain? >> >> In general, however, there should be no difficulty. What happens >> when you reduce your dt ? >> >> You need to keep the CFL < ~0.5 in general. CFL is printed in the logfile, >> >> grep tep logfile >> >> and you'll see it "C= ....." >> >> Paul >> >> >> >> >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Thursday, June 01, 2017 1:57 PM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Corner elements giving large values >> >> Hi NekForum, Paul >> >> I intent to have isothermal conditions at the inner and outer >> cylinders. The differential heating is achieved by imposing a >> temperature difference across the two cylinders. I have used T = 1 and >> 0 values for non-dimensional temperature. I think it should be fine. >> Do you see any problem here. >> >> Thank you >> Swarandeep >> >> On Thu, Jun 1, 2017 at 11:28 PM, wrote: >>> >>> Hi Swarandeep, >>> >>> Is 0/1 a realistic BC for T ? >>> >>> Best, >>> >>> Paul >>> >>> ________________________________________ >>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>> Sent: Thursday, June 01, 2017 12:53 PM >>> To: nek5000-users at lists.mcs.anl.gov >>> Subject: Re: [Nek5000-users] Corner elements giving large values >>> >>> Hi Nek Forum, Paul >>> >>> I am having temp = 1 and temp = 0 at the inner and outer >>> boundaries. Gravity is pointing in the -ve z-direction along the axis >>> of the >>> annulus. The top and bottom planes have zero heat flux conditions. >>> I have Coriolis force added as a body force. >>> The Rayleigh number is around 10000. The resolution is 8, 25, >>> 24 elements in the r, phi and z directions respectively with order 9. >>> My initial condition is very low arbitrary velocity >>> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for >>> param12. The CFL is >>> set to 0.5. Please indicate how to control it. >>> >>> As you have rightly mentioned, I start the simulation with a >>> very low value of the Rayleigh number and gradually increase it. Yet, >>> the corner values only get large. Other regions of the domain are >>> fine. >>> >>> Thank you >>> Swarandeep >>> >>> On Thu, Jun 1, 2017 at 7:38 PM, wrote: >>>> >>>> Hi Swarandeep, >>>> >>>> How large is your Rayleigh number and what is your resolution? >>>> What are your thermal bcs? What are your ICs? >>>> >>>> There should be no particular issue with your configuration - but you do need to >>>> control spatial and temporal resolution. >>>> >>>> >>>> For RB, you often have to start at low Ra and/or small dt because the nonlinear >>>> response can often drive the velocities to be quite high very rapidly before the >>>> CFL constraint kicks in. RB flows are one of the rare instances where I use >>>> variable time step size (param 12 > 0), but even then you have to keep the target >>>> CFL quite small until the flow is established. >>>> >>>> hth, >>>> >>>> Paul >>>> >>>> ________________________________________ >>>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>>> Sent: Thursday, June 01, 2017 7:27 AM >>>> To: nek5000-users at lists.mcs.anl.gov >>>> Subject: [Nek5000-users] Corner elements giving large values >>>> >>>> Dear Nek forum, >>>> >>>> I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. >>>> I am getting spurious large values of velocity at the corner regions >>>> of the annulus i.e. the edges where the inner cylinder joins the >>>> top and bottom >>>> planes. The boundary condition is no-slip and no penetration for velocity >>>> at these edges. Can someone point out a plausible reason. >>>> >>>> Thank you >>>> Swarandeep >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Sun Jun 4 14:42:39 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 5 Jun 2017 01:12:39 +0530 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: Message-ID: Hi Nek Forum, Paul Thanks for the suggestions. I implemented them. Waiting for the simulation to be over. Another issue I have is that the BC for T at the inner cylinder wall does not stay at 0 as I have given in userbc. It becomes 0.2 over some time. Thank you Swarandeep On Fri, Jun 2, 2017 at 10:28 PM, wrote: > > Hi Swarandeep, > > Set p26 in the .rea file to be 0.1 and make certain IFCHAR is F (a few lines below); > > Make certain your dt is positive (param 12) > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Friday, June 02, 2017 10:31 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > > Dear Nek Forum, Paul > > Yes, I have user defined grid such that the locations of the > elements are distributed to lie on the chebyshev grid at the top and > bottom and near the side walls. > Also please tell which param sets the CFL. > > Thank you > Swarandeep > > > On Fri, Jun 2, 2017 at 8:00 PM, wrote: >> >> Dear Swarandeep, >> >> Did you refine your element size at the top and bottom and near the side walls? >> >> Please set your CFL to 0.05 in order to get the flow started. >> >> Best, >> >> >> Paul >> >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Friday, June 02, 2017 1:04 AM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Corner elements giving large values >> >> Hi Nek Forum, Paul >> >> The top and bottom BC are zero heat flux. Please let me n=know >> how to control the CFL. It is 0.5 in the logfile. I am having no >> particular difficulty except these corner elements at the junction of >> inner cylinder and the top/bottom planes. Is there a better way to >> treat edge boundary elements. Also is it possible to have different >> order of elements at different locations. >> >> Thank you >> Swarandeep >> >> On Fri, Jun 2, 2017 at 12:50 AM, wrote: >>> >>> Hi Swarandeep, >>> >>> Sorry - I misunderstood. I guess my question was more about the >>> temperature bcs at the bottom and top of the domain? >>> >>> In general, however, there should be no difficulty. What happens >>> when you reduce your dt ? >>> >>> You need to keep the CFL < ~0.5 in general. CFL is printed in the logfile, >>> >>> grep tep logfile >>> >>> and you'll see it "C= ....." >>> >>> Paul >>> >>> >>> >>> >>> ________________________________________ >>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>> Sent: Thursday, June 01, 2017 1:57 PM >>> To: nek5000-users at lists.mcs.anl.gov >>> Subject: Re: [Nek5000-users] Corner elements giving large values >>> >>> Hi NekForum, Paul >>> >>> I intent to have isothermal conditions at the inner and outer >>> cylinders. The differential heating is achieved by imposing a >>> temperature difference across the two cylinders. I have used T = 1 and >>> 0 values for non-dimensional temperature. I think it should be fine. >>> Do you see any problem here. >>> >>> Thank you >>> Swarandeep >>> >>> On Thu, Jun 1, 2017 at 11:28 PM, wrote: >>>> >>>> Hi Swarandeep, >>>> >>>> Is 0/1 a realistic BC for T ? >>>> >>>> Best, >>>> >>>> Paul >>>> >>>> ________________________________________ >>>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>>> Sent: Thursday, June 01, 2017 12:53 PM >>>> To: nek5000-users at lists.mcs.anl.gov >>>> Subject: Re: [Nek5000-users] Corner elements giving large values >>>> >>>> Hi Nek Forum, Paul >>>> >>>> I am having temp = 1 and temp = 0 at the inner and outer >>>> boundaries. Gravity is pointing in the -ve z-direction along the axis >>>> of the >>>> annulus. The top and bottom planes have zero heat flux conditions. >>>> I have Coriolis force added as a body force. >>>> The Rayleigh number is around 10000. The resolution is 8, 25, >>>> 24 elements in the r, phi and z directions respectively with order 9. >>>> My initial condition is very low arbitrary velocity >>>> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for >>>> param12. The CFL is >>>> set to 0.5. Please indicate how to control it. >>>> >>>> As you have rightly mentioned, I start the simulation with a >>>> very low value of the Rayleigh number and gradually increase it. Yet, >>>> the corner values only get large. Other regions of the domain are >>>> fine. >>>> >>>> Thank you >>>> Swarandeep >>>> >>>> On Thu, Jun 1, 2017 at 7:38 PM, wrote: >>>>> >>>>> Hi Swarandeep, >>>>> >>>>> How large is your Rayleigh number and what is your resolution? >>>>> What are your thermal bcs? What are your ICs? >>>>> >>>>> There should be no particular issue with your configuration - but you do need to >>>>> control spatial and temporal resolution. >>>>> >>>>> >>>>> For RB, you often have to start at low Ra and/or small dt because the nonlinear >>>>> response can often drive the velocities to be quite high very rapidly before the >>>>> CFL constraint kicks in. RB flows are one of the rare instances where I use >>>>> variable time step size (param 12 > 0), but even then you have to keep the target >>>>> CFL quite small until the flow is established. >>>>> >>>>> hth, >>>>> >>>>> Paul >>>>> >>>>> ________________________________________ >>>>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>>>> Sent: Thursday, June 01, 2017 7:27 AM >>>>> To: nek5000-users at lists.mcs.anl.gov >>>>> Subject: [Nek5000-users] Corner elements giving large values >>>>> >>>>> Dear Nek forum, >>>>> >>>>> I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. >>>>> I am getting spurious large values of velocity at the corner regions >>>>> of the annulus i.e. the edges where the inner cylinder joins the >>>>> top and bottom >>>>> planes. The boundary condition is no-slip and no penetration for velocity >>>>> at these edges. Can someone point out a plausible reason. >>>>> >>>>> Thank you >>>>> Swarandeep >>>>> _______________________________________________ >>>>> Nek5000-users mailing list >>>>> Nek5000-users at lists.mcs.anl.gov >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>>> _______________________________________________ >>>>> Nek5000-users mailing list >>>>> Nek5000-users at lists.mcs.anl.gov >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Sun Jun 4 20:20:04 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 5 Jun 2017 01:20:04 +0000 Subject: [Nek5000-users] Corner elements giving large values In-Reply-To: References: , Message-ID: Swarandeep, if you send me your .rea/.usr/SIZE file off-list, I'll take a look. Paul ________________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] Sent: Sunday, June 04, 2017 2:42 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Corner elements giving large values Hi Nek Forum, Paul Thanks for the suggestions. I implemented them. Waiting for the simulation to be over. Another issue I have is that the BC for T at the inner cylinder wall does not stay at 0 as I have given in userbc. It becomes 0.2 over some time. Thank you Swarandeep On Fri, Jun 2, 2017 at 10:28 PM, wrote: > > Hi Swarandeep, > > Set p26 in the .rea file to be 0.1 and make certain IFCHAR is F (a few lines below); > > Make certain your dt is positive (param 12) > > Paul > > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Friday, June 02, 2017 10:31 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > > Dear Nek Forum, Paul > > Yes, I have user defined grid such that the locations of the > elements are distributed to lie on the chebyshev grid at the top and > bottom and near the side walls. > Also please tell which param sets the CFL. > > Thank you > Swarandeep > > > On Fri, Jun 2, 2017 at 8:00 PM, wrote: >> >> Dear Swarandeep, >> >> Did you refine your element size at the top and bottom and near the side walls? >> >> Please set your CFL to 0.05 in order to get the flow started. >> >> Best, >> >> >> Paul >> >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Friday, June 02, 2017 1:04 AM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Corner elements giving large values >> >> Hi Nek Forum, Paul >> >> The top and bottom BC are zero heat flux. Please let me n=know >> how to control the CFL. It is 0.5 in the logfile. I am having no >> particular difficulty except these corner elements at the junction of >> inner cylinder and the top/bottom planes. Is there a better way to >> treat edge boundary elements. Also is it possible to have different >> order of elements at different locations. >> >> Thank you >> Swarandeep >> >> On Fri, Jun 2, 2017 at 12:50 AM, wrote: >>> >>> Hi Swarandeep, >>> >>> Sorry - I misunderstood. I guess my question was more about the >>> temperature bcs at the bottom and top of the domain? >>> >>> In general, however, there should be no difficulty. What happens >>> when you reduce your dt ? >>> >>> You need to keep the CFL < ~0.5 in general. CFL is printed in the logfile, >>> >>> grep tep logfile >>> >>> and you'll see it "C= ....." >>> >>> Paul >>> >>> >>> >>> >>> ________________________________________ >>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>> Sent: Thursday, June 01, 2017 1:57 PM >>> To: nek5000-users at lists.mcs.anl.gov >>> Subject: Re: [Nek5000-users] Corner elements giving large values >>> >>> Hi NekForum, Paul >>> >>> I intent to have isothermal conditions at the inner and outer >>> cylinders. The differential heating is achieved by imposing a >>> temperature difference across the two cylinders. I have used T = 1 and >>> 0 values for non-dimensional temperature. I think it should be fine. >>> Do you see any problem here. >>> >>> Thank you >>> Swarandeep >>> >>> On Thu, Jun 1, 2017 at 11:28 PM, wrote: >>>> >>>> Hi Swarandeep, >>>> >>>> Is 0/1 a realistic BC for T ? >>>> >>>> Best, >>>> >>>> Paul >>>> >>>> ________________________________________ >>>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>>> Sent: Thursday, June 01, 2017 12:53 PM >>>> To: nek5000-users at lists.mcs.anl.gov >>>> Subject: Re: [Nek5000-users] Corner elements giving large values >>>> >>>> Hi Nek Forum, Paul >>>> >>>> I am having temp = 1 and temp = 0 at the inner and outer >>>> boundaries. Gravity is pointing in the -ve z-direction along the axis >>>> of the >>>> annulus. The top and bottom planes have zero heat flux conditions. >>>> I have Coriolis force added as a body force. >>>> The Rayleigh number is around 10000. The resolution is 8, 25, >>>> 24 elements in the r, phi and z directions respectively with order 9. >>>> My initial condition is very low arbitrary velocity >>>> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for >>>> param12. The CFL is >>>> set to 0.5. Please indicate how to control it. >>>> >>>> As you have rightly mentioned, I start the simulation with a >>>> very low value of the Rayleigh number and gradually increase it. Yet, >>>> the corner values only get large. Other regions of the domain are >>>> fine. >>>> >>>> Thank you >>>> Swarandeep >>>> >>>> On Thu, Jun 1, 2017 at 7:38 PM, wrote: >>>>> >>>>> Hi Swarandeep, >>>>> >>>>> How large is your Rayleigh number and what is your resolution? >>>>> What are your thermal bcs? What are your ICs? >>>>> >>>>> There should be no particular issue with your configuration - but you do need to >>>>> control spatial and temporal resolution. >>>>> >>>>> >>>>> For RB, you often have to start at low Ra and/or small dt because the nonlinear >>>>> response can often drive the velocities to be quite high very rapidly before the >>>>> CFL constraint kicks in. RB flows are one of the rare instances where I use >>>>> variable time step size (param 12 > 0), but even then you have to keep the target >>>>> CFL quite small until the flow is established. >>>>> >>>>> hth, >>>>> >>>>> Paul >>>>> >>>>> ________________________________________ >>>>> From: nek5000-users-bounces at lists.mcs.anl.gov [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >>>>> Sent: Thursday, June 01, 2017 7:27 AM >>>>> To: nek5000-users at lists.mcs.anl.gov >>>>> Subject: [Nek5000-users] Corner elements giving large values >>>>> >>>>> Dear Nek forum, >>>>> >>>>> I am simulating Rayleigh Benard flow in a 3D finite cylindrical annulus. >>>>> I am getting spurious large values of velocity at the corner regions >>>>> of the annulus i.e. the edges where the inner cylinder joins the >>>>> top and bottom >>>>> planes. The boundary condition is no-slip and no penetration for velocity >>>>> at these edges. Can someone point out a plausible reason. >>>>> >>>>> Thank you >>>>> Swarandeep >>>>> _______________________________________________ >>>>> Nek5000-users mailing list >>>>> Nek5000-users at lists.mcs.anl.gov >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>>> _______________________________________________ >>>>> Nek5000-users mailing list >>>>> Nek5000-users at lists.mcs.anl.gov >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>>> _______________________________________________ >>>> Nek5000-users mailing list >>>> Nek5000-users at lists.mcs.anl.gov >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Mon Jun 5 23:29:02 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 06 Jun 2017 07:29:02 +0300 Subject: [Nek5000-users] =?utf-8?q?Nek5000_documentation_details=3A_Pn-Pn_?= =?utf-8?q?pressure_solver?= Message-ID: Dear Neks, reading the documentation I got the impression that Pn-Pn solver (low Mach) first solves the pressure where the convective and viscous (!) terms are taken into account. After that using this p^{n+1} we solve for velocity field. It seems that the algorithm consists of only 2 steps (pressure + velocity). However, reading?the paper by Tomboulides, Lee, Orszag (1996) which is referenced inside the code, I see the projection algorithm where first the velocity is updated using the extrapolated convective term, then the Laplacian of pressure is calculated due to convection, after that the velocity is updated using convection and pressure gradient. The last step accounts for viscous term. I am a bit confused, could you please help me out here? Which method is used? PS. Another thing is the coupled Helmholtz solver in Pn-Pn. I see that in case of Pn-Pn-2 each velocity component is treated separately (segregated solver). However, this coupled thing slightly confuses me, why not treating it separately as in Pn-Pn-2? Could you please comment there as well? Thank you. Best regards, Vlad From nek5000-users at lists.mcs.anl.gov Mon Jun 5 12:26:01 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 5 Jun 2017 10:26:01 -0700 Subject: [Nek5000-users] Contents of Nek5000-users digest Corner elements giving large values Message-ID: Swarandeep, Another issue that can arise in RBC simulations is poor numerical conditioning due to the non-dimensionalization. The non-dimensional form of the equations used in the Nek5000 examples is based off the conduction scales which result in a very small time-step and large velocities at larger Ra. I don't think you're case is high enough for it to be a problem, but you might try using the "freefall" scales if you aren't already. I suggest consulting the Scheel, Emran and Schumacher paper for some details on running an RBC simulation with Nek5000. Scheel, Janet D., Mohammad S. Emran, and J?rg Schumacher. "Resolving the fine-scale structure in turbulent Rayleigh?B?nard convection." *New Journal of Physics* 15.11 (2013): 113063. Kind Regards, Phil S. On Mon, Jun 5, 2017 at 10:00 AM, wrote: > Send Nek5000-users mailing list submissions to > nek5000-users at lists.mcs.anl.gov > > To subscribe or unsubscribe via the World Wide Web, visit > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > or, via email, send a message with subject or body 'help' to > nek5000-users-request at lists.mcs.anl.gov > > You can reach the person managing the list at > nek5000-users-owner at lists.mcs.anl.gov > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Nek5000-users digest..." > > > Today's Topics: > > 1. Re: Corner elements giving large values > (nek5000-users at lists.mcs.anl.gov) > 2. Re: Corner elements giving large values > (nek5000-users at lists.mcs.anl.gov) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Mon, 5 Jun 2017 01:12:39 +0530 > From: nek5000-users at lists.mcs.anl.gov > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > Message-ID: > > Content-Type: text/plain; charset="UTF-8" > > Hi Nek Forum, Paul > > Thanks for the suggestions. I implemented them. Waiting for the > simulation to be over. > > Another issue I have is that the BC for T at the inner cylinder wall > does not stay at 0 as I have given in userbc. It becomes 0.2 over some > time. > > Thank you > Swarandeep > > > > > > On Fri, Jun 2, 2017 at 10:28 PM, wrote: > > > > Hi Swarandeep, > > > > Set p26 in the .rea file to be 0.1 and make certain IFCHAR is F (a few > lines below); > > > > Make certain your dt is positive (param 12) > > > > Paul > > > > ________________________________________ > > From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > > Sent: Friday, June 02, 2017 10:31 AM > > To: nek5000-users at lists.mcs.anl.gov > > Subject: Re: [Nek5000-users] Corner elements giving large values > > > > Dear Nek Forum, Paul > > > > Yes, I have user defined grid such that the locations of the > > elements are distributed to lie on the chebyshev grid at the top and > > bottom and near the side walls. > > Also please tell which param sets the CFL. > > > > Thank you > > Swarandeep > > > > > > On Fri, Jun 2, 2017 at 8:00 PM, > wrote: > >> > >> Dear Swarandeep, > >> > >> Did you refine your element size at the top and bottom and near the > side walls? > >> > >> Please set your CFL to 0.05 in order to get the flow started. > >> > >> Best, > >> > >> > >> Paul > >> > >> ________________________________________ > >> From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > >> Sent: Friday, June 02, 2017 1:04 AM > >> To: nek5000-users at lists.mcs.anl.gov > >> Subject: Re: [Nek5000-users] Corner elements giving large values > >> > >> Hi Nek Forum, Paul > >> > >> The top and bottom BC are zero heat flux. Please let me n=know > >> how to control the CFL. It is 0.5 in the logfile. I am having no > >> particular difficulty except these corner elements at the junction of > >> inner cylinder and the top/bottom planes. Is there a better way to > >> treat edge boundary elements. Also is it possible to have different > >> order of elements at different locations. > >> > >> Thank you > >> Swarandeep > >> > >> On Fri, Jun 2, 2017 at 12:50 AM, > wrote: > >>> > >>> Hi Swarandeep, > >>> > >>> Sorry - I misunderstood. I guess my question was more about the > >>> temperature bcs at the bottom and top of the domain? > >>> > >>> In general, however, there should be no difficulty. What happens > >>> when you reduce your dt ? > >>> > >>> You need to keep the CFL < ~0.5 in general. CFL is printed in the > logfile, > >>> > >>> grep tep logfile > >>> > >>> and you'll see it "C= ....." > >>> > >>> Paul > >>> > >>> > >>> > >>> > >>> ________________________________________ > >>> From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > >>> Sent: Thursday, June 01, 2017 1:57 PM > >>> To: nek5000-users at lists.mcs.anl.gov > >>> Subject: Re: [Nek5000-users] Corner elements giving large values > >>> > >>> Hi NekForum, Paul > >>> > >>> I intent to have isothermal conditions at the inner and outer > >>> cylinders. The differential heating is achieved by imposing a > >>> temperature difference across the two cylinders. I have used T = 1 and > >>> 0 values for non-dimensional temperature. I think it should be fine. > >>> Do you see any problem here. > >>> > >>> Thank you > >>> Swarandeep > >>> > >>> On Thu, Jun 1, 2017 at 11:28 PM, > wrote: > >>>> > >>>> Hi Swarandeep, > >>>> > >>>> Is 0/1 a realistic BC for T ? > >>>> > >>>> Best, > >>>> > >>>> Paul > >>>> > >>>> ________________________________________ > >>>> From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > >>>> Sent: Thursday, June 01, 2017 12:53 PM > >>>> To: nek5000-users at lists.mcs.anl.gov > >>>> Subject: Re: [Nek5000-users] Corner elements giving large values > >>>> > >>>> Hi Nek Forum, Paul > >>>> > >>>> I am having temp = 1 and temp = 0 at the inner and outer > >>>> boundaries. Gravity is pointing in the -ve z-direction along the axis > >>>> of the > >>>> annulus. The top and bottom planes have zero heat flux > conditions. > >>>> I have Coriolis force added as a body force. > >>>> The Rayleigh number is around 10000. The resolution is 8, 25, > >>>> 24 elements in the r, phi and z directions respectively with order 9. > >>>> My initial condition is very low arbitrary velocity > >>>> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for > >>>> param12. The CFL is > >>>> set to 0.5. Please indicate how to control it. > >>>> > >>>> As you have rightly mentioned, I start the simulation with a > >>>> very low value of the Rayleigh number and gradually increase it. Yet, > >>>> the corner values only get large. Other regions of the domain are > >>>> fine. > >>>> > >>>> Thank you > >>>> Swarandeep > >>>> > >>>> On Thu, Jun 1, 2017 at 7:38 PM, > wrote: > >>>>> > >>>>> Hi Swarandeep, > >>>>> > >>>>> How large is your Rayleigh number and what is your resolution? > >>>>> What are your thermal bcs? What are your ICs? > >>>>> > >>>>> There should be no particular issue with your configuration - but > you do need to > >>>>> control spatial and temporal resolution. > >>>>> > >>>>> > >>>>> For RB, you often have to start at low Ra and/or small dt because > the nonlinear > >>>>> response can often drive the velocities to be quite high very > rapidly before the > >>>>> CFL constraint kicks in. RB flows are one of the rare instances > where I use > >>>>> variable time step size (param 12 > 0), but even then you have to > keep the target > >>>>> CFL quite small until the flow is established. > >>>>> > >>>>> hth, > >>>>> > >>>>> Paul > >>>>> > >>>>> ________________________________________ > >>>>> From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > >>>>> Sent: Thursday, June 01, 2017 7:27 AM > >>>>> To: nek5000-users at lists.mcs.anl.gov > >>>>> Subject: [Nek5000-users] Corner elements giving large values > >>>>> > >>>>> Dear Nek forum, > >>>>> > >>>>> I am simulating Rayleigh Benard flow in a 3D finite > cylindrical annulus. > >>>>> I am getting spurious large values of velocity at the corner > regions > >>>>> of the annulus i.e. the edges where the inner cylinder joins > the > >>>>> top and bottom > >>>>> planes. The boundary condition is no-slip and no penetration > for velocity > >>>>> at these edges. Can someone point out a plausible reason. > >>>>> > >>>>> Thank you > >>>>> Swarandeep > >>>>> _______________________________________________ > >>>>> Nek5000-users mailing list > >>>>> Nek5000-users at lists.mcs.anl.gov > >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>>>> _______________________________________________ > >>>>> Nek5000-users mailing list > >>>>> Nek5000-users at lists.mcs.anl.gov > >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>>> _______________________________________________ > >>>> Nek5000-users mailing list > >>>> Nek5000-users at lists.mcs.anl.gov > >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>>> _______________________________________________ > >>>> Nek5000-users mailing list > >>>> Nek5000-users at lists.mcs.anl.gov > >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>> _______________________________________________ > >>> Nek5000-users mailing list > >>> Nek5000-users at lists.mcs.anl.gov > >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>> _______________________________________________ > >>> Nek5000-users mailing list > >>> Nek5000-users at lists.mcs.anl.gov > >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >> _______________________________________________ > >> Nek5000-users mailing list > >> Nek5000-users at lists.mcs.anl.gov > >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >> _______________________________________________ > >> Nek5000-users mailing list > >> Nek5000-users at lists.mcs.anl.gov > >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > _______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > _______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > ------------------------------ > > Message: 2 > Date: Mon, 5 Jun 2017 01:20:04 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > Subject: Re: [Nek5000-users] Corner elements giving large values > Message-ID: > > Content-Type: text/plain; charset="us-ascii" > > > Swarandeep, > > if you send me your .rea/.usr/SIZE file off-list, I'll take a look. > > Paul > ________________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > Sent: Sunday, June 04, 2017 2:42 PM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Corner elements giving large values > > Hi Nek Forum, Paul > > Thanks for the suggestions. I implemented them. Waiting for the > simulation to be over. > > Another issue I have is that the BC for T at the inner cylinder wall > does not stay at 0 as I have given in userbc. It becomes 0.2 over some > time. > > Thank you > Swarandeep > > > > > > On Fri, Jun 2, 2017 at 10:28 PM, wrote: > > > > Hi Swarandeep, > > > > Set p26 in the .rea file to be 0.1 and make certain IFCHAR is F (a few > lines below); > > > > Make certain your dt is positive (param 12) > > > > Paul > > > > ________________________________________ > > From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > > Sent: Friday, June 02, 2017 10:31 AM > > To: nek5000-users at lists.mcs.anl.gov > > Subject: Re: [Nek5000-users] Corner elements giving large values > > > > Dear Nek Forum, Paul > > > > Yes, I have user defined grid such that the locations of the > > elements are distributed to lie on the chebyshev grid at the top and > > bottom and near the side walls. > > Also please tell which param sets the CFL. > > > > Thank you > > Swarandeep > > > > > > On Fri, Jun 2, 2017 at 8:00 PM, > wrote: > >> > >> Dear Swarandeep, > >> > >> Did you refine your element size at the top and bottom and near the > side walls? > >> > >> Please set your CFL to 0.05 in order to get the flow started. > >> > >> Best, > >> > >> > >> Paul > >> > >> ________________________________________ > >> From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > >> Sent: Friday, June 02, 2017 1:04 AM > >> To: nek5000-users at lists.mcs.anl.gov > >> Subject: Re: [Nek5000-users] Corner elements giving large values > >> > >> Hi Nek Forum, Paul > >> > >> The top and bottom BC are zero heat flux. Please let me n=know > >> how to control the CFL. It is 0.5 in the logfile. I am having no > >> particular difficulty except these corner elements at the junction of > >> inner cylinder and the top/bottom planes. Is there a better way to > >> treat edge boundary elements. Also is it possible to have different > >> order of elements at different locations. > >> > >> Thank you > >> Swarandeep > >> > >> On Fri, Jun 2, 2017 at 12:50 AM, > wrote: > >>> > >>> Hi Swarandeep, > >>> > >>> Sorry - I misunderstood. I guess my question was more about the > >>> temperature bcs at the bottom and top of the domain? > >>> > >>> In general, however, there should be no difficulty. What happens > >>> when you reduce your dt ? > >>> > >>> You need to keep the CFL < ~0.5 in general. CFL is printed in the > logfile, > >>> > >>> grep tep logfile > >>> > >>> and you'll see it "C= ....." > >>> > >>> Paul > >>> > >>> > >>> > >>> > >>> ________________________________________ > >>> From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > >>> Sent: Thursday, June 01, 2017 1:57 PM > >>> To: nek5000-users at lists.mcs.anl.gov > >>> Subject: Re: [Nek5000-users] Corner elements giving large values > >>> > >>> Hi NekForum, Paul > >>> > >>> I intent to have isothermal conditions at the inner and outer > >>> cylinders. The differential heating is achieved by imposing a > >>> temperature difference across the two cylinders. I have used T = 1 and > >>> 0 values for non-dimensional temperature. I think it should be fine. > >>> Do you see any problem here. > >>> > >>> Thank you > >>> Swarandeep > >>> > >>> On Thu, Jun 1, 2017 at 11:28 PM, > wrote: > >>>> > >>>> Hi Swarandeep, > >>>> > >>>> Is 0/1 a realistic BC for T ? > >>>> > >>>> Best, > >>>> > >>>> Paul > >>>> > >>>> ________________________________________ > >>>> From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > >>>> Sent: Thursday, June 01, 2017 12:53 PM > >>>> To: nek5000-users at lists.mcs.anl.gov > >>>> Subject: Re: [Nek5000-users] Corner elements giving large values > >>>> > >>>> Hi Nek Forum, Paul > >>>> > >>>> I am having temp = 1 and temp = 0 at the inner and outer > >>>> boundaries. Gravity is pointing in the -ve z-direction along the axis > >>>> of the > >>>> annulus. The top and bottom planes have zero heat flux > conditions. > >>>> I have Coriolis force added as a body force. > >>>> The Rayleigh number is around 10000. The resolution is 8, 25, > >>>> 24 elements in the r, phi and z directions respectively with order 9. > >>>> My initial condition is very low arbitrary velocity > >>>> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for > >>>> param12. The CFL is > >>>> set to 0.5. Please indicate how to control it. > >>>> > >>>> As you have rightly mentioned, I start the simulation with a > >>>> very low value of the Rayleigh number and gradually increase it. Yet, > >>>> the corner values only get large. Other regions of the domain are > >>>> fine. > >>>> > >>>> Thank you > >>>> Swarandeep > >>>> > >>>> On Thu, Jun 1, 2017 at 7:38 PM, > wrote: > >>>>> > >>>>> Hi Swarandeep, > >>>>> > >>>>> How large is your Rayleigh number and what is your resolution? > >>>>> What are your thermal bcs? What are your ICs? > >>>>> > >>>>> There should be no particular issue with your configuration - but > you do need to > >>>>> control spatial and temporal resolution. > >>>>> > >>>>> > >>>>> For RB, you often have to start at low Ra and/or small dt because > the nonlinear > >>>>> response can often drive the velocities to be quite high very > rapidly before the > >>>>> CFL constraint kicks in. RB flows are one of the rare instances > where I use > >>>>> variable time step size (param 12 > 0), but even then you have to > keep the target > >>>>> CFL quite small until the flow is established. > >>>>> > >>>>> hth, > >>>>> > >>>>> Paul > >>>>> > >>>>> ________________________________________ > >>>>> From: nek5000-users-bounces at lists.mcs.anl.gov [ > nek5000-users-bounces at lists.mcs.anl.gov] on behalf of > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] > >>>>> Sent: Thursday, June 01, 2017 7:27 AM > >>>>> To: nek5000-users at lists.mcs.anl.gov > >>>>> Subject: [Nek5000-users] Corner elements giving large values > >>>>> > >>>>> Dear Nek forum, > >>>>> > >>>>> I am simulating Rayleigh Benard flow in a 3D finite > cylindrical annulus. > >>>>> I am getting spurious large values of velocity at the corner > regions > >>>>> of the annulus i.e. the edges where the inner cylinder joins > the > >>>>> top and bottom > >>>>> planes. The boundary condition is no-slip and no penetration > for velocity > >>>>> at these edges. Can someone point out a plausible reason. > >>>>> > >>>>> Thank you > >>>>> Swarandeep > >>>>> _______________________________________________ > >>>>> Nek5000-users mailing list > >>>>> Nek5000-users at lists.mcs.anl.gov > >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>>>> _______________________________________________ > >>>>> Nek5000-users mailing list > >>>>> Nek5000-users at lists.mcs.anl.gov > >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>>> _______________________________________________ > >>>> Nek5000-users mailing list > >>>> Nek5000-users at lists.mcs.anl.gov > >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>>> _______________________________________________ > >>>> Nek5000-users mailing list > >>>> Nek5000-users at lists.mcs.anl.gov > >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>> _______________________________________________ > >>> Nek5000-users mailing list > >>> Nek5000-users at lists.mcs.anl.gov > >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >>> _______________________________________________ > >>> Nek5000-users mailing list > >>> Nek5000-users at lists.mcs.anl.gov > >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >> _______________________________________________ > >> Nek5000-users mailing list > >> Nek5000-users at lists.mcs.anl.gov > >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >> _______________________________________________ > >> Nek5000-users mailing list > >> Nek5000-users at lists.mcs.anl.gov > >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > _______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > _______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > ------------------------------ > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > End of Nek5000-users Digest, Vol 100, Issue 4 > ********************************************* > -- Phil Sakievich Post Doctoral Research Associate - Mechanical and Aerospace Engineering Arizona State University - Ira A. Fulton School for Engineering of Matter Transport and Energy Tempe, Arizona -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Jun 6 04:19:41 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 6 Jun 2017 12:19:41 +0300 Subject: [Nek5000-users] Nek5000 documentation details: Pn-Pn pressure solver In-Reply-To: References: Message-ID: Dear Vlad, the low Mach Pn-Pn approach is based on the 1997 (JSC) and 1997 (JCP) papers you mention and it consists of 3 steps as you describe, i.e.: a) first the velocity is updated using the extrapolated convective term, b) then the Laplacian of pressure is calculated due to convection, after that c) the velocity is updated using the pressure gradient and accounts for viscous term The coupled Helmholtz solver is used for the velocities only when using ifstrs=true, that is when you want to include the full stress tensor. Otherwise, it is using separate Helmholtz solves for each of the velocity components, similar to Pn-Pn-2. Hope this helps clarify things. All the best, Ananias On Tue, Jun 6, 2017 at 7:29 AM, wrote: > Dear Neks, > > reading the documentation I got the impression that Pn-Pn solver (low > Mach) first solves the pressure where the convective and viscous (!) terms > are taken into account. After that using this p^{n+1} we solve for velocity > field. It seems that the algorithm consists of only 2 steps (pressure + > velocity). > > However, reading the paper by Tomboulides, Lee, Orszag (1996) which is > referenced inside the code, I see the projection algorithm where first the > velocity is updated using the extrapolated convective term, then the > Laplacian of pressure is calculated due to convection, after that the > velocity is updated using convection and pressure gradient. The last step > accounts for viscous term. > > I am a bit confused, could you please help me out here? Which method is > used? > > PS. Another thing is the coupled Helmholtz solver in Pn-Pn. I see that in > case of Pn-Pn-2 each velocity component is treated separately (segregated > solver). However, this coupled thing slightly confuses me, why not treating > it separately as in Pn-Pn-2? Could you please comment there as well? Thank > you. > > Best regards, > Vlad > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Jun 6 07:36:54 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 6 Jun 2017 18:06:54 +0530 Subject: [Nek5000-users] Contents of Nek5000-users digest Corner elements giving large values In-Reply-To: References: Message-ID: Hi Phil, Thanks for the suggestion. I will certainly look into it. Swarandeep On Mon, Jun 5, 2017 at 10:56 PM, wrote: > Swarandeep, > > Another issue that can arise in RBC simulations is poor numerical > conditioning due to the non-dimensionalization. The non-dimensional form of > the equations used in the Nek5000 examples is based off the conduction > scales which result in a very small time-step and large velocities at larger > Ra. I don't think you're case is high enough for it to be a problem, but > you might try using the "freefall" scales if you aren't already. > > I suggest consulting the Scheel, Emran and Schumacher paper for some details > on running an RBC simulation with Nek5000. > > Scheel, Janet D., Mohammad S. Emran, and J?rg Schumacher. "Resolving the > fine-scale structure in turbulent Rayleigh?B?nard convection." New Journal > of Physics 15.11 (2013): 113063. > > Kind Regards, > > Phil S. > > On Mon, Jun 5, 2017 at 10:00 AM, > wrote: >> >> Send Nek5000-users mailing list submissions to >> nek5000-users at lists.mcs.anl.gov >> >> To subscribe or unsubscribe via the World Wide Web, visit >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> or, via email, send a message with subject or body 'help' to >> nek5000-users-request at lists.mcs.anl.gov >> >> You can reach the person managing the list at >> nek5000-users-owner at lists.mcs.anl.gov >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of Nek5000-users digest..." >> >> >> Today's Topics: >> >> 1. Re: Corner elements giving large values >> (nek5000-users at lists.mcs.anl.gov) >> 2. Re: Corner elements giving large values >> (nek5000-users at lists.mcs.anl.gov) >> >> >> ---------------------------------------------------------------------- >> >> Message: 1 >> Date: Mon, 5 Jun 2017 01:12:39 +0530 >> From: nek5000-users at lists.mcs.anl.gov >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Corner elements giving large values >> Message-ID: >> >> Content-Type: text/plain; charset="UTF-8" >> >> Hi Nek Forum, Paul >> >> Thanks for the suggestions. I implemented them. Waiting for the >> simulation to be over. >> >> Another issue I have is that the BC for T at the inner cylinder wall >> does not stay at 0 as I have given in userbc. It becomes 0.2 over some >> time. >> >> Thank you >> Swarandeep >> >> >> >> >> >> On Fri, Jun 2, 2017 at 10:28 PM, wrote: >> > >> > Hi Swarandeep, >> > >> > Set p26 in the .rea file to be 0.1 and make certain IFCHAR is F (a few >> > lines below); >> > >> > Make certain your dt is positive (param 12) >> > >> > Paul >> > >> > ________________________________________ >> > From: nek5000-users-bounces at lists.mcs.anl.gov >> > [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> > Sent: Friday, June 02, 2017 10:31 AM >> > To: nek5000-users at lists.mcs.anl.gov >> > Subject: Re: [Nek5000-users] Corner elements giving large values >> > >> > Dear Nek Forum, Paul >> > >> > Yes, I have user defined grid such that the locations of the >> > elements are distributed to lie on the chebyshev grid at the top and >> > bottom and near the side walls. >> > Also please tell which param sets the CFL. >> > >> > Thank you >> > Swarandeep >> > >> > >> > On Fri, Jun 2, 2017 at 8:00 PM, >> > wrote: >> >> >> >> Dear Swarandeep, >> >> >> >> Did you refine your element size at the top and bottom and near the >> >> side walls? >> >> >> >> Please set your CFL to 0.05 in order to get the flow started. >> >> >> >> Best, >> >> >> >> >> >> Paul >> >> >> >> ________________________________________ >> >> From: nek5000-users-bounces at lists.mcs.anl.gov >> >> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> >> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> >> Sent: Friday, June 02, 2017 1:04 AM >> >> To: nek5000-users at lists.mcs.anl.gov >> >> Subject: Re: [Nek5000-users] Corner elements giving large values >> >> >> >> Hi Nek Forum, Paul >> >> >> >> The top and bottom BC are zero heat flux. Please let me n=know >> >> how to control the CFL. It is 0.5 in the logfile. I am having no >> >> particular difficulty except these corner elements at the junction of >> >> inner cylinder and the top/bottom planes. Is there a better way to >> >> treat edge boundary elements. Also is it possible to have different >> >> order of elements at different locations. >> >> >> >> Thank you >> >> Swarandeep >> >> >> >> On Fri, Jun 2, 2017 at 12:50 AM, >> >> wrote: >> >>> >> >>> Hi Swarandeep, >> >>> >> >>> Sorry - I misunderstood. I guess my question was more about the >> >>> temperature bcs at the bottom and top of the domain? >> >>> >> >>> In general, however, there should be no difficulty. What happens >> >>> when you reduce your dt ? >> >>> >> >>> You need to keep the CFL < ~0.5 in general. CFL is printed in the >> >>> logfile, >> >>> >> >>> grep tep logfile >> >>> >> >>> and you'll see it "C= ....." >> >>> >> >>> Paul >> >>> >> >>> >> >>> >> >>> >> >>> ________________________________________ >> >>> From: nek5000-users-bounces at lists.mcs.anl.gov >> >>> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> >>> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> >>> Sent: Thursday, June 01, 2017 1:57 PM >> >>> To: nek5000-users at lists.mcs.anl.gov >> >>> Subject: Re: [Nek5000-users] Corner elements giving large values >> >>> >> >>> Hi NekForum, Paul >> >>> >> >>> I intent to have isothermal conditions at the inner and outer >> >>> cylinders. The differential heating is achieved by imposing a >> >>> temperature difference across the two cylinders. I have used T = 1 and >> >>> 0 values for non-dimensional temperature. I think it should be fine. >> >>> Do you see any problem here. >> >>> >> >>> Thank you >> >>> Swarandeep >> >>> >> >>> On Thu, Jun 1, 2017 at 11:28 PM, >> >>> wrote: >> >>>> >> >>>> Hi Swarandeep, >> >>>> >> >>>> Is 0/1 a realistic BC for T ? >> >>>> >> >>>> Best, >> >>>> >> >>>> Paul >> >>>> >> >>>> ________________________________________ >> >>>> From: nek5000-users-bounces at lists.mcs.anl.gov >> >>>> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> >>>> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> >>>> Sent: Thursday, June 01, 2017 12:53 PM >> >>>> To: nek5000-users at lists.mcs.anl.gov >> >>>> Subject: Re: [Nek5000-users] Corner elements giving large values >> >>>> >> >>>> Hi Nek Forum, Paul >> >>>> >> >>>> I am having temp = 1 and temp = 0 at the inner and outer >> >>>> boundaries. Gravity is pointing in the -ve z-direction along the axis >> >>>> of the >> >>>> annulus. The top and bottom planes have zero heat flux >> >>>> conditions. >> >>>> I have Coriolis force added as a body force. >> >>>> The Rayleigh number is around 10000. The resolution is 8, 25, >> >>>> 24 elements in the r, phi and z directions respectively with order 9. >> >>>> My initial condition is very low arbitrary velocity >> >>>> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for >> >>>> param12. The CFL is >> >>>> set to 0.5. Please indicate how to control it. >> >>>> >> >>>> As you have rightly mentioned, I start the simulation with a >> >>>> very low value of the Rayleigh number and gradually increase it. Yet, >> >>>> the corner values only get large. Other regions of the domain are >> >>>> fine. >> >>>> >> >>>> Thank you >> >>>> Swarandeep >> >>>> >> >>>> On Thu, Jun 1, 2017 at 7:38 PM, >> >>>> wrote: >> >>>>> >> >>>>> Hi Swarandeep, >> >>>>> >> >>>>> How large is your Rayleigh number and what is your resolution? >> >>>>> What are your thermal bcs? What are your ICs? >> >>>>> >> >>>>> There should be no particular issue with your configuration - but >> >>>>> you do need to >> >>>>> control spatial and temporal resolution. >> >>>>> >> >>>>> >> >>>>> For RB, you often have to start at low Ra and/or small dt because >> >>>>> the nonlinear >> >>>>> response can often drive the velocities to be quite high very >> >>>>> rapidly before the >> >>>>> CFL constraint kicks in. RB flows are one of the rare instances >> >>>>> where I use >> >>>>> variable time step size (param 12 > 0), but even then you have to >> >>>>> keep the target >> >>>>> CFL quite small until the flow is established. >> >>>>> >> >>>>> hth, >> >>>>> >> >>>>> Paul >> >>>>> >> >>>>> ________________________________________ >> >>>>> From: nek5000-users-bounces at lists.mcs.anl.gov >> >>>>> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> >>>>> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> >>>>> Sent: Thursday, June 01, 2017 7:27 AM >> >>>>> To: nek5000-users at lists.mcs.anl.gov >> >>>>> Subject: [Nek5000-users] Corner elements giving large values >> >>>>> >> >>>>> Dear Nek forum, >> >>>>> >> >>>>> I am simulating Rayleigh Benard flow in a 3D finite >> >>>>> cylindrical annulus. >> >>>>> I am getting spurious large values of velocity at the corner >> >>>>> regions >> >>>>> of the annulus i.e. the edges where the inner cylinder joins >> >>>>> the >> >>>>> top and bottom >> >>>>> planes. The boundary condition is no-slip and no penetration >> >>>>> for velocity >> >>>>> at these edges. Can someone point out a plausible reason. >> >>>>> >> >>>>> Thank you >> >>>>> Swarandeep >> >>>>> _______________________________________________ >> >>>>> Nek5000-users mailing list >> >>>>> Nek5000-users at lists.mcs.anl.gov >> >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>>>> _______________________________________________ >> >>>>> Nek5000-users mailing list >> >>>>> Nek5000-users at lists.mcs.anl.gov >> >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>>> _______________________________________________ >> >>>> Nek5000-users mailing list >> >>>> Nek5000-users at lists.mcs.anl.gov >> >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>>> _______________________________________________ >> >>>> Nek5000-users mailing list >> >>>> Nek5000-users at lists.mcs.anl.gov >> >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>> _______________________________________________ >> >>> Nek5000-users mailing list >> >>> Nek5000-users at lists.mcs.anl.gov >> >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>> _______________________________________________ >> >>> Nek5000-users mailing list >> >>> Nek5000-users at lists.mcs.anl.gov >> >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> _______________________________________________ >> >> Nek5000-users mailing list >> >> Nek5000-users at lists.mcs.anl.gov >> >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> _______________________________________________ >> >> Nek5000-users mailing list >> >> Nek5000-users at lists.mcs.anl.gov >> >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> > _______________________________________________ >> > Nek5000-users mailing list >> > Nek5000-users at lists.mcs.anl.gov >> > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> > _______________________________________________ >> > Nek5000-users mailing list >> > Nek5000-users at lists.mcs.anl.gov >> > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> ------------------------------ >> >> Message: 2 >> Date: Mon, 5 Jun 2017 01:20:04 +0000 >> From: nek5000-users at lists.mcs.anl.gov >> To: "nek5000-users at lists.mcs.anl.gov" >> >> Subject: Re: [Nek5000-users] Corner elements giving large values >> Message-ID: >> >> Content-Type: text/plain; charset="us-ascii" >> >> >> Swarandeep, >> >> if you send me your .rea/.usr/SIZE file off-list, I'll take a look. >> >> Paul >> ________________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov >> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> Sent: Sunday, June 04, 2017 2:42 PM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Corner elements giving large values >> >> Hi Nek Forum, Paul >> >> Thanks for the suggestions. I implemented them. Waiting for the >> simulation to be over. >> >> Another issue I have is that the BC for T at the inner cylinder wall >> does not stay at 0 as I have given in userbc. It becomes 0.2 over some >> time. >> >> Thank you >> Swarandeep >> >> >> >> >> >> On Fri, Jun 2, 2017 at 10:28 PM, wrote: >> > >> > Hi Swarandeep, >> > >> > Set p26 in the .rea file to be 0.1 and make certain IFCHAR is F (a few >> > lines below); >> > >> > Make certain your dt is positive (param 12) >> > >> > Paul >> > >> > ________________________________________ >> > From: nek5000-users-bounces at lists.mcs.anl.gov >> > [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> > nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> > Sent: Friday, June 02, 2017 10:31 AM >> > To: nek5000-users at lists.mcs.anl.gov >> > Subject: Re: [Nek5000-users] Corner elements giving large values >> > >> > Dear Nek Forum, Paul >> > >> > Yes, I have user defined grid such that the locations of the >> > elements are distributed to lie on the chebyshev grid at the top and >> > bottom and near the side walls. >> > Also please tell which param sets the CFL. >> > >> > Thank you >> > Swarandeep >> > >> > >> > On Fri, Jun 2, 2017 at 8:00 PM, >> > wrote: >> >> >> >> Dear Swarandeep, >> >> >> >> Did you refine your element size at the top and bottom and near the >> >> side walls? >> >> >> >> Please set your CFL to 0.05 in order to get the flow started. >> >> >> >> Best, >> >> >> >> >> >> Paul >> >> >> >> ________________________________________ >> >> From: nek5000-users-bounces at lists.mcs.anl.gov >> >> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> >> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> >> Sent: Friday, June 02, 2017 1:04 AM >> >> To: nek5000-users at lists.mcs.anl.gov >> >> Subject: Re: [Nek5000-users] Corner elements giving large values >> >> >> >> Hi Nek Forum, Paul >> >> >> >> The top and bottom BC are zero heat flux. Please let me n=know >> >> how to control the CFL. It is 0.5 in the logfile. I am having no >> >> particular difficulty except these corner elements at the junction of >> >> inner cylinder and the top/bottom planes. Is there a better way to >> >> treat edge boundary elements. Also is it possible to have different >> >> order of elements at different locations. >> >> >> >> Thank you >> >> Swarandeep >> >> >> >> On Fri, Jun 2, 2017 at 12:50 AM, >> >> wrote: >> >>> >> >>> Hi Swarandeep, >> >>> >> >>> Sorry - I misunderstood. I guess my question was more about the >> >>> temperature bcs at the bottom and top of the domain? >> >>> >> >>> In general, however, there should be no difficulty. What happens >> >>> when you reduce your dt ? >> >>> >> >>> You need to keep the CFL < ~0.5 in general. CFL is printed in the >> >>> logfile, >> >>> >> >>> grep tep logfile >> >>> >> >>> and you'll see it "C= ....." >> >>> >> >>> Paul >> >>> >> >>> >> >>> >> >>> >> >>> ________________________________________ >> >>> From: nek5000-users-bounces at lists.mcs.anl.gov >> >>> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> >>> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> >>> Sent: Thursday, June 01, 2017 1:57 PM >> >>> To: nek5000-users at lists.mcs.anl.gov >> >>> Subject: Re: [Nek5000-users] Corner elements giving large values >> >>> >> >>> Hi NekForum, Paul >> >>> >> >>> I intent to have isothermal conditions at the inner and outer >> >>> cylinders. The differential heating is achieved by imposing a >> >>> temperature difference across the two cylinders. I have used T = 1 and >> >>> 0 values for non-dimensional temperature. I think it should be fine. >> >>> Do you see any problem here. >> >>> >> >>> Thank you >> >>> Swarandeep >> >>> >> >>> On Thu, Jun 1, 2017 at 11:28 PM, >> >>> wrote: >> >>>> >> >>>> Hi Swarandeep, >> >>>> >> >>>> Is 0/1 a realistic BC for T ? >> >>>> >> >>>> Best, >> >>>> >> >>>> Paul >> >>>> >> >>>> ________________________________________ >> >>>> From: nek5000-users-bounces at lists.mcs.anl.gov >> >>>> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> >>>> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> >>>> Sent: Thursday, June 01, 2017 12:53 PM >> >>>> To: nek5000-users at lists.mcs.anl.gov >> >>>> Subject: Re: [Nek5000-users] Corner elements giving large values >> >>>> >> >>>> Hi Nek Forum, Paul >> >>>> >> >>>> I am having temp = 1 and temp = 0 at the inner and outer >> >>>> boundaries. Gravity is pointing in the -ve z-direction along the axis >> >>>> of the >> >>>> annulus. The top and bottom planes have zero heat flux >> >>>> conditions. >> >>>> I have Coriolis force added as a body force. >> >>>> The Rayleigh number is around 10000. The resolution is 8, 25, >> >>>> 24 elements in the r, phi and z directions respectively with order 9. >> >>>> My initial condition is very low arbitrary velocity >> >>>> perturbation (~10^(-8)). The time resolution is set to 10^(-5) for >> >>>> param12. The CFL is >> >>>> set to 0.5. Please indicate how to control it. >> >>>> >> >>>> As you have rightly mentioned, I start the simulation with a >> >>>> very low value of the Rayleigh number and gradually increase it. Yet, >> >>>> the corner values only get large. Other regions of the domain are >> >>>> fine. >> >>>> >> >>>> Thank you >> >>>> Swarandeep >> >>>> >> >>>> On Thu, Jun 1, 2017 at 7:38 PM, >> >>>> wrote: >> >>>>> >> >>>>> Hi Swarandeep, >> >>>>> >> >>>>> How large is your Rayleigh number and what is your resolution? >> >>>>> What are your thermal bcs? What are your ICs? >> >>>>> >> >>>>> There should be no particular issue with your configuration - but >> >>>>> you do need to >> >>>>> control spatial and temporal resolution. >> >>>>> >> >>>>> >> >>>>> For RB, you often have to start at low Ra and/or small dt because >> >>>>> the nonlinear >> >>>>> response can often drive the velocities to be quite high very >> >>>>> rapidly before the >> >>>>> CFL constraint kicks in. RB flows are one of the rare instances >> >>>>> where I use >> >>>>> variable time step size (param 12 > 0), but even then you have to >> >>>>> keep the target >> >>>>> CFL quite small until the flow is established. >> >>>>> >> >>>>> hth, >> >>>>> >> >>>>> Paul >> >>>>> >> >>>>> ________________________________________ >> >>>>> From: nek5000-users-bounces at lists.mcs.anl.gov >> >>>>> [nek5000-users-bounces at lists.mcs.anl.gov] on behalf of >> >>>>> nek5000-users at lists.mcs.anl.gov [nek5000-users at lists.mcs.anl.gov] >> >>>>> Sent: Thursday, June 01, 2017 7:27 AM >> >>>>> To: nek5000-users at lists.mcs.anl.gov >> >>>>> Subject: [Nek5000-users] Corner elements giving large values >> >>>>> >> >>>>> Dear Nek forum, >> >>>>> >> >>>>> I am simulating Rayleigh Benard flow in a 3D finite >> >>>>> cylindrical annulus. >> >>>>> I am getting spurious large values of velocity at the corner >> >>>>> regions >> >>>>> of the annulus i.e. the edges where the inner cylinder joins >> >>>>> the >> >>>>> top and bottom >> >>>>> planes. The boundary condition is no-slip and no penetration >> >>>>> for velocity >> >>>>> at these edges. Can someone point out a plausible reason. >> >>>>> >> >>>>> Thank you >> >>>>> Swarandeep >> >>>>> _______________________________________________ >> >>>>> Nek5000-users mailing list >> >>>>> Nek5000-users at lists.mcs.anl.gov >> >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>>>> _______________________________________________ >> >>>>> Nek5000-users mailing list >> >>>>> Nek5000-users at lists.mcs.anl.gov >> >>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>>> _______________________________________________ >> >>>> Nek5000-users mailing list >> >>>> Nek5000-users at lists.mcs.anl.gov >> >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>>> _______________________________________________ >> >>>> Nek5000-users mailing list >> >>>> Nek5000-users at lists.mcs.anl.gov >> >>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>> _______________________________________________ >> >>> Nek5000-users mailing list >> >>> Nek5000-users at lists.mcs.anl.gov >> >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >>> _______________________________________________ >> >>> Nek5000-users mailing list >> >>> Nek5000-users at lists.mcs.anl.gov >> >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> _______________________________________________ >> >> Nek5000-users mailing list >> >> Nek5000-users at lists.mcs.anl.gov >> >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> _______________________________________________ >> >> Nek5000-users mailing list >> >> Nek5000-users at lists.mcs.anl.gov >> >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> > _______________________________________________ >> > Nek5000-users mailing list >> > Nek5000-users at lists.mcs.anl.gov >> > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> > _______________________________________________ >> > Nek5000-users mailing list >> > Nek5000-users at lists.mcs.anl.gov >> > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> ------------------------------ >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> End of Nek5000-users Digest, Vol 100, Issue 4 >> ********************************************* > > > > > -- > Phil Sakievich > > Post Doctoral Research Associate - Mechanical and Aerospace Engineering > Arizona State University - Ira A. Fulton School for Engineering of Matter > Transport and Energy > Tempe, Arizona > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > From nek5000-users at lists.mcs.anl.gov Tue Jun 6 08:27:40 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 6 Jun 2017 13:27:40 +0000 Subject: [Nek5000-users] An abrupt error of "unconverged Helmh3 fluid" Message-ID: Dear Neks, I am using the moving mesh to simulate surface gravity waves with some random heat flux at the surface. The codes ran well for a while, but it suddenly pop out an error of unconverged Helmh3 fluid, as shown in the below log file (Step 14650). Could anyone give me a hint on this kind of error? Thank you, Peng Step 14647, t= 7.3235000E+01, DT= 5.0000000E-03, C= 0.036 2.8716E+04 1.9695E+00 Solving for Hmholtz scalars 14647 Hmholtz TEMP 4 6.0654E-16 1.4699E-01 1.0000E-12 14647 Scalars done 7.3235E+01 3.9568E-02 Solving for fluid 14647 Helmh3 fluid 4 1.9934E-13 8.0532E-04 1.0000E-12 14647 U-PRES gmres 14 4.5296E-11 2.2779E-03 1.0000E-10 8.2958E-02 2.5954E-01 14647 Fluid done 7.3235E+01 3.7241E-01 Step 14648, t= 7.3240000E+01, DT= 5.0000000E-03, C= 0.035 2.8718E+04 1.9642E+00 Solving for Hmholtz scalars 14648 Hmholtz TEMP 4 6.0987E-16 1.4701E-01 1.0000E-12 14648 Scalars done 7.3240E+01 3.9551E-02 Solving for fluid 14648 Helmh3 fluid 4 1.9904E-13 8.0535E-04 1.0000E-12 14648 U-PRES gmres 14 4.5647E-11 2.2779E-03 1.0000E-10 8.3140E-02 2.5985E-01 14648 Fluid done 7.3240E+01 3.7297E-01 Step 14649, t= 7.3245000E+01, DT= 5.0000000E-03, C= 0.034 2.8720E+04 1.9658E+00 Solving for Hmholtz scalars 14649 Hmholtz TEMP 4 6.1309E-16 1.4704E-01 1.0000E-12 14649 Scalars done 7.3245E+01 3.9058E-02 Solving for fluid 14649 Helmh3 fluid 4 1.9875E-13 8.0535E-04 1.0000E-12 14649 U-PRES gmres 14 4.5884E-11 2.2778E-03 1.0000E-10 8.3552E-02 2.5956E-01 14649 Fluid done 7.3245E+01 3.7226E-01 Step 14650, t= 7.3250000E+01, DT= 5.0000000E-03, C= 0.033 2.8722E+04 1.9614E+00 Solving for Hmholtz scalars 14650 Hmholtz TEMP 4 6.1615E-16 1.4707E-01 1.0000E-12 14650 Scalars done 7.3250E+01 3.9481E-02 Solving for fluid 14650 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 14650 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0555E-01 2.3985E+00 14650 Fluid done 7.3250E+01 5.4078E+00 Step 14651, t= 7.3255000E+01, DT= 5.0000000E-03, C= NaN 2.8729E+04 7.0068E+00 Solving for Hmholtz scalars 14651 Error Hmholtz TEMP 200 NaN NaN 1.0000E-12 14651 Scalars done 7.3255E+01 1.0293E+00 Solving for fluid 14651 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 14651 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0602E-01 2.4098E+00 14651 Fluid done 7.3255E+01 5.3955E+00 Step 14652, t= 7.3260000E+01, DT= 5.0000000E-03, C= NaN 2.8738E+04 9.1272E+00 Solving for Hmholtz scalars 14652 Error Hmholtz TEMP 200 NaN NaN 1.0000E-12 14652 Scalars done 7.3260E+01 1.0415E+00 Solving for fluid 14652 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 14652 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0562E-01 2.3933E+00 14652 Fluid done 7.3260E+01 5.3900E+00 From nek5000-users at lists.mcs.anl.gov Tue Jun 6 08:33:50 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 6 Jun 2017 13:33:50 +0000 Subject: [Nek5000-users] An abrupt error of "unconverged Helmh3 fluid" In-Reply-To: References: Message-ID: Pend, can you please give more details on the case you are trying to run with Nek5000? How did you set up your case? Did you start from one of the examples? Thanks. > On Jun 6, 2017, at 9:27 AM, nek5000-users at lists.mcs.anl.gov wrote: > > Dear Neks, > > I am using the moving mesh to simulate surface gravity waves with some random heat flux at the surface. > The codes ran well for a while, but it suddenly pop out an error of unconverged Helmh3 fluid, as shown in the below log file (Step 14650). > > Could anyone give me a hint on this kind of error? > > Thank you, > Peng > > > Step 14647, t= 7.3235000E+01, DT= 5.0000000E-03, C= 0.036 2.8716E+04 1.9695E+00 > Solving for Hmholtz scalars > 14647 Hmholtz TEMP 4 6.0654E-16 1.4699E-01 1.0000E-12 > 14647 Scalars done 7.3235E+01 3.9568E-02 > Solving for fluid > 14647 Helmh3 fluid 4 1.9934E-13 8.0532E-04 1.0000E-12 > 14647 U-PRES gmres 14 4.5296E-11 2.2779E-03 1.0000E-10 8.2958E-02 2.5954E-01 > 14647 Fluid done 7.3235E+01 3.7241E-01 > Step 14648, t= 7.3240000E+01, DT= 5.0000000E-03, C= 0.035 2.8718E+04 1.9642E+00 > Solving for Hmholtz scalars > 14648 Hmholtz TEMP 4 6.0987E-16 1.4701E-01 1.0000E-12 > 14648 Scalars done 7.3240E+01 3.9551E-02 > Solving for fluid > 14648 Helmh3 fluid 4 1.9904E-13 8.0535E-04 1.0000E-12 > 14648 U-PRES gmres 14 4.5647E-11 2.2779E-03 1.0000E-10 8.3140E-02 2.5985E-01 > 14648 Fluid done 7.3240E+01 3.7297E-01 > Step 14649, t= 7.3245000E+01, DT= 5.0000000E-03, C= 0.034 2.8720E+04 1.9658E+00 > Solving for Hmholtz scalars > 14649 Hmholtz TEMP 4 6.1309E-16 1.4704E-01 1.0000E-12 > 14649 Scalars done 7.3245E+01 3.9058E-02 > Solving for fluid > 14649 Helmh3 fluid 4 1.9875E-13 8.0535E-04 1.0000E-12 > 14649 U-PRES gmres 14 4.5884E-11 2.2778E-03 1.0000E-10 8.3552E-02 2.5956E-01 > 14649 Fluid done 7.3245E+01 3.7226E-01 > Step 14650, t= 7.3250000E+01, DT= 5.0000000E-03, C= 0.033 2.8722E+04 1.9614E+00 > Solving for Hmholtz scalars > 14650 Hmholtz TEMP 4 6.1615E-16 1.4707E-01 1.0000E-12 > 14650 Scalars done 7.3250E+01 3.9481E-02 > Solving for fluid > 14650 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 > 14650 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0555E-01 2.3985E+00 > 14650 Fluid done 7.3250E+01 5.4078E+00 > Step 14651, t= 7.3255000E+01, DT= 5.0000000E-03, C= NaN 2.8729E+04 7.0068E+00 > Solving for Hmholtz scalars > 14651 Error Hmholtz TEMP 200 NaN NaN 1.0000E-12 > 14651 Scalars done 7.3255E+01 1.0293E+00 > Solving for fluid > 14651 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 > 14651 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0602E-01 2.4098E+00 > 14651 Fluid done 7.3255E+01 5.3955E+00 > Step 14652, t= 7.3260000E+01, DT= 5.0000000E-03, C= NaN 2.8738E+04 9.1272E+00 > Solving for Hmholtz scalars > 14652 Error Hmholtz TEMP 200 NaN NaN 1.0000E-12 > 14652 Scalars done 7.3260E+01 1.0415E+00 > Solving for fluid > 14652 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 > 14652 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0562E-01 2.3933E+00 > 14652 Fluid done 7.3260E+01 5.3900E+00 > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > From nek5000-users at lists.mcs.anl.gov Tue Jun 6 08:52:25 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 6 Jun 2017 15:52:25 +0200 Subject: [Nek5000-users] Information in output file Message-ID: Deak Neks, I am new to NEK5000. I need help in understanding the contents of the output file. I understand that the four columns corresponding to "Hlmholtz" lines are iteration and tolerance information (kindly correct me if otherwise). What do the numbers in the four lines that follow represent? What are alpha and alph12? 23 1.8000920E+02 Perturbation Solve: 1 23 Hmholtz VELX: 12 4.8636E-13 2.1640E-01 1.0000E-12 23 Hmholtz VELY: 12 2.5736E-13 1.5550E-01 1.0000E-12 23 Hmholtz VELZ: 12 7.1335E-13 2.6858E-01 1.0000E-12 23 20 alpha: 3.4675E-01 2.0377E-02 1.6467E-02 -5.1081E-02 -1.1948E-02 3.2566E-02 -3.4915E-03 -9.6668E-03 -9.1184E-03 1.0323E-03 23 20 2.8358E-02 5.7262E-07 4.9524E+04 alph12 23 U-PRES gmres: 120 4.9518E-11 1.0000E-11 5.7262E-07 2.0179E+00 3.8401E+00 Step 24, t= 1.8000960E+02, DT= 4.0000000E-04, C= .273 9.2289E+01 4.1700E+00 PERTURBATION SOLVE PERTURBATION SOLVE Thanks. -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Jun 6 11:09:36 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 06 Jun 2017 19:09:36 +0300 Subject: [Nek5000-users] =?utf-8?q?Nek5000_documentation_details=3A_Pn-Pn_?= =?utf-8?q?pressure_solver?= In-Reply-To: References: Message-ID: Dear Ananias, ? thank you for a prompt and clear response! About the coupled Helmholtz solver, it is used to solve for three velocity components at once. Is it due to \nabla \mu^{n+1} \nabla v^{n+1} term? Thus, in the equation for v_x, for example, there are terms with derivatives of v_y and v_z, since they are at n+1 time step, they should go to the matrix, and not to the RHS of the equation.. Right? ? The second issue is that the same term with additional \nabla appears in the equation for Laplacian p^{n+1}. Do you treat it explicitly here? I mean at the time step n instead of n+1? ? Is there no conflict between implicit treatment of viscous terms at the `velocity' step while doing it explicitly during `pressure' step? ? Best regards, Vlad >???????, 6 ???? 2017, 17:51 +07:00 ?? nek5000-users at lists.mcs.anl.gov: > >Dear Vlad, >the low Mach Pn-Pn approach is based on the 1997 (JSC) and 1997 (JCP) papers you mention and it consists of 3 steps as you describe, i.e.: >a) first the velocity is updated using the extrapolated convective term,? >b) then the Laplacian of pressure is calculated due to convection, after that? >c) the velocity is updated using the pressure gradient and accounts for viscous term >The coupled Helmholtz solver is used for the velocities only when using ifstrs=true, that >is when you want to include the full stress tensor. Otherwise, it is using separate Helmholtz solves for each of the velocity components, similar to Pn-Pn-2.? >Hope this helps clarify things. >All the best, >Ananias > > >On Tue, Jun 6, 2017 at 7:29 AM, < nek5000-users at lists.mcs.anl.gov > wrote: >>Dear Neks, >> >>reading the documentation I got the impression that Pn-Pn solver (low Mach) first solves the pressure where the convective and viscous (!) terms are taken into account. After that using this p^{n+1} we solve for velocity field. It seems that the algorithm consists of only 2 steps (pressure + velocity). >> >>However, reading?the paper by Tomboulides, Lee, Orszag (1996) which is referenced inside the code, I see the projection algorithm where first the velocity is updated using the extrapolated convective term, then the Laplacian of pressure is calculated due to convection, after that the velocity is updated using convection and pressure gradient. The last step accounts for viscous term. >> >>I am a bit confused, could you please help me out here? Which method is used? >> >>PS. Another thing is the coupled Helmholtz solver in Pn-Pn. I see that in case of Pn-Pn-2 each velocity component is treated separately (segregated solver). However, this coupled thing slightly confuses me, why not treating it separately as in Pn-Pn-2? Could you please comment there as well? Thank you. >> >>Best regards, >>Vlad >>_______________________________________________ >>Nek5000-users mailing list >>Nek5000-users at lists.mcs.anl.gov >>https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > >_______________________________________________ >Nek5000-users mailing list >Nek5000-users at lists.mcs.anl.gov >https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Jun 6 12:59:53 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 6 Jun 2017 20:59:53 +0300 Subject: [Nek5000-users] Nek5000 documentation details: Pn-Pn pressure solver In-Reply-To: References: Message-ID: Dear Vlad, this is correct, the coupled Helmholtz solve is used in the case of the full stress tensor because in that case the stress tensor is not diagonal. The splitting approach is based on an irrotational-solenoidal decomposition of the velocity (which is described in the 1997 JSC paper); the divergence of the former, which appears in the rhs of the pressure equation is treated implicitly (it is zero in the case of constant viscosity and incompressible flow), whereas the divergence of the latter is treated explicitly through the vorticity (which also appears in the pressure rhs and the pressure BC and is again zero in the case of constant viscosity and incompressible flow; this is not the case in the pressure BC) . It was proved in the JSC and JCP papers that this splitting approach, which allows for an uncoupled solution of the pressure and velocity equations, leads to a high-order overall accuracy in time. Best, Ananias On Tue, Jun 6, 2017 at 7:09 PM, wrote: > Dear Ananias, > > thank you for a prompt and clear response! About the coupled Helmholtz > solver, it is used to solve for three velocity components at once. Is it > due to \nabla \mu^{n+1} \nabla v^{n+1} term? Thus, in the equation for v_x, > for example, there are terms with derivatives of v_y and v_z, since they > are at n+1 time step, they should go to the matrix, and not to the RHS of > the equation.. Right? > > The second issue is that the same term with additional \nabla appears in > the equation for Laplacian p^{n+1}. Do you treat it explicitly here? I mean > at the time step n instead of n+1? > > Is there no conflict between implicit treatment of viscous terms at the > `velocity' step while doing it explicitly during `pressure' step? > > Best regards, > Vlad > > > > ???????, 6 ???? 2017, 17:51 +07:00 ?? nek5000-users at lists.mcs.anl.gov: > > > Dear Vlad, > the low Mach Pn-Pn approach is based on the 1997 (JSC) and 1997 (JCP) > papers you mention and it consists of 3 steps as you describe, i.e.: > a) first the velocity is updated using the extrapolated convective term, > b) then the Laplacian of pressure is calculated due to convection, after > that > c) the velocity is updated using the pressure gradient and accounts for > viscous term > The coupled Helmholtz solver is used for the velocities only when using > ifstrs=true, that > is when you want to include the full stress tensor. Otherwise, it is using > separate Helmholtz solves for each of the velocity components, similar to > Pn-Pn-2. > Hope this helps clarify things. > All the best, > Ananias > > > On Tue, Jun 6, 2017 at 7:29 AM, wrote: > > Dear Neks, > > reading the documentation I got the impression that Pn-Pn solver (low > Mach) first solves the pressure where the convective and viscous (!) terms > are taken into account. After that using this p^{n+1} we solve for velocity > field. It seems that the algorithm consists of only 2 steps (pressure + > velocity). > > However, reading the paper by Tomboulides, Lee, Orszag (1996) which is > referenced inside the code, I see the projection algorithm where first the > velocity is updated using the extrapolated convective term, then the > Laplacian of pressure is calculated due to convection, after that the > velocity is updated using convection and pressure gradient. The last step > accounts for viscous term. > > I am a bit confused, could you please help me out here? Which method is > used? > > PS. Another thing is the coupled Helmholtz solver in Pn-Pn. I see that in > case of Pn-Pn-2 each velocity component is treated separately (segregated > solver). However, this coupled thing slightly confuses me, why not treating > it separately as in Pn-Pn-2? Could you please comment there as well? Thank > you. > > Best regards, > Vlad > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Jun 6 21:51:44 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 07 Jun 2017 05:51:44 +0300 Subject: [Nek5000-users] =?utf-8?q?Nek5000_documentation_details=3A_Pn-Pn_?= =?utf-8?q?pressure_solver?= In-Reply-To: References: Message-ID: Dear Ananias, thank you for your answer again, but I think, you told about another part of viscous term. In 1997 JSP paper it is clearly explained the situation, when \mu doesn't depend on temperature. But if I, for example, use the ? Sutherland's law there is another extra term with (\nabla \mu) and first derivations of velocity as it is shown on the figure below. I saw in the code and it seems like they are treated explicitly in the pressure solver, using the meaning of velocity at n-th time step. Is it so? Best regards, Vlad >?????, 7 ???? 2017, 2:43 +07:00 ?? nek5000-users at lists.mcs.anl.gov: > >Dear Vlad, > >this is correct, the coupled Helmholtz solve is used in the case of the full stress tensor >because in that case the stress tensor is not diagonal. > >The splitting approach is based on an irrotational-solenoidal decomposition of the velocity >(which is described in the 1997 JSC paper); the divergence of the former, which appears in >the rhs of the pressure equation is treated implicitly (it is zero in the case of constant viscosity >and incompressible flow), whereas the divergence of the latter is treated explicitly through the >vorticity (which also appears in the pressure rhs and the pressure BC and is again zero in the >case of constant viscosity and incompressible flow; this is not the case in the pressure BC) . > >It was proved in the JSC and JCP papers that this splitting approach, which allows for an >uncoupled solution of the pressure and velocity equations, leads to a high-order overall >accuracy in time. > >Best, >Ananias > >On Tue, Jun 6, 2017 at 7:09 PM, < nek5000-users at lists.mcs.anl.gov > wrote: >>Dear Ananias, >>? >>thank you for a prompt and clear response! About the coupled Helmholtz solver, it is used to solve for three velocity components at once. Is it due to \nabla \mu^{n+1} \nabla v^{n+1} term? Thus, in the equation for v_x, for example, there are terms with derivatives of v_y and v_z, since they are at n+1 time step, they should go to the matrix, and not to the RHS of the equation.. Right? >>? >>The second issue is that the same term with additional \nabla appears in the equation for Laplacian p^{n+1}. Do you treat it explicitly here? I mean at the time step n instead of n+1? >>? >>Is there no conflict between implicit treatment of viscous terms at the `velocity' step while doing it explicitly during `pressure' step? >>? >>Best regards, >>Vlad >> >> >>>???????, 6 ???? 2017, 17:51 +07:00 ?? nek5000-users at lists.mcs.anl.gov : >>> >>> >>>Dear Vlad, >>>the low Mach Pn-Pn approach is based on the 1997 (JSC) and 1997 (JCP) papers you mention and it consists of 3 steps as you describe, i.e.: >>>a) first the velocity is updated using the extrapolated convective term,? >>>b) then the Laplacian of pressure is calculated due to convection, after that? >>>c) the velocity is updated using the pressure gradient and accounts for viscous term >>>The coupled Helmholtz solver is used for the velocities only when using ifstrs=true, that >>>is when you want to include the full stress tensor. Otherwise, it is using separate Helmholtz solves for each of the velocity components, similar to Pn-Pn-2.? >>>Hope this helps clarify things. >>>All the best, >>>Ananias >>> >>> >>>On Tue, Jun 6, 2017 at 7:29 AM, < nek5000-users at lists.mcs.anl.gov > wrote: >>>>Dear Neks, >>>> >>>>reading the documentation I got the impression that Pn-Pn solver (low Mach) first solves the pressure where the convective and viscous (!) terms are taken into account. After that using this p^{n+1} we solve for velocity field. It seems that the algorithm consists of only 2 steps (pressure + velocity). >>>> >>>>However, reading?the paper by Tomboulides, Lee, Orszag (1996) which is referenced inside the code, I see the projection algorithm where first the velocity is updated using the extrapolated convective term, then the Laplacian of pressure is calculated due to convection, after that the velocity is updated using convection and pressure gradient. The last step accounts for viscous term. >>>> >>>>I am a bit confused, could you please help me out here? Which method is used? >>>> >>>>PS. Another thing is the coupled Helmholtz solver in Pn-Pn. I see that in case of Pn-Pn-2 each velocity component is treated separately (segregated solver). However, this coupled thing slightly confuses me, why not treating it separately as in Pn-Pn-2? Could you please comment there as well? Thank you. >>>> >>>>Best regards, >>>>Vlad >>>>_______________________________________________ >>>>Nek5000-users mailing list >>>>Nek5000-users at lists.mcs.anl.gov >>>>https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >>> >>>_______________________________________________ >>>Nek5000-users mailing list >>>Nek5000-users at lists.mcs.anl.gov >>>https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >>_______________________________________________ >>Nek5000-users mailing list >>Nek5000-users at lists.mcs.anl.gov >>https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> > >_______________________________________________ >Nek5000-users mailing list >Nek5000-users at lists.mcs.anl.gov >https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: image.png Type: image/png Size: 23241 bytes Desc: not available URL: From nek5000-users at lists.mcs.anl.gov Wed Jun 7 08:47:39 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 7 Jun 2017 16:47:39 +0300 Subject: [Nek5000-users] Nek5000 documentation details: Pn-Pn pressure solver In-Reply-To: References: Message-ID: Hi Vlad, what you say is correct about the treatment of the viscous term in the case of variable dynamic viscosity, i.e. the second term in your equation is indeed treated explicitly, by extrapolating the velocity using previous time steps (the gradient of mu is treated implicitly though as the dynamic viscosity was already updated to its n+1 value). The semi-implicit treatment of the full stress tensor in the case of variable dynamic viscosity is a fairly recent development (2015), which has not been published and is not described in the 1997/1998 papers or anywhere else in more detail. Note that because of the implicit treatment of del mu, the inclusion of this explicit term in the rhs of the pressure equation is not adding a severe diffusion-like CFL restriction to the time step (which is normally related to second order velocity spatial derivatives , i.e. Laplacian). I don't believe further discussion on this topic is of interest to the majority of Nek users, so if necessary let's continue any additional conversation off-line. All the best, Ananias On Wed, Jun 7, 2017 at 5:51 AM, wrote: > Dear Ananias, > > thank you for your answer again, but I think, you told about another part > of viscous term. In 1997 JSP paper it is clearly explained the situation, > when \mu doesn't depend on temperature. But if I, for example, use the Sutherland's > law there is another extra term with (\nabla \mu) and first derivations of > velocity as it is shown on the figure below. > > > I saw in the code and it seems like they are treated explicitly in the > pressure solver, using the meaning of velocity at n-th time step. Is it so? > > Best regards, > Vlad > > > ?????, 7 ???? 2017, 2:43 +07:00 ?? nek5000-users at lists.mcs.anl.gov: > > > Dear Vlad, > > this is correct, the coupled Helmholtz solve is used in the case of the > full stress tensor > because in that case the stress tensor is not diagonal. > > The splitting approach is based on an irrotational-solenoidal > decomposition of the velocity > (which is described in the 1997 JSC paper); the divergence of the former, > which appears in > the rhs of the pressure equation is treated implicitly (it is zero in the > case of constant viscosity > and incompressible flow), whereas the divergence of the latter is treated > explicitly through the > vorticity (which also appears in the pressure rhs and the pressure BC and > is again zero in the > case of constant viscosity and incompressible flow; this is not the case > in the pressure BC) . > > It was proved in the JSC and JCP papers that this splitting approach, > which allows for an > uncoupled solution of the pressure and velocity equations, leads to a > high-order overall > accuracy in time. > > Best, > Ananias > > On Tue, Jun 6, 2017 at 7:09 PM, wrote: > > Dear Ananias, > > thank you for a prompt and clear response! About the coupled Helmholtz > solver, it is used to solve for three velocity components at once. Is it > due to \nabla \mu^{n+1} \nabla v^{n+1} term? Thus, in the equation for v_x, > for example, there are terms with derivatives of v_y and v_z, since they > are at n+1 time step, they should go to the matrix, and not to the RHS of > the equation.. Right? > > The second issue is that the same term with additional \nabla appears in > the equation for Laplacian p^{n+1}. Do you treat it explicitly here? I mean > at the time step n instead of n+1? > > Is there no conflict between implicit treatment of viscous terms at the > `velocity' step while doing it explicitly during `pressure' step? > > Best regards, > Vlad > > > > ???????, 6 ???? 2017, 17:51 +07:00 ?? nek5000-users at lists.mcs.anl.gov: > > > Dear Vlad, > the low Mach Pn-Pn approach is based on the 1997 (JSC) and 1997 (JCP) > papers you mention and it consists of 3 steps as you describe, i.e.: > a) first the velocity is updated using the extrapolated convective term, > b) then the Laplacian of pressure is calculated due to convection, after > that > c) the velocity is updated using the pressure gradient and accounts for > viscous term > The coupled Helmholtz solver is used for the velocities only when using > ifstrs=true, that > is when you want to include the full stress tensor. Otherwise, it is using > separate Helmholtz solves for each of the velocity components, similar to > Pn-Pn-2. > Hope this helps clarify things. > All the best, > Ananias > > > On Tue, Jun 6, 2017 at 7:29 AM, wrote: > > Dear Neks, > > reading the documentation I got the impression that Pn-Pn solver (low > Mach) first solves the pressure where the convective and viscous (!) terms > are taken into account. After that using this p^{n+1} we solve for velocity > field. It seems that the algorithm consists of only 2 steps (pressure + > velocity). > > However, reading the paper by Tomboulides, Lee, Orszag (1996) which is > referenced inside the code, I see the projection algorithm where first the > velocity is updated using the extrapolated convective term, then the > Laplacian of pressure is calculated due to convection, after that the > velocity is updated using convection and pressure gradient. The last step > accounts for viscous term. > > I am a bit confused, could you please help me out here? Which method is > used? > > PS. Another thing is the coupled Helmholtz solver in Pn-Pn. I see that in > case of Pn-Pn-2 each velocity component is treated separately (segregated > solver). However, this coupled thing slightly confuses me, why not treating > it separately as in Pn-Pn-2? Could you please comment there as well? Thank > you. > > Best regards, > Vlad > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: image.png Type: image/png Size: 23241 bytes Desc: not available URL: From nek5000-users at lists.mcs.anl.gov Wed Jun 7 09:09:36 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 7 Jun 2017 14:09:36 +0000 Subject: [Nek5000-users] An abrupt error of "unconverged Helmh3 In-Reply-To: References: Message-ID: Hi, Thanks for your reply. The simulation is in a 3D box with moving meshes. Surface waves are excited by the fluctuation of ambient pressure. Based on my tests, the error of unconverged Helmh3 fluid seems to be related with random heat flux, which is set as flux = -1.0e-3 big = 1.0e7*ix + 1.0e8*iy + 1.0e9*ieg random = sin(big) flux = flux + random*flux/10.0 The heat flux changes buoyancy, affecting fluid motions. When the random flux is removed, the simulation runs well. Thanks. > On Jun 6, 2017, at 12:09 PM, nek5000-users-request at lists.mcs.anl.gov wrote: > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Tue, 6 Jun 2017 13:27:40 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > Subject: [Nek5000-users] An abrupt error of "unconverged Helmh3 fluid" > Message-ID: > > Content-Type: text/plain; charset="us-ascii" > > Dear Neks, > > I am using the moving mesh to simulate surface gravity waves with some random heat flux at the surface. > The codes ran well for a while, but it suddenly pop out an error of unconverged Helmh3 fluid, as shown in the below log file (Step 14650). > > Could anyone give me a hint on this kind of error? > > Thank you, > Peng > > > Step 14647, t= 7.3235000E+01, DT= 5.0000000E-03, C= 0.036 2.8716E+04 1.9695E+00 > Solving for Hmholtz scalars > 14647 Hmholtz TEMP 4 6.0654E-16 1.4699E-01 1.0000E-12 > 14647 Scalars done 7.3235E+01 3.9568E-02 > Solving for fluid > 14647 Helmh3 fluid 4 1.9934E-13 8.0532E-04 1.0000E-12 > 14647 U-PRES gmres 14 4.5296E-11 2.2779E-03 1.0000E-10 8.2958E-02 2.5954E-01 > 14647 Fluid done 7.3235E+01 3.7241E-01 > Step 14648, t= 7.3240000E+01, DT= 5.0000000E-03, C= 0.035 2.8718E+04 1.9642E+00 > Solving for Hmholtz scalars > 14648 Hmholtz TEMP 4 6.0987E-16 1.4701E-01 1.0000E-12 > 14648 Scalars done 7.3240E+01 3.9551E-02 > Solving for fluid > 14648 Helmh3 fluid 4 1.9904E-13 8.0535E-04 1.0000E-12 > 14648 U-PRES gmres 14 4.5647E-11 2.2779E-03 1.0000E-10 8.3140E-02 2.5985E-01 > 14648 Fluid done 7.3240E+01 3.7297E-01 > Step 14649, t= 7.3245000E+01, DT= 5.0000000E-03, C= 0.034 2.8720E+04 1.9658E+00 > Solving for Hmholtz scalars > 14649 Hmholtz TEMP 4 6.1309E-16 1.4704E-01 1.0000E-12 > 14649 Scalars done 7.3245E+01 3.9058E-02 > Solving for fluid > 14649 Helmh3 fluid 4 1.9875E-13 8.0535E-04 1.0000E-12 > 14649 U-PRES gmres 14 4.5884E-11 2.2778E-03 1.0000E-10 8.3552E-02 2.5956E-01 > 14649 Fluid done 7.3245E+01 3.7226E-01 > Step 14650, t= 7.3250000E+01, DT= 5.0000000E-03, C= 0.033 2.8722E+04 1.9614E+00 > Solving for Hmholtz scalars > 14650 Hmholtz TEMP 4 6.1615E-16 1.4707E-01 1.0000E-12 > 14650 Scalars done 7.3250E+01 3.9481E-02 > Solving for fluid > 14650 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 > 14650 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0555E-01 2.3985E+00 > 14650 Fluid done 7.3250E+01 5.4078E+00 > Step 14651, t= 7.3255000E+01, DT= 5.0000000E-03, C= NaN 2.8729E+04 7.0068E+00 > Solving for Hmholtz scalars > 14651 Error Hmholtz TEMP 200 NaN NaN 1.0000E-12 > 14651 Scalars done 7.3255E+01 1.0293E+00 > Solving for fluid > 14651 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 > 14651 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0602E-01 2.4098E+00 > 14651 Fluid done 7.3255E+01 5.3955E+00 > Step 14652, t= 7.3260000E+01, DT= 5.0000000E-03, C= NaN 2.8738E+04 9.1272E+00 > Solving for Hmholtz scalars > 14652 Error Hmholtz TEMP 200 NaN NaN 1.0000E-12 > 14652 Scalars done 7.3260E+01 1.0415E+00 > Solving for fluid > 14652 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 > 14652 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0562E-01 2.3933E+00 > 14652 Fluid done 7.3260E+01 5.3900E+00 > > ------------------------------ > > Message: 2 > Date: Tue, 6 Jun 2017 13:33:50 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > Subject: Re: [Nek5000-users] An abrupt error of "unconverged Helmh3 > fluid" > Message-ID: > > Content-Type: text/plain; charset="us-ascii" > > Pend, > > can you please give more details on the case you are trying to run with Nek5000? How did you set up your case? Did you start from one of the examples? > > Thanks. >> On Jun 6, 2017, at 9:27 AM, nek5000-users at lists.mcs.anl.gov wrote: >> >> Dear Neks, >> >> I am using the moving mesh to simulate surface gravity waves with some random heat flux at the surface. >> The codes ran well for a while, but it suddenly pop out an error of unconverged Helmh3 fluid, as shown in the below log file (Step 14650). >> >> Could anyone give me a hint on this kind of error? >> >> Thank you, >> Peng >> >> >> Step 14647, t= 7.3235000E+01, DT= 5.0000000E-03, C= 0.036 2.8716E+04 1.9695E+00 >> Solving for Hmholtz scalars >> 14647 Hmholtz TEMP 4 6.0654E-16 1.4699E-01 1.0000E-12 >> 14647 Scalars done 7.3235E+01 3.9568E-02 >> Solving for fluid >> 14647 Helmh3 fluid 4 1.9934E-13 8.0532E-04 1.0000E-12 >> 14647 U-PRES gmres 14 4.5296E-11 2.2779E-03 1.0000E-10 8.2958E-02 2.5954E-01 >> 14647 Fluid done 7.3235E+01 3.7241E-01 >> Step 14648, t= 7.3240000E+01, DT= 5.0000000E-03, C= 0.035 2.8718E+04 1.9642E+00 >> Solving for Hmholtz scalars >> 14648 Hmholtz TEMP 4 6.0987E-16 1.4701E-01 1.0000E-12 >> 14648 Scalars done 7.3240E+01 3.9551E-02 >> Solving for fluid >> 14648 Helmh3 fluid 4 1.9904E-13 8.0535E-04 1.0000E-12 >> 14648 U-PRES gmres 14 4.5647E-11 2.2779E-03 1.0000E-10 8.3140E-02 2.5985E-01 >> 14648 Fluid done 7.3240E+01 3.7297E-01 >> Step 14649, t= 7.3245000E+01, DT= 5.0000000E-03, C= 0.034 2.8720E+04 1.9658E+00 >> Solving for Hmholtz scalars >> 14649 Hmholtz TEMP 4 6.1309E-16 1.4704E-01 1.0000E-12 >> 14649 Scalars done 7.3245E+01 3.9058E-02 >> Solving for fluid >> 14649 Helmh3 fluid 4 1.9875E-13 8.0535E-04 1.0000E-12 >> 14649 U-PRES gmres 14 4.5884E-11 2.2778E-03 1.0000E-10 8.3552E-02 2.5956E-01 >> 14649 Fluid done 7.3245E+01 3.7226E-01 >> Step 14650, t= 7.3250000E+01, DT= 5.0000000E-03, C= 0.033 2.8722E+04 1.9614E+00 >> Solving for Hmholtz scalars >> 14650 Hmholtz TEMP 4 6.1615E-16 1.4707E-01 1.0000E-12 >> 14650 Scalars done 7.3250E+01 3.9481E-02 >> Solving for fluid >> 14650 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 >> 14650 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0555E-01 2.3985E+00 >> 14650 Fluid done 7.3250E+01 5.4078E+00 >> Step 14651, t= 7.3255000E+01, DT= 5.0000000E-03, C= NaN 2.8729E+04 7.0068E+00 >> Solving for Hmholtz scalars >> 14651 Error Hmholtz TEMP 200 NaN NaN 1.0000E-12 >> 14651 Scalars done 7.3255E+01 1.0293E+00 >> Solving for fluid >> 14651 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 >> 14651 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0602E-01 2.4098E+00 >> 14651 Fluid done 7.3255E+01 5.3955E+00 >> Step 14652, t= 7.3260000E+01, DT= 5.0000000E-03, C= NaN 2.8738E+04 9.1272E+00 >> Solving for Hmholtz scalars >> 14652 Error Hmholtz TEMP 200 NaN NaN 1.0000E-12 >> 14652 Scalars done 7.3260E+01 1.0415E+00 >> Solving for fluid >> 14652 201 Unconverged Helmh3 fluid rbnorm = NaN 0.100000E-11 >> 14652 U-PRES gmres 120 NaN NaN 1.0000E-10 7.0562E-01 2.3933E+00 >> 14652 Fluid done 7.3260E+01 5.3900E+00 >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://urldefense.proofpoint.com/v2/url?u=https-3A__lists.mcs.anl.gov_mailman_listinfo_nek5000-2Dusers&d=DwICAg&c=y2w-uYmhgFWijp_IQN0DhA&r=QEl5evD27KmYmPJRHh_RJPXwtG0VKYwu68jKU6Xd1hw&m=HnNT_MhdQLlwY0JA866_ttLabHwe_juGfYHIpwi2jOM&s=tKCD3fDJwCYITa5L-EBwfbmcIBJB_pu7q_VczBM-nGo&e= >> > From nek5000-users at lists.mcs.anl.gov Thu Jun 8 10:19:02 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 8 Jun 2017 11:19:02 -0400 Subject: [Nek5000-users] ifort Issues; Runtime Error Message-ID: Hi NEKS... I'm running my problem on a new machine (Comet at SDSC), and they recommend the Intel compilers, which I tried... Got compilation to work; then run started and crashed with "... error reading .re2 file...". Made sure all scripts, .re2 file, basename, etc... appeared OK; finally re-compiler with GNU and it then ran fine... Any ideas? Are there any compiler options I should be using with ifort that I'm missing? Also, just wanted to note that the makefiles for tools and core do not descend into all the sub-directories when running "make clean"; for tools, you have to go to each sub-directory to clean the objects; in core, the 3rd-party folders have to be manually cleaned or the existing objects will be used instead of the recompiled versions... Thanks, Murph --- This email has been checked for viruses by AVG. http://www.avg.com From nek5000-users at lists.mcs.anl.gov Thu Jun 8 11:06:01 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 8 Jun 2017 18:06:01 +0200 Subject: [Nek5000-users] ifort Issues; Runtime Error In-Reply-To: References: Message-ID: Is this with the master? Note only makenek realclean will delete all 3rd party objects. Please open an issue if maketools clean doesn't work for you. On 8 Jun 2017, at 17:19, "nek5000-users at lists.mcs.anl.gov " > wrote: Hi NEKS... I'm running my problem on a new machine (Comet at SDSC), and they recommend the Intel compilers, which I tried...? Got compilation to work; then run started and crashed with "... error reading .re2 file...".? Made sure all scripts, .re2 file, basename, etc... appeared OK; finally re-compiler with GNU and it then ran fine... Any ideas?? Are there any compiler options I should be using with ifort that I'm missing? Also, just wanted to note that the makefiles for tools and core do not descend into all the sub-directories when running "make clean"; for tools, you have to go to each sub-directory to clean the objects; in core, the 3rd-party folders have to be manually cleaned or the existing objects will be used instead of the recompiled versions... Thanks, Murph --- This email has been checked for viruses by AVG. http://www.avg.com _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Jun 12 11:15:53 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 12 Jun 2017 18:15:53 +0200 Subject: [Nek5000-users] Courant for passive scalar ? In-Reply-To: References: Message-ID: Dear Neks I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime. I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity. has this already been implemented ? or how do people avoid this ? any suggestion is more than welcome. thanks Agnese Agnese Seminara -------------------------------- CNRS Institut de Physique de Nice Parc Valrose avenue J Vallot 06108 Nice, France +33 (0) 492 076 775 http://sites.unice.fr/site/aseminara/ -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Jun 12 13:00:26 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 12 Jun 2017 20:00:26 +0200 Subject: [Nek5000-users] Courant for passive scalar ? In-Reply-To: References: Message-ID: Hi, I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk. Best, Jan > Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov: > > Dear Neks > > I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime. > > I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity. > > has this already been implemented ? > or how do people avoid this ? > > any suggestion is more than welcome. > thanks > Agnese > > > > Agnese Seminara > -------------------------------- > CNRS > Institut de Physique de Nice > Parc Valrose > avenue J Vallot > 06108 Nice, France > +33 (0) 492 076 775 > http://sites.unice.fr/site/aseminara/ > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Jun 13 05:12:49 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 13 Jun 2017 11:12:49 +0100 Subject: [Nek5000-users] prenek Message-ID: Hi nek, I have a 2D mesh. When I used global refine function, the errors occurred. do you know why? OCT/Multi-SPLIT quad or multi-split? (q/m) Number of elements after OctSplit operation: 7680 Are you sure you want to split (y/n)? inside gencen 7680 F 4 done gencen 7680 start locglob_lexico: 4 7680 30720 2.00000009E-03 locglob: 1 1 30720 locglob: 2 3664 30720 locglob: 1 7761 30720 locglob: 2 7761 30720 Performing unique_vertex2 self_chk 30720 done locglob_lexico: 7761 7761 30720 4 Performing makecell self_chk 30720 RMIN: 0.153545E-03 EPS,RMIN: 0.500000E-01 0.153545E-03 I/O Error: No String Terminator sent to PRS zeroing y: 1 82 -0.75715434E-10 I/O Error: No String Terminator sent to PRS zeroing y: 4 82 -0.14778734E-11 I/O Error: No String Terminator sent to PRS zeroing y: 4 83 0.72759583E-10 I/O Error: No String Terminator sent to PRS zeroing y: 4 123 0.15279482E-09 ...... Does it mean my mesh is still not right? -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Jun 14 03:44:13 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 14 Jun 2017 10:44:13 +0200 Subject: [Nek5000-users] Courant for passive scalar ? In-Reply-To: References: Message-ID: Thanks Jan ! so it is confirmed that the tilmestep is based exclusively on the velocity field ? how do people make sure that the passive scalars do not blow up ? anybody ? any help is greatly appreciated !! agnese Agnese Seminara -------------------------------- CNRS Institut de Physique de Nice Parc Valrose avenue J Vallot 06108 Nice, France +33 (0) 492 076 775 http://sites.unice.fr/site/aseminara/ > On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote: > > Send Nek5000-users mailing list submissions to > nek5000-users at lists.mcs.anl.gov > > To subscribe or unsubscribe via the World Wide Web, visit > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > or, via email, send a message with subject or body 'help' to > nek5000-users-request at lists.mcs.anl.gov > > You can reach the person managing the list at > nek5000-users-owner at lists.mcs.anl.gov > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Nek5000-users digest..." > Today's Topics: > > 1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov) > 2. prenek (nek5000-users at lists.mcs.anl.gov) > > From: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Courant for passive scalar ? > Date: June 12, 2017 at 8:00:26 PM GMT+2 > To: nek5000-users at lists.mcs.anl.gov > Reply-To: nek5000-users at lists.mcs.anl.gov > > > Hi, > I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk. > > Best, > Jan > > Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov : > >> Dear Neks >> >> I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime. >> >> I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity. >> >> has this already been implemented ? >> or how do people avoid this ? >> >> any suggestion is more than welcome. >> thanks >> Agnese >> >> >> >> Agnese Seminara >> -------------------------------- >> CNRS >> Institut de Physique de Nice >> Parc Valrose >> avenue J Vallot >> 06108 Nice, France >> +33 (0) 492 076 775 >> http://sites.unice.fr/site/aseminara/ >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > From: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] prenek > Date: June 13, 2017 at 12:12:49 PM GMT+2 > To: nek5000-users at lists.mcs.anl.gov > Reply-To: nek5000-users at lists.mcs.anl.gov > > > Hi nek, > > I have a 2D mesh. When I used global refine function, the errors occurred. do you know why? > > OCT/Multi-SPLIT > quad or multi-split? (q/m) > Number of elements after OctSplit operation: 7680 > Are you sure you want to split (y/n)? > inside gencen 7680 F 4 > done gencen 7680 > start locglob_lexico: 4 7680 30720 2.00000009E-03 > locglob: 1 1 30720 > locglob: 2 3664 30720 > locglob: 1 7761 30720 > locglob: 2 7761 30720 > Performing unique_vertex2 self_chk 30720 > done locglob_lexico: 7761 7761 30720 4 > Performing makecell self_chk 30720 > > RMIN: 0.153545E-03 > EPS,RMIN: 0.500000E-01 0.153545E-03 > I/O Error: No String Terminator sent to PRS > zeroing y: 1 82 -0.75715434E-10 > I/O Error: No String Terminator sent to PRS > zeroing y: 4 82 -0.14778734E-11 > I/O Error: No String Terminator sent to PRS > zeroing y: 4 83 0.72759583E-10 > I/O Error: No String Terminator sent to PRS > zeroing y: 4 123 0.15279482E-09 > ...... > > Does it mean my mesh is still not right? > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Jun 14 07:37:35 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 14 Jun 2017 14:37:35 +0200 Subject: [Nek5000-users] Courant for passive scalar ? In-Reply-To: References: Message-ID: How do you define a courant number for a scalar? -----Original message----- > From:nek5000-users at lists.mcs.anl.gov > Sent: Wednesday 14th June 2017 10:43 > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Courant for passive scalar ? > > Thanks Jan ! >
so it is confirmed that the tilmestep is based exclusively on the velocity field ? > how do people make sure that the passive scalars do not blow up ?? > anybody ? >
any help is greatly appreciated !! > agnese >

Agnese Seminara > -------------------------------- > CNRS > Institut de Physique de Nice > Parc Valrose? > avenue J Vallot > 06108 Nice, France > +33 (0) 492 076 775 > http://sites.unice.fr/site/aseminara/
On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote: >
Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

??1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov)
??2. prenek (nek5000-users@ lists.mc s.anl.gov)

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi, > I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk. >
Best, > Jan >
Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov :

Dear Neks >
I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime. >
I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity.? >
has this already been implemented ? > or how do people avoid this ? >
any suggestion is more than welcome. > thanks > Agnese >


Agnese Seminara > -------------------------------- > CNRS > Institut de Physique de Nice > Parc Valrose? > avenue J Vallot > 06108 Nice, France > +33 (0) 492 076 775 > http://sites.unice.fr/site/aseminara/
_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



From: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi nek, >
I have a 2D mesh. When I used global refine function, the errors occurred. do you know why? >
?OCT/Multi-SPLIT > quad or multi-split? (q/m) > Number of elements after OctSplit operation: ? ?7680 > Are you sure you want to split (y/n)? > ?inside gencen ? ? ? ? 7680 F ? ? ? ? ? 4 > ?done gencen ? ? ? ? 7680 > ?start locglob_lexico: ? ? ? ? ? 4 ? ? ? ?7680 ? ? ? 30720 ? 2.00000009E-03 > ?locglob: ? ? ? ? ? 1 ? ? ? ? ? 1 ? ? ? 30720 > ?locglob: ? ? ? ? ? 2 ? ? ? ?3664 ? ? ? 30720 > ?locglob: ? ? ? ? ? 1 ? ? ? ?7761 ? ? ? 30720 > ?locglob: ? ? ? ? ? 2 ? ? ? ?7761 ? ? ? 30720 > ?Performing unique_vertex2 self_chk ? ? ? 30720 > ?done locglob_lexico: ? ? ?7761 ? ? ?7761 ? ? 30720 ? ? ? ? 4 > ?Performing makecell self_chk ? ? ? 30720 >
? RMIN: 0.153545E-03 > ? EPS,RMIN: 0.500000E-01 0.153545E-03 > ?I/O Error: No String Terminator sent to PRS > ?zeroing y: ? ? 1 ?82 -0.75715434E-10 > ?I/O Error: No String Terminator sent to PRS > ?zeroing y: ? ? 4 ?82 -0.14778734E-11 > ?I/O Error: No String Terminator sent to PRS > ?zeroing y: ? ? 4 ?83 ?0.72759583E-10 > ?I/O Error: No String Terminator sent to PRS > ?zeroing y: ? ? 4 123 ?0.15279482E-09 > ...... >
Does it mean my mesh is still not right? >

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users

_______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Fri Jun 9 00:04:01 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 9 Jun 2017 01:04:01 -0400 Subject: [Nek5000-users] Problem with set_object Message-ID: Dear Neks, I'm trying to calculate the drag on a cylinder surface in a flow. I started from "ext_cyl", "TurbChannel" and some threads on this list to understand how to set objects. I'm using torque_calc in user check: scale = 2. ! Cd = F/(.5 rho U^2 ) = 2*F if (mod(istep,10).eq.0) call torque_calc(scale,x0,.true.,.false.) The problem is that the cylinder is rotating, so its boundary conditions are "v ", and the values are set to be the tangential velocity that I need, for ux and uy (2D case), so I can't follow exactly the procedure en "ext_cyl". What I've been doing is something similar to the following: https://lists.mcs.anl.gov/mailman/htdig/nek5000-users/2011-September/001473.html but it can't get the right Cd value. I'm testing my .usr file with "exy_cyl" case, trying to obtain the same result but it doesn't work. For example, with Re=20, I should get a Cd=2.0 aprox., but I'm getting 25.0 aprox. My set object subroutine is: subroutine set_obj ! define objects for surface integrals c include 'SIZE' include 'TOTAL' integer e,f,eg real xx,yy,nx,ny,rn nobj = 1 iobj = 0 do ii=nhis+1,nhis+nobj iobj = iobj+1 hcode(10,ii) = 'I' hcode( 1,ii) = 'F' hcode( 2,ii) = 'F' hcode( 3,ii) = 'F' lochis(1,ii) = iobj enddo nhis = nhis + nobj if (maxobj.lt.nobj) call exitti('increase maxobj in SIZE$',nobj) nxyz = nx1*ny1*nz1 nface = 2*ndim do e=1,nelv do ix=1,nx1 do iy=1,ny1 do iz=1,nz1 xx=xm1(ix,iy,iz,e) yy=ym1(ix,iy,iz,e) if ((sqrt(xx*xx+yy*yy)).lt.(0.51)) then do f=1,nface if (cbc(f,e,1).eq.'W ') then iobj = 1 if (iobj.gt.0) then nmember(iobj) = nmember(iobj) + 1 mem = nmember(iobj) eg = lglel(e) object(iobj,mem,1) = eg object(iobj,mem,2) = f endif endif enddo endif enddo enddo enddo enddo c write(6,*) 'number',(nmember(k),k=1,4) c return end So what I'm trying to do is to capture the cylinder (radius=0.5) on a 0.51 radius circle, and to identify the faces with "W " B.C. I think it should work and give me the same result as the one obtained with the original ext_cyl.usr. What am I doing wrong? Thanks in advace, Juan Pablo. -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Jun 14 11:34:50 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 14 Jun 2017 16:34:50 +0000 Subject: [Nek5000-users] Moving mesh boundary conditions Message-ID: Hi all, I am new to Nek5000, and am using the ALE capability in Nek500. I am trying to override the default way Nek5000 sets the mesh boundary conditions based on the fluid/thermal boundary conditions (section 5.7 in the PRENEK manual). On one boundary I have a Dirichlet velocity condition, 'v ', and on another I have an outflow condition, 'O '. On both these boundaries I would like to allow the mesh to move freely, without any constraint. At the moment Nek5000 either fixes the mesh or allows the mesh nodes to slide freely in the tangential direction with a zero normal velocity. It seems that here https://github.com/Nek5000/Nek5000/blob/master/core/mvmesh.f#L41 the mesh boundary conditions are set for the velocity boundary. Would the mesh be free to move in any direction on the velocity boundary if in mvmesh.f I change: IF (CBF(1:1).EQ.'V' .OR. CBF(1:1).EQ.'v' .OR. $ CBF(1:1).EQ.'W' ) THEN IFLD = 1 CB = 'FIX' IF (IFMELT .OR. CBM.EQ.'+') CB='SYM' GOTO 200 ENDIF to IF (CBF(1:1).EQ.'V' .OR. CBF(1:1).EQ.'v' .OR. $ CBF(1:1).EQ.'W' ) THEN IFLD = 1 IF (IFMELT .OR. CBM.EQ.'+') CB='SYM' GOTO 200 ENDIF ? I did not notice an equivalent statement for the outflow condition 'O ' in mvmesh.f, is this set somewhere else? Any advice would be very much appreciated. Thanks very much, Louis Steytler Department of Mechanical Science and Engineering University of Illinois at Urbana-Champaign 1206 West Green Street Urbana, Il 61801 steytle1 at illinois.edu -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Jun 15 02:30:38 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 15 Jun 2017 09:30:38 +0200 Subject: [Nek5000-users] Courant number for passive scalar In-Reply-To: References: Message-ID: I havent seen the equivalent of a Courant criterion for passive scalars. what I meant is the scalar is intermittent, so one should have a way to detect the regions where gradients are extremely large (fronts) and make sure to decrease the timestep there. fronts do not correspond to regions where velocities are especially large, so the classic courant criterion will miss them. does this make sense ? any suggestions ? agnese > On Jun 14, 2017, at 2:39 PM, nek5000-users-request at lists.mcs.anl.gov wrote: > > Send Nek5000-users mailing list submissions to > nek5000-users at lists.mcs.anl.gov > > To subscribe or unsubscribe via the World Wide Web, visit > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > or, via email, send a message with subject or body 'help' to > nek5000-users-request at lists.mcs.anl.gov > > You can reach the person managing the list at > nek5000-users-owner at lists.mcs.anl.gov > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Nek5000-users digest..." > Today's Topics: > > 1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov) > 2. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov) > > From: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Courant for passive scalar ? > Date: June 14, 2017 at 10:44:13 AM GMT+2 > To: nek5000-users at lists.mcs.anl.gov > Reply-To: nek5000-users at lists.mcs.anl.gov > > > Thanks Jan ! > > so it is confirmed that the tilmestep is based exclusively on the velocity field ? > how do people make sure that the passive scalars do not blow up ? > anybody ? > > any help is greatly appreciated !! > agnese > > > Agnese Seminara > -------------------------------- > CNRS > Institut de Physique de Nice > Parc Valrose > avenue J Vallot > 06108 Nice, France > +33 (0) 492 076 775 > http://sites.unice.fr/site/aseminara/ >> On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote: >> >> Send Nek5000-users mailing list submissions to >> nek5000-users at lists.mcs.anl.gov >> >> To subscribe or unsubscribe via the World Wide Web, visit >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> or, via email, send a message with subject or body 'help' to >> nek5000-users-request at lists.mcs.anl.gov >> >> You can reach the person managing the list at >> nek5000-users-owner at lists.mcs.anl.gov >> >> When replying, please edit your Subject line so it is more specific >> than "Re: Contents of Nek5000-users digest..." >> Today's Topics: >> >> 1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov) >> 2. prenek (nek5000-users at lists.mcs.anl.gov) >> >> From: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Courant for passive scalar ? >> Date: June 12, 2017 at 8:00:26 PM GMT+2 >> To: nek5000-users at lists.mcs.anl.gov >> Reply-To: nek5000-users at lists.mcs.anl.gov >> >> >> Hi, >> I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk. >> >> Best, >> Jan >> >> Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov : >> >>> Dear Neks >>> >>> I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime. >>> >>> I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity. >>> >>> has this already been implemented ? >>> or how do people avoid this ? >>> >>> any suggestion is more than welcome. >>> thanks >>> Agnese >>> >>> >>> >>> Agnese Seminara >>> -------------------------------- >>> CNRS >>> Institut de Physique de Nice >>> Parc Valrose >>> avenue J Vallot >>> 06108 Nice, France >>> +33 (0) 492 076 775 >>> http://sites.unice.fr/site/aseminara/ >>> _______________________________________________ >>> Nek5000-users mailing list >>> Nek5000-users at lists.mcs.anl.gov >>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> >> From: nek5000-users at lists.mcs.anl.gov >> Subject: [Nek5000-users] prenek >> Date: June 13, 2017 at 12:12:49 PM GMT+2 >> To: nek5000-users at lists.mcs.anl.gov >> Reply-To: nek5000-users at lists.mcs.anl.gov >> >> >> Hi nek, >> >> I have a 2D mesh. When I used global refine function, the errors occurred. do you know why? >> >> OCT/Multi-SPLIT >> quad or multi-split? (q/m) >> Number of elements after OctSplit operation: 7680 >> Are you sure you want to split (y/n)? >> inside gencen 7680 F 4 >> done gencen 7680 >> start locglob_lexico: 4 7680 30720 2.00000009E-03 >> locglob: 1 1 30720 >> locglob: 2 3664 30720 >> locglob: 1 7761 30720 >> locglob: 2 7761 30720 >> Performing unique_vertex2 self_chk 30720 >> done locglob_lexico: 7761 7761 30720 4 >> Performing makecell self_chk 30720 >> >> RMIN: 0.153545E-03 >> EPS,RMIN: 0.500000E-01 0.153545E-03 >> I/O Error: No String Terminator sent to PRS >> zeroing y: 1 82 -0.75715434E-10 >> I/O Error: No String Terminator sent to PRS >> zeroing y: 4 82 -0.14778734E-11 >> I/O Error: No String Terminator sent to PRS >> zeroing y: 4 83 0.72759583E-10 >> I/O Error: No String Terminator sent to PRS >> zeroing y: 4 123 0.15279482E-09 >> ...... >> >> Does it mean my mesh is still not right? >> >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > > From: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Courant for passive scalar ? > Date: June 14, 2017 at 2:37:35 PM GMT+2 > To: nek5000-users at lists.mcs.anl.gov > Reply-To: nek5000-users at lists.mcs.anl.gov > > > How do you define a courant number for a scalar? > > -----Original message----- >> From:nek5000-users at lists.mcs.anl.gov >> Sent: Wednesday 14th June 2017 10:43 >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Courant for passive scalar ? >> >> Thanks Jan ! >>
so it is confirmed that the tilmestep is based exclusively on the velocity field ? >> how do people make sure that the passive scalars do not blow up ? >> anybody ? >>
any help is greatly appreciated !! >> agnese >>

Agnese Seminara >> -------------------------------- >> CNRS >> Institut de Physique de Nice >> Parc Valrose >> avenue J Vallot >> 06108 Nice, France >> +33 (0) 492 076 775 >> http://sites.unice.fr/site/aseminara/
On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote: >>
Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov)
2. prenek (nek5000-users@ > lists.mc > s.anl.gov)

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi, >> I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk. >>
Best, >> Jan >>
Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov :

Dear Neks >>
I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime. >>
I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity. >>
has this already been implemented ? >> or how do people avoid this ? >>
any suggestion is more than welcome. >> thanks >> Agnese >>


Agnese Seminara >> -------------------------------- >> CNRS >> Institut de Physique de Nice >> Parc Valrose >> avenue J Vallot >> 06108 Nice, France >> +33 (0) 492 076 775 >> http://sites.unice.fr/site/aseminara/
_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



From: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi nek, >>
I have a 2D mesh. When I used global refine function, the errors occurred. do you know why? >>
OCT/Multi-SPLIT >> quad or multi-split? (q/m) >> Number of elements after OctSplit operation: 7680 >> Are you sure you want to split (y/n)? >> inside gencen 7680 F 4 >> done gencen 7680 >> start locglob_lexico: 4 7680 30720 2.00000009E-03 >> locglob: 1 1 30720 >> locglob: 2 3664 30720 >> locglob: 1 7761 30720 >> locglob: 2 7761 30720 >> Performing unique_vertex2 self_chk 30720 >> done locglob_lexico: 7761 7761 30720 4 >> Performing makecell self_chk 30720 >>
RMIN: 0.153545E-03 >> EPS,RMIN: 0.500000E-01 0.153545E-03 >> I/O Error: No String Terminator sent to PRS >> zeroing y: 1 82 -0.75715434E-10 >> I/O Error: No String Terminator sent to PRS >> zeroing y: 4 82 -0.14778734E-11 >> I/O Error: No String Terminator sent to PRS >> zeroing y: 4 83 0.72759583E-10 >> I/O Error: No String Terminator sent to PRS >> zeroing y: 4 123 0.15279482E-09 >> ...... >>
Does it mean my mesh is still not right? >>

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users

_______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Jun 15 03:38:03 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 15 Jun 2017 10:38:03 +0200 Subject: [Nek5000-users] Courant number for passive scalar Message-ID: How does the blow-up manifest itself? Try to run it with 2nd order in time and keep the CFL<0.5. This should be stable assuming there is no source term introducing an additional time step constraint. Does your mesh resolve these steep gradients? Cheers, Stefan -----Original message----- > From:nek5000-users at lists.mcs.anl.gov > Sent: Thursday 15th June 2017 9:29 > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Courant number for passive scalar > > I havent seen the equivalent of a Courant criterion for passive scalars.?? > what I meant is the scalar is intermittent, so one should have a way to detect the regions where gradients are extremely large (fronts) and make sure to decrease the timestep there. >
fronts do not correspond to regions where velocities are especially large, so the classic courant criterion will miss them.?? >
does this make sense ? > any suggestions ? >
agnese >


On Jun 14, 2017, at 2:39 PM, nek5000-users-request at lists.mcs.anl.gov wrote: >
Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

????1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov)
????2. Re: Courant for passiv e scalar ? (nek5000-users at lists.mcs.anl.gov)

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 14, 2017 at 10:44:13 AM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Thanks Jan ! >
so it is confirmed that the tilmestep is based exclusively on the velocity field ? > how do people make sure that the passive scalars do not blow up ??? > anybody ? >
any help is greatly appreciated !! > agnese >

Agnese Seminara > -------------------------------- > CNRS > Institut de Physique de Nice > Parc Valrose?? > avenue J Vallot > 06108 Nice, France > +33 (0) 492 076 775 > http://sites.unice.fr/site/aseminara/
On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote: >
Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

????1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov)
????2. prenek (nek5000-users@ lists.mc s.anl.gov)

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi, > I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk. >
Best, > Jan >
Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov :

Dear Neks >
I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime. >
I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity.?? >
has this already been implemented ? > or how do people avoid this ? >
any suggestion is more than welcome. > thanks > Agnese >


Agnese Seminara > -------------------------------- > CNRS > Institut de Physique de Nice > Parc Valrose?? > avenue J Vallot > 06108 Nice, France > +33 (0) 492 076 775 > http://sites.unice.fr/site/aseminara/
_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



From: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi nek, >
I have a 2D mesh. When I used global refine function, the errors occurred. do you know why? >
??OCT/Multi-SPLIT > quad or multi-split? (q/m) > Number of elements after OctSplit operation: ?? ??7680 > Are you sure you want to split (y/n)? > ??inside gencen ?? ?? ?? ?? 7680 F ?? ?? ?? ?? ?? 4 > ??done gencen ?? ?? ?? ?? 7680 > ??start locglob_lexico: ?? ?? ?? ?? ?? 4 ?? ?? ?? ??7680 ?? ?? ?? 30720 ?? 2.00000009E-03 > ??locglob: ?? ?? ?? ?? ?? 1 ?? ?? ?? ?? ?? 1 ?? ?? ?? 30720 > ??locglob: ?? ?? ?? ?? ?? 2 ?? ?? ?? ??3664 ?? ?? ?? 30720 > ??locglob: ?? ?? ?? ?? ?? 1 ?? ?? ?? ??7761 ?? ?? ?? 30720 > ??locglob: ?? ?? ?? ?? ?? 2 ?? ?? ?? ??7761 ?? ?? ?? 30720 > ??Performing unique_vertex2 self_chk ?? ?? ?? 30720 > ??done locglob_lexico: ?? ?? ??7761 ?? ?? ??7761 ?? ?? 30720 ?? ?? ?? ?? 4 > ??Performing makecell self_chk ?? ?? ?? 30720 >
?? RMIN: 0.153545E-03 > ?? EPS,RMIN: 0.500000E-01 0.153545E-03 > ??I/O Error: No String Terminator sent to PRS > ??zeroing y: ?? ?? 1 ??82 -0.75715434E-10 > ??I/O Error: No String Terminator sent to PRS > ??zeroing y: ?? ?? 4 ??82 -0.14778734E-11 > ??I/O Error: No String Terminator sent to PRS > ??zeroing y: ?? ?? 4 ??83 ??0.72759583E-10 > ??I/O Error: No String Terminator sent to PRS > ??zeroing y: ?? ?? 4 123 ??0.15279482E-09 > ...... >
Does it mean my mesh is still not right? >

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users




From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 14, 2017 at 2:37:35 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


How do you define a courant number for a scalar?

-----Original message-----
From:nek5000-users at lists.mcs.anl.gov
Sent: Wednesday 14th June 2017 10:43
To: nek5000-users at lists.mcs .anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?

Thanks Jan !

so it is confirmed that the tilmestep is based exclusively on the velocity field ?
how do people make sure that the passive scalars do not blow up ???
anybody ?

any help is greatly appreciated !!
agnese


Agnese Seminara
--------------------------------
CNRS
Institut de Physique de Nice
Parc Valrose??
avenue J Vallot
06108 Nice, France
+33 (0) 492 076 775
http://sites.unice.fr/site/aseminara/
On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote:

Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

????1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov)
????2. prenek ( nek5000-users@
lists.mc
s.anl.gov)

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi,
I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk.

Best,
Jan

Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov :

Dear Neks

I am running a simulation of a channel, with a passive scalar. Usi ng diffu sivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime.

I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity.??

has this already been implemented ?
or how do people avoid this ?

any suggestion is more than welcome.
thanks
Agnese



Agnese Seminara
--------------------------------
CNRS
Institut de Physique de Nice
Parc Valrose ??
avenue J Vallot
06108 Nice, France
+33 (0) 492 076 775
http://sites.unice.fr/site/aseminara/
_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



From: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi nek,

I have a 2D mesh. When I used global refine fu nction, the errors occurred. do you know why?

??OCT/Multi-SPLIT
quad or multi-split? (q/m)
Number of elements after OctSplit operation: ?? ??7680
Are you sure you want to split (y/n)?
??inside gencen ?? ?? ?? ?? 7680 F ?? ?? ?? ?? ?? 4
??done gencen ?? ?? ?? ?? 7680
??start locglob_lexico: ?? ?? ?? ?? ?? 4 ?? ?? ?? ??7680 ?? ?? ?? 30720 ?? 2.00000009E-03
??locglob: ?? ?? ?? ?? ?? 1 ?? ?? ?? ?? ?? 1 ?? ?? ?? 30720
??locglob: ?? ?? ?? ?? ?? 2 ?? ?? ?? ??3664 ?? ?? ?? 30720
??locglob: ?? ?? ?? ?? ?? 1 ?? ?? ?? ??7761 ?? ?? ? A0 30720
??locglob: ?? ?? ?? ?? ?? 2 ?? ?? ?? ??7761 ?? ?? ?? 30720
??Performing unique_vertex2 self_chk ?? ?? ?? 30720
??done locglob_lexico: ?? ?? ??7761 ?? ?? ??7761 ?? ?? 30720 ?? ?? ?? ?? 4
??Performing makecell self_chk ?? ?? ?? 30720

?? RMIN: 0.153545E-03
?? EPS,RMIN: 0.500000E-01 0.153545E-03
??I/O Error: No String Terminator sent to PRS
??zeroing y: ?? ?? 1 ??82 -0.75715434E-10
??I/O Error: No String Terminator sent to PRS
??zeroing y: ?? ?? 4 ??82 -0.14778734E-11
??I/O Error: No String Terminator sent to PRS
??zeroing y: ?? ?? 4 ??83 ??0.72759583E-10
?=A 0I/O Error: No String Terminator sent to PRS
??zeroing y: ?? ?? 4 123 ??0.15279482E-09
......

Does it mean my mesh is still not right?


_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo /nek5000 -users

_______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Thu Jun 15 04:08:16 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 15 Jun 2017 09:08:16 +0000 Subject: [Nek5000-users] Periodic box of isotropic turbulence Message-ID: Hi all, Just wondering if there are any examples of simulation of isotropic turbulence in a triply periodic box (using linear forcing or otherwise). I'm hoping to obtain a flow similar to that of https://pdfs.semanticscholar.org/f7b3/c77341d66da230177f6528344a3afd647c2d.pdf which seems to have a relatively simple forcing function proportional to velocity. However, when using F_i = A*u_i in the userf subroutine, I end up with something that isn't isotropic. Any help or advice would be greatly appreciated, Cheers, Lee -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Jun 16 10:31:26 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 16 Jun 2017 15:31:26 +0000 Subject: [Nek5000-users] prenek Message-ID: Hi, If you post your mesh here, I can take a look at it and try to figure out what is going on. Ketan From nek5000-users at lists.mcs.anl.gov Sun Jun 18 03:41:15 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sun, 18 Jun 2017 10:41:15 +0200 Subject: [Nek5000-users] Crash in intp_setup Message-ID: Hi all, my simulations (latest Nek Release) crash when calling intp_setup(): call intp_setup(), tol= 0.100000000000000003E-12 2017-06-18 10:21:11.770 (WARN ) [0x400012691b0] :3452573:ibm.runjob.client.Job: terminated by signal 11 2017-06-18 10:21:11.771 (WARN ) [0x400012691b0] :3452573:ibm.runjob.client.Job: abnormal termination by signal 11 from rank 805 Are there any other parameters i have to set large enough to avoid this? I want to run the interpolation for a large number of points. Earlier mails refer to setting INTP_MAXPTS, but I can?t find this anywhere with grep in the core folder. Jan From nek5000-users at lists.mcs.anl.gov Sun Jun 18 04:00:56 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sun, 18 Jun 2017 11:00:56 +0200 Subject: [Nek5000-users] Crash in intp_setup In-Reply-To: References: Message-ID: Check the hemi example to see how to use the interpolation wrapper. Try to distribute your points for all available ranks to lower the memory footprint per rank. On 18 Jun 2017, at 10:40, "nek5000-users at lists.mcs.anl.gov " > wrote: Hi all, my simulations (latest Nek Release) crash when calling intp_setup(): call intp_setup(), tol= 0.100000000000000003E-12 2017-06-18 10:21:11.770 (WARN ) [0x400012691b0] :3452573:ibm.runjob.client.Job: terminated by signal 11 2017-06-18 10:21:11.771 (WARN ) [0x400012691b0] :3452573:ibm.runjob.client.Job: abnormal termination by signal 11 from rank 805 Are there any other parameters i have to set large enough to avoid this? I want to run the interpolation for a large number of points. Earlier mails refer to setting INTP_MAXPTS, but I can?t find this anywhere with grep in the core folder. Jan _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Sun Jun 18 05:08:41 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sun, 18 Jun 2017 12:08:41 +0200 Subject: [Nek5000-users] Crash in intp_setup In-Reply-To: References: Message-ID: Thanks, I'll have a look. From nek5000-users at lists.mcs.anl.gov Sun Jun 18 15:06:49 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sun, 18 Jun 2017 20:06:49 +0000 Subject: [Nek5000-users] Courant number for passive scalar In-Reply-To: References: Message-ID: Have you tried reducing dt by, say, 2x ? That should be enough... Also, you can try filtering by setting p103=0.05 in the .rea file. The velocity and passive scalars have the same CFL restriction, but velocity is regularized by pressure so it tends to be more stable than high Peclet passive scalar simulations. hth, Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov Sent: Thursday, June 15, 2017 3:38:03 AM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Courant number for passive scalar How does the blow-up manifest itself? Try to run it with 2nd order in time and keep the CFL<0.5. This should be stable assuming there is no source term introducing an additional time step constraint. Does your mesh resolve these steep gradients? Cheers, Stefan -----Original message----- > From:nek5000-users at lists.mcs.anl.gov > Sent: Thursday 15th June 2017 9:29 > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Courant number for passive scalar > > I havent seen the equivalent of a Courant criterion for passive scalars. > what I meant is the scalar is intermittent, so one should have a way to detect the regions where gradients are extremely large (fronts) and make sure to decrease the timestep there. >
fronts do not correspond to regions where velocities are especially large, so the classic courant criterion will miss them. >
does this make sense ? > any suggestions ? >
agnese >


On Jun 14, 2017, at 2:39 PM, nek5000-users-request at lists.mcs.anl.gov wrote: >
Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov)
2. Re: Courant for passiv e scalar ? (nek5000-users at lists.mcs.anl.gov)

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 14, 2017 at 10:44:13 AM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Thanks Jan ! >
so it is confirmed that the tilmestep is based exclusively on the velocity field ? > how do people make sure that the passive scalars do not blow up ? > anybody ? >
any help is greatly appreciated !! > agnese >

Agnese Seminara > -------------------------------- > CNRS > Institut de Physique de Nice > Parc Valrose > avenue J Vallot > 06108 Nice, France > +33 (0) 492 076 775 > http://sites.unice.fr/site/aseminara/
On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote: >
Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov)
2. prenek (nek5000-users@ lists.mc s.anl.gov)

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi, > I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk. >
Best, > Jan >
Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov :

Dear Neks >
I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime. >
I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity. >
has this already been implemented ? > or how do people avoid this ? >
any suggestion is more than welcome. > thanks > Agnese >


Agnese Seminara > -------------------------------- > CNRS > Institut de Physique de Nice > Parc Valrose > avenue J Vallot > 06108 Nice, France > +33 (0) 492 076 775 > http://sites.unice.fr/site/aseminara/
_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



From: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi nek, >
I have a 2D mesh. When I used global refine function, the errors occurred. do you know why? >
OCT/Multi-SPLIT > quad or multi-split? (q/m) > Number of elements after OctSplit operation: 7680 > Are you sure you want to split (y/n)? > inside gencen 7680 F 4 > done gencen 7680 > start locglob_lexico: 4 7680 30720 2.00000009E-03 > locglob: 1 1 30720 > locglob: 2 3664 30720 > locglob: 1 7761 30720 > locglob: 2 7761 30720 > Performing unique_vertex2 self_chk 30720 > done locglob_lexico: 7761 7761 30720 4 > Performing makecell self_chk 30720 >
RMIN: 0.153545E-03 > EPS,RMIN: 0.500000E-01 0.153545E-03 > I/O Error: No String Terminator sent to PRS > zeroing y: 1 82 -0.75715434E-10 > I/O Error: No String Terminator sent to PRS > zeroing y: 4 82 -0.14778734E-11 > I/O Error: No String Terminator sent to PRS > zeroing y: 4 83 0.72759583E-10 > I/O Error: No String Terminator sent to PRS > zeroing y: 4 123 0.15279482E-09 > ...... >
Does it mean my mesh is still not right? >

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users




From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 14, 2017 at 2:37:35 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


How do you define a courant number for a scalar?

-----Original message-----
From:nek5000-users at lists.mcs.anl.gov
Sent: Wednesday 14th June 2017 10:43
To: nek5000-users at lists.mcs .anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?

Thanks Jan !

so it is confirmed that the tilmestep is based exclusively on the velocity field ?
how do people make sure that the passive scalars do not blow up ?
anybody ?

any help is greatly appreciated !!
agnese


Agnese Seminara
--------------------------------
CNRS
Institut de Physique de Nice
Parc Valrose
avenue J Vallot
06108 Nice, France
+33 (0) 492 076 775
http://sites.unice.fr/site/aseminara/
On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote:

Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov)
2. prenek ( nek5000-users@
lists.mc
s.anl.gov)

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi,
I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk.

Best,
Jan

Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov :

Dear Neks

I am running a simulation of a channel, with a passive scalar. Usi ng diffu sivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime.

I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity.

has this already been implemented ?
or how do people avoid this ?

any suggestion is more than welcome.
thanks
Agnese



Agnese Seminara
--------------------------------
CNRS
Institut de Physique de Nice
Parc Valrose
avenue J Vallot
06108 Nice, France
+33 (0) 492 076 775
http://sites.unice.fr/site/aseminara/
_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



From: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi nek,

I have a 2D mesh. When I used global refine fu nction, the errors occurred. do you know why?

OCT/Multi-SPLIT
quad or multi-split? (q/m)
Number of elements after OctSplit operation: 7680
Are you sure you want to split (y/n)?
inside gencen 7680 F 4
done gencen 7680
start locglob_lexico: 4 7680 30720 2.00000009E-03
locglob: 1 1 30720
locglob: 2 3664 30720
locglob: 1 7761 ? A0 30720
locglob: 2 7761 30720
Performing unique_vertex2 self_chk 30720
done locglob_lexico: 7761 7761 30720 4
Performing makecell self_chk 30720

RMIN: 0.153545E-03
EPS,RMIN: 0.500000E-01 0.153545E-03
I/O Error: No String Terminator sent to PRS
zeroing y: 1 82 -0.75715434E-10
I/O Error: No String Terminator sent to PRS
zeroing y: 4 82 -0.14778734E-11
I/O Error: No String Terminator sent to PRS
zeroing y: 4 83 0.72759583E-10
?= 0I/O Error: No String Terminator sent to PRS
zeroing y: 4 123 0.15279482E-09
......

Does it mean my mesh is still not right?


_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo /nek5000 -users

_______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Jun 20 04:04:17 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 20 Jun 2017 11:04:17 +0200 Subject: [Nek5000-users] Courant for passive scalar In-Reply-To: References: Message-ID: Hi Paul and Stephan how does the blow up manifests itself : I didnt dig a lot into the blow up. the flow converges nicely, and once its at steady state I inject the scalar at the inlet through the boundary condition (localized source). it develops a nice turbulent plume, that keeps going for a long time and then suddenly it produces lots of NaN. this picture is consistent with the idea that sometimes passive scalars develop fronts, where abrupt changes occur across a small scale (bachelor scale; for me ~ kolmogorov scale cause diffusivity = viscosity). the classic Courant condition for advection will miss the presence of fronts, cause these are completely uncorrelated to the velocity field. (see eg PRL 84:2385, 2000). i am thinking of including a further check to account for the presence of fronts. but i am sure others have faced similar problems, and so i wanted to check whether someone has already implemented something. in the meantime, i?m using courant 0.25 and see if this is enough, although its clearly suboptimal. cheers Agnese ps. Paul : filtering was already on. > > Have you tried reducing dt by, say, 2x ? > > That should be enough... > > Also, you can try filtering by setting p103=0.05 > in the .rea file. > > The velocity and passive scalars have the same CFL restriction, but velocity is regularized by pressure so it tends to be more stable than high Peclet passive scalar simulations. > > hth, > > Paul > > From: nek5000-users-bounces at lists.mcs.anl.gov > on behalf of nek5000-users at lists.mcs.anl.gov > > Sent: Thursday, June 15, 2017 3:38:03 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Courant number for passive scalar > > How does the blow-up manifest itself? > > Try to run it with 2nd order in time and keep the CFL<0.5. This should be stable assuming there is no source term introducing an additional time step constraint. > > Does your mesh resolve these steep gradients? > > Cheers, > Stefan > > -----Original message----- > > From:nek5000-users at lists.mcs.anl.gov > > > Sent: Thursday 15th June 2017 9:29 > > To: nek5000-users at lists.mcs.anl.gov > > Subject: Re: [Nek5000-users] Courant number for passive scalar > > > > I havent seen the equivalent of a Courant criterion for passive scalars. > > what I meant is the scalar is intermittent, so one should have a way to detect the regions where gradients are extremely large (fronts) and make sure to decrease the timestep there. > >
fronts do not correspond to regions where velocities are especially large, so the classic courant criterion will miss them. > >
does this make sense ? > > any suggestions ? > >
agnese > >


On Jun 14, 2017, at 2:39 PM, nek5000-users-request at lists.mcs.anl.gov > wrote: > >
Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov >

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
class="" />or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov )
2. Re: Courant for passiv > e scalar > ? (nek5000-users at lists.mcs.anl.gov )

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 14, 2017 at 10:44:13 AM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Thanks Jan ! > >
so it is confirmed that the tilmestep is based exclusively on the velocity field ? > > how do people make sure that the passive scalars do not blow up ? > > anybody ? > >
any help is greatly appreciated !! > > agnese > >

Agnese Seminara > > -------------------------------- > > CNRS > > Institut de Physique de Nice > > Parc Valrose > > avenue J Vallot > > 06108 Nice, France > > +33 (0) 492 076 775 > > http://sites.unice.fr/site/aseminara/ >
On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov > wrote: > >
Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov >

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
class="" />or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov )
2. prenek (nek5000-users@ > lists.mc > s.anl.gov )

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi, > > I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk. > >
Best, > > Jan > >
Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov >:

Dear Neks > >
I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime. > >
I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity. > >
has this already been implemented ? > > or how do people avoid this ? > >
any suggestion is more than welcome. > > thanks > > Agnese > >


Agnese Seminara > > -------------------------------- > > CNRS > > Institut de Physique de Nice > > Parc Valrose > > avenue J Vallot > > 06108 Nice, France > > +33 (0) 492 076 775 > > http://sites.unice.fr/site/aseminara/ >
_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov >
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >



From: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi nek, > >
I have a 2D mesh. When I used global refine function, the errors occurred. do you know why? > >
OCT/Multi-SPLIT > > quad or multi-split? (q/m) > > Number of elements after OctSplit operation: 7680 > > Are you sure you want to split (y/n)? > > inside gencen 7680 F 4 > > done gencen 7680 > > start locglob_lexico: 4 7680 30720 2.00000009E-03 > > locglob: 1 1 30720 > > locglob: 2 3664 30720 > > locglob: 1 7761 30720 > > locglob: 2 7761 30720 > > Performing unique_vertex2 self_chk 30720 > > done locglob_lexico: 7761 7761 30720 4 > > Performing makecell self_chk 30720 > >
RMIN: 0.153545E-03 > > EPS,RMIN: 0.500000E-01 0.153545E-03 > > I/O Error: No String Terminator sent to PRS > > zeroing y: 1 82 -0.75715434E-10 > > I/O Error: No String Terminator sent to PRS > > zeroing y: 4 82 -0.14778734E-11 > > I/O Error: No String Terminator sent to PRS > > zeroing y: 4 83 0.72759583E-10 > > I/O Error: No String Terminator sent to PRS > > zeroing y: 4 123 0.15279482E-09 > > ...... > >
Does it mean my mesh is still not right? > >

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users




From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 14, 2017 at 2:37:35 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov >
Reply-To: nek5000-users at lists.mcs.anl.gov


How do you define a courant number for a scalar?

-----Original message-----
From:nek5000-users at lists.mcs.anl.gov >
Sent: Wednesday 14th June 2017 10:43
To: nek5000-users at lists.mcs > .anl.gov >
Subject: Re: [Nek5000-users] Courant for passive scalar ?

Thanks Jan !

so it is confirmed that the tilmestep is based exclusively on the velocity field ?
how do people make sure that the passive scalars do not blow up ?
anybody ?

any help is greatly appreciated !!
agnese


Agnese Seminara
--------------------------------
CNRS
Institut de Physique de Nice
Parc Valrose
avenue J Vallot
06108 Nice, France
+33 (0) 492 076 775
http://sites.unice.fr/site/aseminara/ >
On Jun 13, 2017, at 7:00 PM, nek5000-users-request at lists.mcs.anl.gov < mailto:nek5000-users-request at lists.mcs.a > nl.gov > > wrote:

Send Nek5000-users mailing list submissions to
nek5000-users at lists.mcs.anl.gov >

To subscribe or unsubscribe via the World Wide Web, visit
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
class="" />or, via email, send a message with subject or body 'help' to
nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov )
2. > prenek ( > nek5000-users@
lists.mc
s.anl.gov )

From: nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi,
I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk.

Best,
Jan

Am 12.06.2017 um 18:15 schrieb nek5000-users at lists.mcs.anl.gov >:

Dear Neks

I am running a simulation of a channel, with a passive scalar. Usi > ng diffu > sivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime.

I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity.

has this already been implemented ?
or how do people avoid this ?

any suggestion is more than welcome.
thanks
Agnese



Agnese Seminara
--------------------------------
CNRS
Institut de Physique de Nice
Parc > Valrose >
avenue J Vallot
06108 Nice, France
+33 (0) 492 076 775
http://sites.unice.fr/site/aseminara/ >
_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov >
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >



From: nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To: nek5000-users at lists.mcs.anl.gov
Reply-To: nek5000-users at lists.mcs.anl.gov


Hi nek,

I have a 2D mesh. When I used global refine fu > nction, > the errors occurred. do you know why?

OCT/Multi-SPLIT
quad or multi-split? (q/m)
Number of elements after OctSplit operation: 7680
Are you sure you want to split (y/n)?
inside gencen 7680 F 4
done gencen 7680
start locglob_lexico: 4 7680 30720 2.00000009E-03
locglob: 1 1 30720
locglob: 2 3664 30720
locglob: 1 7761 ? A0 30720 >
locglob: 2 7761 30720
Performing unique_vertex2 self_chk 30720
done locglob_lexico: 7761 7761 30720 4
Performing makecell self_chk 30720

RMIN: 0.153545E-03
EPS,RMIN: 0.500000E-01 0.153545E-03
I/O Error: No String Terminator sent to PRS
zeroing y: 1 82 -0.75715434E-10
I/O Error: No String Terminator sent to PRS
zeroing y: 4 82 -0.14778734E-11
I/O Error: No String Terminator sent to PRS
zeroing y: 4 83 0.72759583E-10
/>?= > 0I/O Error: No String Terminator sent to PRS
zeroing y: 4 123 0.15279482E-09
......

Does it mean my mesh is still not right?


_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo > /nek5000 > -users

_______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Jun 20 06:22:49 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 20 Jun 2017 13:22:49 +0200 Subject: [Nek5000-users] Courant for passive scalar Message-ID: I have seen a similar behavior with passive scalars and LES type resolutions. If your scalar is not well resolved you'll get oscillations driving it crazy. May be that's why you'll see NaNs after a while. Note, the current implementation does not preserve limits and/or monotonicity using techniques artificial viscosity, flux limiting etc. Cheers, Stefan -----Original message----- > From:nek5000-users at lists.mcs.anl.gov > Sent: Tuesday 20th June 2017 11:04 > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Courant for passive scalar > > Hi Paul and Stephan >
how does the blow up manifests itself : I didnt dig a lot into the blow up. the flow converges nicely, and once its at steady state I inject the scalar at the inlet through the boundary condition (localized source). it develops a nice turbulent plume, that keeps going for a long time and then suddenly it produces lots of NaN.?? >
this picture is consistent with the idea that sometimes passive scalars develop fronts, where abrupt changes occur across a small scale (bachelor scale; for me ~ kolmogorov scale cause diffusivity = viscosity). the classic Courant condition for advection will miss ??the presence of fronts, cause these are completely uncorrelated to the velocity field. (see eg PRL 84:2385, 2000).?? >
i am thinking of including a further check to account for the presence of fronts. but i am sure others have faced similar problems, and so i wanted to check whether someone has already implemented something.?? >
in the meantime, i???m using courant 0.25 and see if this is enough, although its clearly suboptimal. ?? >
cheers > Agnese >
ps. Paul : filtering was already on. >

Have you tried reducing dt by, say, 2x ? >
That should be enough... >
Also, you can try filtering by setting p103=0.05 > in the .rea file. >
The velocity and passive scalars have the same CFL restriction, but velocity is regularized by pressure so it tends to be more stable than high Peclet passive scalar simulations. >
hth, >
Paul >
----------- > From:??nek5000-users-bounces at lists.mcs.anl.gov ??> on behalf of??nek5000-users at lists.mcs.anl.gov ??>
Sent:??Thursday, June 15, 2017 3:38:03 AM
To:??nek5000-users at lists.mcs.anl.gov
Subject:??Re: [Nek5000-users] Courant number for passive scalar > How does the blow-up manifest itself?

Try to run it with 2nd order in time and keep the CFL<0.5. This should be stable assuming there is no source term introducing an additional time step constraint.

Does your mesh resolve these steep gradients???

Cheers,
Stefan
??
-----Original message-----
> From:nek5000-users at lists.mcs.anl.gov ??>
> Sent: Thursday 15th June 2017 9:29
> To:??nek5000-users at lists.mcs.anl.gov
> Subject: Re: [Nek5000-users] Courant number for passive scalar
>??
> I havent seen the equivalent of a Courant criterion for passive scalars.??
> wha t I meant is the scalar is intermittent, so one should have a way to detect the regions where gradients are extremely large (fronts) and make sure to decrease the timestep there.
>
fronts do not correspond to regions where velocities are especially large, so the classic courant criterion will miss them.??
>
does this make sense ?
> any suggestions ?
>
agnese
>


On Jun 14, 2017, at 2:39 PM,??nek5000-users-request at lists.mcs.anl.gov ??> wrote:
>
Send Nek5000-users mailing list submissions to
??nek5000-users at lists.mcs.anl.gov =C 2?>

To subscribe or unsubscribe via the World Wide Web, visit
????https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
??class="" />or, via email, send a message with subject or body 'help' to
??????????????nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
??????????????nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

????1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov )
????2. Re: Courant for passiv
??e scalar
?? ? (nek5000-users at lists.mcs.anl.gov )

From:??nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 14, 2017 at 10:44:13 AM GMT+2
To:??nek5000-users at lists.mcs.anl.gov
Reply-To:??nek5000-users at lists.mcs.anl.gov


Thanks Jan !
>
so it is confirmed that the tilmestep is based exclusively on the velocity field ?
> how do people make su re that the passive scalars do not blow up ???
> anybody ?
>
any help is greatly appreciated !!
> agnese
>

Agnese Seminara
> --------------------------------
> CNRS
> Institut de Physique de Nice
> Parc Valrose??
> avenue J Vallot
> 06108 Nice, France
> +33 (0) 492 076 775
>??http://sites.unice.fr/site/aseminara/ ??>
On Jun 13, 2017, at 7:00 PM,??nek5000-users-request at lists.mcs.anl.gov ??> wrote:
>
Send Nek5000-users mailing list submissions to
??nek5000-users at lists.mcs.anl.gov ??>

To subscribe or unsubscribe via the World Wide Web, visit
????https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
??class="" />or, via email, send a message with subject or body 'help' to
??????????????nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
??????????????nek5000-users-owner at lists.mcs.anl.gov

When replying, please edit your S ubject l ine so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

????1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov )
????2. prenek (nek5000-users@
??lists.mc
??s.anl.gov )

From:??nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To:??nek5000-users at lists.mcs.anl.gov
Reply-To:??nek5000-users at lists.mcs.anl.gov


Hi,
> I haven' t tried this myself, but I guess you could wr ite a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk.
>
Best,
> Jan
>
Am 12.06.2017 um 18:15 schrieb??nek5000-users at lists.mcs.anl.gov ??>:

Dear Neks
>
I am running a simulation of a channel, with a passive scalar. Using diffusivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 000 so its in the turbulent regime.
>
I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to a void the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity.??
>
has this already been implemented ?
> or how do people avoid this ?
>
any suggestion is more than welcome.
> thanks
> Agnese
>


Agnese Seminara
> --------------------------------
> CNRS
> Institut de Physique de Nice
> Parc Valrose??
> avenue J Vallot
> 06108 Nice, France
> +33 (0) 492 076 775
>??http://sites.unice.fr/site/aseminara/ ??>
_______________________________________________
Ne k5000-users mailing list
Nek5000-users at lists.mcs.anl.gov ??>
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users ??>



From:??nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To:??nek5000-users at lists.mcs.anl.gov
Reply-To:??nek5000-users at lists.mcs.anl.gov


Hi nek,
>
I have a 2D mesh. When I used global refine function, the errors occurred. do you know why?
>
??OCT/Multi-SPLIT
> quad or multi-split? (q/m)
> Number of elements after OctSplit operation: ?? ??7680
> Are you sure you want to split (y/n)?
> ??inside gencen ?? ?? ?? ?? 7680 F ?? ?? ?? ?? ?? 4
> ??done gencen ?? ?? ?? ?? 7680
> ??start locglob_lexico: ?? ?? ?? ?? ?? 4 ?? ?? ?? ??7680 ?? ?? ?? 30720 ?? 2.00000009E-03
> ??locglob: ?? ?? ?? ?? ?? 1 ?? ?? ?? ?? ?? 1 ?? ?? ?? 30720
> ??locglob: ?? ?? ?? ?? ?? 2 ?? ?? ?? ??3664 ?? ?? ?? 30720
> ??locglob: ? ? ? A0 ?? ?? ?? 1 ?? ?? ?? ??7761 ?? ?? ?? 30720
> ??locglob: ?? ?? ?? ?? ?? 2 ?? ?? ?? ??7761 ?? ?? ?? 30720
> ??Performing unique_vertex2 self_chk ?? ?? ?? 30720
> ??done locglob_lexico: ?? ?? ??7761 ?? ?? ??7761 ?? ?? 30720 ?? ?? ?? ?? 4
> ??Performing makecell self_chk ?? ?? ?? 30720
>
?? RMIN: 0.153545E-03
> ?? EPS,RMIN: 0.500000E-01 0.153545E-03
> ??I/O Error: No String Terminator sent to PRS
> ??zeroing y: ?? ?? 1 ??82 -0.75715434E-10
> ??I/O Error: No String Terminator sent to PRS
> ??zeroing y: ?? ?? 4 ??82 -0.14778734E-11
> ??I/O Error: No String Terminato r sent t o PRS
> ??zeroing y: ?? ?? 4 ??83 ??0.72759583E-10
> ??I/O Error: No String Terminator sent to PRS
> ??zeroing y: ?? ?? 4 123 ??0.15279482E-09
> ......
>
Does it mean my mesh is still not right?
>

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users




From:??nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 14, 2017 at 2:37:35 PM GMT+2
To:??nek5000-users at lists.mcs.anl.gov ??>
Reply-To:??nek5000-users at lists.mcs.anl.gov


How do you define a courant number for a scalar?

-----Original message-----
From:nek5000-users at lists.mcs.anl.gov ??>
Sent: Wednesday 14th June 2017 10:43
To:??nek5000-users at lists.mcs
??.anl.gov
??
Subject: Re: [Nek5000-users] Courant for passive scalar ?

Thanks Jan !

so it is confirme d that t he tilmestep is based exclusively on the velocity field ?
how do people make sure that the passive scalars do not blow up ???
anybody ?

any help is greatly appreciated !!
agnese


Agnese Seminara
--------------------------------
CNRS
Institut de Physique de Nice
Parc Valrose??
avenue J Vallot
06108 Nice, France
+33 (0) 492 076 775
http://sites.unice.fr/site/aseminara/ >
On Jun 13, 2017, at 7:00 PM,??nek5000-users-request at lists.mcs.anl.gov ??< <>mailto:nek5000-users-request at lists.mcs.a
??nl.gov >??
??wrote:

Send Nek5000-users mailing list submissions to
????????????nek5000-users at lists.mcs.anl.gov ??>

To subscribe or unsubscribe via the World Wide Web, visit
??https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
??class="" />or, via email, send a message with subject or body 'help' to
??????????????nek5000-users-request at lists.mcs.anl.gov

You can reach the person managing the list at
???????????? ??ne k5000-users-owner at lists.mcs.anl.gov

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Nek5000-users digest..."
Today's Topics:

????1. Re: Courant for passive scalar ? (nek5000-users at lists.mcs.anl.gov )
????2.??
??prenek (
??nek5000-users@
lists.mc
??s.anl.gov )

From:??nek5000-users at lists.mcs.anl.gov
Subject: Re: [Nek5000-users] Courant for passive scalar ?
Date: June 12, 2017 at 8:00:26 PM GMT+2
To:??nek5000-users at lists.mcs.anl.gov
Rep ly-To:=C 2?nek5000-users at lists.mcs.anl.gov


Hi,
I haven' t tried this myself, but I guess you could write a subroutine that calculates the maximum timestep from the scalar field and sets dt accordingly. Then simple call the routine from userchk.

Best,
Jan

Am 12.06.2017 um 18:15 schrieb??nek5000-users at lists.mcs.anl.gov ??>:

Dear Neks

I am running a simulation of a channel, with a passive scalar. Usi
??ng diffu
??sivity=viscosity, the velocity field converges nicely, whereas the passive scalar blows up at some random point down the simulation. Reynolds number is 10 00 0 so its in the turbulent regime.

I suspect this is due to the fact that the passive scalar is intermittent, so sometimes large fluctuations will occur, totally uncorrelated to velocity fluctuations. to avoid the code to blow up, one would need to setup the timestep based on a courant criterion for the scalar, and not for the velocity.??

has this already been implemented ?
or how do people avoid this ?

any suggestion is more than welcome.
thanks
Agnese



Agnese Seminara
--------------------------------
CNRS
Institut de Physique de Nice
Parc
?? Valrose
????
avenue J Vallot
06108 Nice, France
+33 (0) 492 076 775
http://sites.unice.fr/site/aseminara/ ??>
_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov ??>
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users ??>



From:??nek5000-users at lists.mcs.anl.gov
Subject: [Nek5000-users] prenek
Date: June 13, 2017 at 12:12:49 PM GMT+2
To:??ne k5000-us ers at lists.mcs.anl.gov
Reply-To:??nek5000-users at lists.mcs.anl.gov


Hi nek,

I have a 2D mesh. When I used global refine fu
??nction,??
??the errors occurred. do you know why?

??OCT/Multi-SPLIT
quad or multi-split? (q/m)
Number of elements after OctSplit operation: ?? ??7680
Are you sure you want to split (y/n)?
??inside gencen ?? ?? ?? ?? 7680 F ?? ?? ?? ?? ?? 4
??done gencen ?? ?? ?? ?? 7680
??start locglob_lexico: ?? ?? ?? ?? ?? 4 ?? ?? ?? ??7680 ?? ?? ?? 30720 ?? 2.00000009E-03
?? locglob: ?? ?? ?? ?? ?? 1 ?? ?? ?? ?? ?? 1 ?? ?? ?? 30720
??locglob: ?? ?? ?? ?? ?? 2 ?? ?? ?? ??3664 ?? ?? ?? 30720
??locglob: ?? ?? ?? ?? ?? 1 ?? ?? ?? ??7761 ?? ?? ??? A0 30720
??
??locglob: ?? ?? ?? ?? ?? 2 ?? ?? ?? ??7761 ?? ?? ?? 30720
??Performing unique_vertex2 self_chk ?? ?? ?? 30720
??done locglob_lexico: ?? ?? ??7761 ?? ?? ??7761 ?? ?? 30720 ?? ?? ?? ?? 4
??Performing makecell self_chk ?? ?? ?? 30720

?? RMIN: 0.153545E-03
?? EPS,RMIN: 0.500000E-01 0.153545E-03
??I/O Error: No String Ter minator sent to PRS
??zeroing y: ?? ?? 1 ??82 -0.75715434E-10
??I/O Error: No String Terminator sent to PRS
??zeroing y: ?? ?? 4 ??82 -0.14778734E-11
??I/O Error: No String Terminator sent to PRS
??zeroing y: ?? ?? 4 ??83 ??0.72759583E-10
?? />???=
??0I/O Error: No String Terminator sent to PRS
??zeroing y: ?? ?? 4 123 ??0.15279482E-09
......

Does it mean my mesh is still not right?


_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users

_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https://lists.mcs.anl.gov/mailman/listinfo
??/nek5000
??-users

_______________________________________________
> Nek5000-users mailing list
>??Nek5000-users at lists.mcs.anl.gov
>??https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users


_______________________________________________
Nek5000-users mailing list
Nek5000-users at lists.mcs.anl.gov
https ://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Tue Jun 20 08:18:59 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 20 Jun 2017 15:18:59 +0200 Subject: [Nek5000-users] Interpolation on a finer mesh Message-ID: Dear Neks, I am simulating turbulent pipe flow with Nek5000. As a first step, I have generated a fully developed turbulent pipe flow within a domain of L_z = 5 D at Re_\tau = 360. Now, I would like to use this flow field as an initial / restart condition for a turbulent flow at a Re_\tau=720 with a finer mesh. The geometry of the finer mesh is contained in a .rea file. For doing the interpolation, I have found the "interpolation wrapper for usage in .usr file" in core/intp_usr.f. Can someone give a short example on how to use these subroutines to interpolate results from 3 or 4 restart files onto my new geometry? Kind Regards, Steffen Straub -- Karlsruhe Institute of Technology (KIT) Institute of Fluid Mechanics M.Sc. Steffen Straub Doctoral Researcher Kaiserstra?e 10 Building 10.23 76131 Karlsruhe, Germany Phone: +49 721 608-43027 E-mail: steffen.straub at kit.edu Web: http://www.istm.kit.edu KIT ? The Research University in the Helmholtz Association Since 2010, the KIT has been certified as a family-friendly university. From nek5000-users at lists.mcs.anl.gov Tue Jun 20 08:39:10 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 20 Jun 2017 15:39:10 +0200 Subject: [Nek5000-users] Interpolation on a finer mesh Message-ID: You can use our generic fld reader. Just add this to userchk(): if(istep.eq.0) call gfldr('myfldfilename') This will interpolate your field file data to your new mesh. You're all set :) Cheers, Stefan -----Original message----- > From:nek5000-users at lists.mcs.anl.gov > Sent: Tuesday 20th June 2017 15:19 > To: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] Interpolation on a finer mesh > > Dear Neks, > > I am simulating turbulent pipe flow with Nek5000. As a first step, I > have generated a fully developed turbulent pipe flow within a domain of > L_z = 5 D at Re_\tau = 360. Now, I would like to use this flow field as > an initial / restart condition for a turbulent flow at a Re_\tau=720 > with a finer mesh. > The geometry of the finer mesh is contained in a .rea file. > > > For doing the interpolation, I have found the "interpolation wrapper for > usage in .usr file" in core/intp_usr.f. > Can someone give a short example on how to use these subroutines to > interpolate results from 3 or 4 restart files onto my new geometry? > > > Kind Regards, > Steffen Straub > > > -- > Karlsruhe Institute of Technology (KIT) > Institute of Fluid Mechanics > > M.Sc. Steffen Straub > Doctoral Researcher > > Kaiserstra?e 10 > Building 10.23 > 76131 Karlsruhe, Germany > > Phone: +49 721 608-43027 > E-mail: steffen.straub at kit.edu > Web: http://www.istm.kit.edu > > KIT ? The Research University in the Helmholtz Association > > Since 2010, the KIT has been certified as a family-friendly university. > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > From nek5000-users at lists.mcs.anl.gov Tue Jun 20 08:44:15 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 20 Jun 2017 15:44:15 +0200 Subject: [Nek5000-users] Interpolation on a finer mesh Message-ID: Note: * this feature is only available in the latest MASTER on GitHub * only single .fXXXX fld files are supported * selective restarts (using only a subset of the fld file) is currently not supported Stefan -----Original message----- > From:Stefan Kerkemeier > Sent: Tuesday 20th June 2017 15:39 > To: nek5000-users at lists.mcs.anl.gov > Subject: RE: [Nek5000-users] Interpolation on a finer mesh > > You can use our generic fld reader. > > Just add this to userchk(): > > if(istep.eq.0) call gfldr('myfldfilename') > > This will interpolate your field file data to your new mesh. You're all set :) > > Cheers, > Stefan > > -----Original message----- > > From:nek5000-users at lists.mcs.anl.gov > > Sent: Tuesday 20th June 2017 15:19 > > To: nek5000-users at lists.mcs.anl.gov > > Subject: [Nek5000-users] Interpolation on a finer mesh > > > > Dear Neks, > > > > I am simulating turbulent pipe flow with Nek5000. As a first step, I > > have generated a fully developed turbulent pipe flow within a domain of > > L_z = 5 D at Re_\tau = 360. Now, I would like to use this flow field as > > an initial / restart condition for a turbulent flow at a Re_\tau=720 > > with a finer mesh. > > The geometry of the finer mesh is contained in a .rea file. > > > > > > For doing the interpolation, I have found the "interpolation wrapper for > > usage in .usr file" in core/intp_usr.f. > > Can someone give a short example on how to use these subroutines to > > interpolate results from 3 or 4 restart files onto my new geometry? > > > > > > Kind Regards, > > Steffen Straub > > > > > > -- > > Karlsruhe Institute of Technology (KIT) > > Institute of Fluid Mechanics > > > > M.Sc. Steffen Straub > > Doctoral Researcher > > > > Kaiserstra?e 10 > > Building 10.23 > > 76131 Karlsruhe, Germany > > > > Phone: +49 721 608-43027 > > E-mail: steffen.straub at kit.edu > > Web: http://www.istm.kit.edu > > > > KIT ? The Research University in the Helmholtz Association > > > > Since 2010, the KIT has been certified as a family-friendly university. > > > > _______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > From nek5000-users at lists.mcs.anl.gov Wed Jun 21 04:12:19 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 21 Jun 2017 11:12:19 +0200 Subject: [Nek5000-users] Interpolation on a finer mesh In-Reply-To: References: Message-ID: Hello Stefan, thanks for your quick reply. I have just tested the generic fld reader gfldr as you suggested. It is super easy and works like a charm. Cheers, Steffen On 06/20/2017 07:00 PM, nek5000-users-request at lists.mcs.anl.gov wrote: > Message: 2 Date: Tue, 20 Jun 2017 15:39:10 +0200 From: > nek5000-users at lists.mcs.anl.gov To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] > Interpolation on a finer mesh Message-ID: > > Content-Type: text/plain; charset=utf-8 You can use our generic fld > reader. Just add this to userchk(): if(istep.eq.0) call > gfldr('myfldfilename') This will interpolate your field file data to > your new mesh. You're all set :) Cheers, Stefan -----Original > message----- >> From:nek5000-users at lists.mcs.anl.gov >> Sent: Tuesday 20th June 2017 15:19 >> To:nek5000-users at lists.mcs.anl.gov >> Subject: [Nek5000-users] Interpolation on a finer mesh >> >> Dear Neks, >> >> I am simulating turbulent pipe flow with Nek5000. As a first step, I >> have generated a fully developed turbulent pipe flow within a domain of >> L_z = 5 D at Re_\tau = 360. Now, I would like to use this flow field as >> an initial / restart condition for a turbulent flow at a Re_\tau=720 >> with a finer mesh. >> The geometry of the finer mesh is contained in a .rea file. >> >> >> For doing the interpolation, I have found the "interpolation wrapper for >> usage in .usr file" in core/intp_usr.f. >> Can someone give a short example on how to use these subroutines to >> interpolate results from 3 or 4 restart files onto my new geometry? >> >> >> Kind Regards, >> Steffen Straub -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Jun 21 08:37:20 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 21 Jun 2017 15:37:20 +0200 Subject: [Nek5000-users] Restart using a solution computed on different domain Message-ID: Hello, I have carried out a simulation for the flow past a cylinder at Re=100. The extent of the domain in the is [Lx,Ly,Lz] = [30, 10, 10]; x is the streamwise, y transverse and z is the spanwise direction. I have another computational domain with size [50, 10, 10]. The mesh near the cylinder in the two meshes is identical. If I use the fully developed solution obtained using the first domain as a restart file for computation using the second domain, will the first solution be projected on the second mesh automatically by NEK? Thanks. N. From nek5000-users at lists.mcs.anl.gov Wed Jun 21 08:57:36 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 21 Jun 2017 15:57:36 +0200 Subject: [Nek5000-users] Restart using a solution computed on different domain In-Reply-To: References: Message-ID: You can use the generic fld reader to read the old field file data. This will interpolate all fields onto the new mesh. Not sure what you want to do with the part which is not covered by your old domain. For this area you have to provide somehow a meaningful IC. Cheers Stefan On 21 Jun 2017, at 15:37, "nek5000-users at lists.mcs.anl.gov " > wrote: Hello, ? I have carried out a simulation for the flow past a cylinder at Re=100. The extent of the domain in the is [Lx,Ly,Lz] = [30, 10, 10]; x is the streamwise, y transverse and z is the spanwise direction. I have another computational domain with size [50, 10, 10]. The mesh near the cylinder in the two meshes is identical. If I use the fully developed solution obtained using the first domain as a restart file for computation using the second domain, will the first solution be projected on the second mesh automatically by NEK? Thanks. N. _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Jun 21 09:23:25 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 21 Jun 2017 16:23:25 +0200 Subject: [Nek5000-users] Restart using a solution computed on different domain In-Reply-To: References: Message-ID: Thanks for the reply Stefan. If I do not input any value in the part of domain 2 that is not covered by domain 1, what will be the default value of the flow variables? Would the values be assigned randomly? P.S. As of now I am not much concerned with the initial condition. I just need a fully developed solution at a given Re. I am doing this exercise to quickly reach a fully developed state on the bigger domain. Many thanks, N. On Wednesday 21 June 2017 03:57 PM, nek5000-users at lists.mcs.anl.gov wrote: > You can use the generic fld reader to read the old field file data. > This will interpolate all fields onto the new mesh. Not sure what you > want to do with the part which is not covered by your old domain. For > this area you have to provide somehow a meaningful IC. > > Cheers > Stefan > > On 21 Jun 2017, at 15:37, "nek5000-users at lists.mcs.anl.gov > " > > wrote: > >> Hello, >> >> I have carried out a simulation for the flow past a cylinder at Re=100. >> The extent of the domain in the is [Lx,Ly,Lz] = [30, 10, 10]; x is the >> streamwise, y transverse and z is the spanwise direction. I have another >> computational domain with size [50, 10, 10]. The mesh near the >> cylinder in the two meshes is identical. If I use the fully developed >> solution >> obtained using the first domain as a restart file for computation >> using the second domain, will the first solution be projected on the >> second mesh >> automatically by NEK? >> >> Thanks. >> N. >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> > > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Jun 22 13:13:17 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 22 Jun 2017 20:13:17 +0200 Subject: [Nek5000-users] Numerical treatment of variable viscosity and diffusivity Message-ID: Hi Neks, I am reporting a turbulent pipe case with temperature dependent viscosity and diffusivity and I would like to have a better understanding of the numerical treatment of the momentum and thermal equations when those properties varies. I would highly appreciate if you could indicate a reference (paper/thesis/document) where you describe how the code works in that situation. Thanks in advance. Antonio -- ANTONIO ANTORANZ PERALES Universidad Carlos III de Madrid -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Jun 23 08:12:37 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 23 Jun 2017 13:12:37 +0000 Subject: [Nek5000-users] Help: compute dissipation Message-ID: Hi NEK users, I am trying to calculate dissipation for fully developed turbulent channel flow. I tried the code below, but result I got from this totally disagrees with Vreman?s 2014 paper. x-axis is streamwise, y-axis is wall-normal and z-axis is spanwise. Any errors in the code? Any thoughts? real diss_sqr(lx1,ly1,lz1, lelt) real s11 real s12 real s13 real s22 real s23 real s33 common /plane/ uavg_pl(ly1*lely) $ , vavg_pl(ly1*lely) $ , wavg_pl(ly1*lely) $ , urms_pl(ly1*lely) $ , vrms_pl(ly1*lely) $ , wrms_pl(ly1*lely) $ , uvms_pl(ly1*lely) $ , diss_sqr_pl(ly1*lely) $ , yy(ly1*lely) $ , w1(ly1*lely),w2(ly1*lely) $ , ffx_avg, dragx_avg parameter (lr=lx1*ly1*lz1) common /scruz/ ur(lr),us(lr),ut(lr) $ , vr(lr),vs(lr),vt(lr) $ , wr(lr),ws(lr),wt(lr) common /avg/ uavg(lx1,ly1,lz1,lelv) & , vavg(lx1,ly1,lz1,lelv) & , wavg(lx1,ly1,lz1,lelv) & , urms(lx1,ly1,lz1,lelv) & , vrms(lx1,ly1,lz1,lelv) & , wrms(lx1,ly1,lz1,lelv) & , uvms(lx1,ly1,lz1,lelv) & , dissavg(lx1,ly1,lz1,lelv) common /scrns/ sij (lx1*ly1*lz1,6,lelv) if(icalld.eq.0) then call rzero(uavg,n) call rzero(urms,n) call rzero(vrms,n) call rzero(wrms,n) call rzero(uvms,n) call rzero(dissavg,n) call rzero(wavg,n) call rzero(vavg,n) call rzero(diss_sqr,n) call comp_sij(sij,6,vx,vy,vz,ur,us,ut,vr,vs,vt,wr,ws,wt) do e=1,nelt do i = 1,nx1*ny1*nz1 s11=sij(i,1,e) s12=sij(i,4,e) s22=sij(i,2,e) s13=sij(i,6,e) s23=sij(i,5,e) s33=sij(i,3,e) diss_sqr(i,1,1,e) = (s11*s11) + $ (s22*s22) + $ (s33*s33) + $ (s12*s12)*2 + (s23*s23)*2 + $ (s13*s13)*2 enddo enddo call avg1(dissavg,diss_sqr,alpha,beta,n,'diss',ifverbose) call planar_average_s(diss_sqr_pl,dissavg,w1,w2) do i=1,ny1*nely/2 uavg_pl(i) = 0.5 * (uavg_pl(i) + uavg_pl(m-i+1)) vavg_pl(i) = 0.5 * (vavg_pl(i) + vavg_pl(m-i+1)) wavg_pl(i) = 0.5 * (wavg_pl(i) + wavg_pl(m-i+1)) diss_sqr_pl(i) = 0.5 * (diss_sqr_pl(i) + diss_sqr_pl(m-i+1)) urms_pl(i) = 0.5 * (urms_pl(i) + urms_pl(m-i+1)) vrms_pl(i) = 0.5 * (vrms_pl(i) + vrms_pl(m-i+1)) wrms_pl(i) = 0.5 * (wrms_pl(i) + wrms_pl(m-i+1)) enddo Many thanks. -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Jun 23 09:09:57 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 23 Jun 2017 16:09:57 +0200 Subject: [Nek5000-users] Nek5000 documentation details: Pn-Pn pressure solver In-Reply-To: References: Message-ID: Dear Ananias You say that the coupled Helmholtz solver is used in the case of variable viscosity because in that case the stress tensor is not diagonal. Can you provide some reference about the coupled Helmholtz solver, please. Best, Andrey 2017-06-07 15:47 GMT+02:00 : > Hi Vlad, > > what you say is correct about the treatment of the viscous term in the > case of variable > dynamic viscosity, i.e. the second term in your equation is indeed treated > explicitly, by > extrapolating the velocity using previous time steps (the gradient of mu > is treated implicitly > though as the dynamic viscosity was already updated to its n+1 value). > > The semi-implicit treatment of the full stress tensor in the case of > variable dynamic > viscosity is a fairly recent development (2015), which has not been > published and is not > described in the 1997/1998 papers or anywhere else in more detail. Note > that because of > the implicit treatment of del mu, the inclusion of this explicit term in > the rhs of the pressure > equation is not adding a severe diffusion-like CFL restriction to the time > step (which is > normally related to second order velocity spatial derivatives , i.e. > Laplacian). > > I don't believe further discussion on this topic is of interest to the > majority of Nek users, > so if necessary let's continue any additional conversation off-line. > > All the best, > Ananias > > On Wed, Jun 7, 2017 at 5:51 AM, wrote: > >> Dear Ananias, >> >> thank you for your answer again, but I think, you told about another part >> of viscous term. In 1997 JSP paper it is clearly explained the situation, >> when \mu doesn't depend on temperature. But if I, for example, use the Sutherland's >> law there is another extra term with (\nabla \mu) and first derivations of >> velocity as it is shown on the figure below. >> >> >> I saw in the code and it seems like they are treated explicitly in the >> pressure solver, using the meaning of velocity at n-th time step. Is it so? >> >> Best regards, >> Vlad >> >> >> ?????, 7 ???? 2017, 2:43 +07:00 ?? nek5000-users at lists.mcs.anl.gov: >> >> >> Dear Vlad, >> >> this is correct, the coupled Helmholtz solve is used in the case of the >> full stress tensor >> because in that case the stress tensor is not diagonal. >> >> The splitting approach is based on an irrotational-solenoidal >> decomposition of the velocity >> (which is described in the 1997 JSC paper); the divergence of the former, >> which appears in >> the rhs of the pressure equation is treated implicitly (it is zero in the >> case of constant viscosity >> and incompressible flow), whereas the divergence of the latter is treated >> explicitly through the >> vorticity (which also appears in the pressure rhs and the pressure BC and >> is again zero in the >> case of constant viscosity and incompressible flow; this is not the case >> in the pressure BC) . >> >> It was proved in the JSC and JCP papers that this splitting approach, >> which allows for an >> uncoupled solution of the pressure and velocity equations, leads to a >> high-order overall >> accuracy in time. >> >> Best, >> Ananias >> >> On Tue, Jun 6, 2017 at 7:09 PM, wrote: >> >> Dear Ananias, >> >> thank you for a prompt and clear response! About the coupled Helmholtz >> solver, it is used to solve for three velocity components at once. Is it >> due to \nabla \mu^{n+1} \nabla v^{n+1} term? Thus, in the equation for v_x, >> for example, there are terms with derivatives of v_y and v_z, since they >> are at n+1 time step, they should go to the matrix, and not to the RHS of >> the equation.. Right? >> >> The second issue is that the same term with additional \nabla appears in >> the equation for Laplacian p^{n+1}. Do you treat it explicitly here? I mean >> at the time step n instead of n+1? >> >> Is there no conflict between implicit treatment of viscous terms at the >> `velocity' step while doing it explicitly during `pressure' step? >> >> Best regards, >> Vlad >> >> >> >> ???????, 6 ???? 2017, 17:51 +07:00 ?? nek5000-users at lists.mcs.anl.gov: >> >> >> Dear Vlad, >> the low Mach Pn-Pn approach is based on the 1997 (JSC) and 1997 (JCP) >> papers you mention and it consists of 3 steps as you describe, i.e.: >> a) first the velocity is updated using the extrapolated convective term, >> b) then the Laplacian of pressure is calculated due to convection, after >> that >> c) the velocity is updated using the pressure gradient and accounts for >> viscous term >> The coupled Helmholtz solver is used for the velocities only when using >> ifstrs=true, that >> is when you want to include the full stress tensor. Otherwise, it is >> using separate Helmholtz solves for each of the velocity components, >> similar to Pn-Pn-2. >> Hope this helps clarify things. >> All the best, >> Ananias >> >> >> On Tue, Jun 6, 2017 at 7:29 AM, wrote: >> >> Dear Neks, >> >> reading the documentation I got the impression that Pn-Pn solver (low >> Mach) first solves the pressure where the convective and viscous (!) terms >> are taken into account. After that using this p^{n+1} we solve for velocity >> field. It seems that the algorithm consists of only 2 steps (pressure + >> velocity). >> >> However, reading the paper by Tomboulides, Lee, Orszag (1996) which is >> referenced inside the code, I see the projection algorithm where first the >> velocity is updated using the extrapolated convective term, then the >> Laplacian of pressure is calculated due to convection, after that the >> velocity is updated using convection and pressure gradient. The last step >> accounts for viscous term. >> >> I am a bit confused, could you please help me out here? Which method is >> used? >> >> PS. Another thing is the coupled Helmholtz solver in Pn-Pn. I see that in >> case of Pn-Pn-2 each velocity component is treated separately (segregated >> solver). However, this coupled thing slightly confuses me, why not treating >> it separately as in Pn-Pn-2? Could you please comment there as well? Thank >> you. >> >> Best regards, >> Vlad >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> >> > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: image.png Type: image/png Size: 23241 bytes Desc: not available URL: From nek5000-users at lists.mcs.anl.gov Fri Jun 23 10:17:51 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 23 Jun 2017 15:17:51 +0000 Subject: [Nek5000-users] Help: compute dissipation In-Reply-To: References: Message-ID: Hi, Have you tested your routine on plane Poiseiulle flow, u=1-y^2 ? Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov Sent: Friday, June 23, 2017 8:12:37 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Help: compute dissipation Hi NEK users, I am trying to calculate dissipation for fully developed turbulent channel flow. I tried the code below, but result I got from this totally disagrees with Vreman?s 2014 paper. x-axis is streamwise, y-axis is wall-normal and z-axis is spanwise. Any errors in the code? Any thoughts? real diss_sqr(lx1,ly1,lz1, lelt) real s11 real s12 real s13 real s22 real s23 real s33 common /plane/ uavg_pl(ly1*lely) $ , vavg_pl(ly1*lely) $ , wavg_pl(ly1*lely) $ , urms_pl(ly1*lely) $ , vrms_pl(ly1*lely) $ , wrms_pl(ly1*lely) $ , uvms_pl(ly1*lely) $ , diss_sqr_pl(ly1*lely) $ , yy(ly1*lely) $ , w1(ly1*lely),w2(ly1*lely) $ , ffx_avg, dragx_avg parameter (lr=lx1*ly1*lz1) common /scruz/ ur(lr),us(lr),ut(lr) $ , vr(lr),vs(lr),vt(lr) $ , wr(lr),ws(lr),wt(lr) common /avg/ uavg(lx1,ly1,lz1,lelv) & , vavg(lx1,ly1,lz1,lelv) & , wavg(lx1,ly1,lz1,lelv) & , urms(lx1,ly1,lz1,lelv) & , vrms(lx1,ly1,lz1,lelv) & , wrms(lx1,ly1,lz1,lelv) & , uvms(lx1,ly1,lz1,lelv) & , dissavg(lx1,ly1,lz1,lelv) common /scrns/ sij (lx1*ly1*lz1,6,lelv) if(icalld.eq.0) then call rzero(uavg,n) call rzero(urms,n) call rzero(vrms,n) call rzero(wrms,n) call rzero(uvms,n) call rzero(dissavg,n) call rzero(wavg,n) call rzero(vavg,n) call rzero(diss_sqr,n) call comp_sij(sij,6,vx,vy,vz,ur,us,ut,vr,vs,vt,wr,ws,wt) do e=1,nelt do i = 1,nx1*ny1*nz1 s11=sij(i,1,e) s12=sij(i,4,e) s22=sij(i,2,e) s13=sij(i,6,e) s23=sij(i,5,e) s33=sij(i,3,e) diss_sqr(i,1,1,e) = (s11*s11) + $ (s22*s22) + $ (s33*s33) + $ (s12*s12)*2 + (s23*s23)*2 + $ (s13*s13)*2 enddo enddo call avg1(dissavg,diss_sqr,alpha,beta,n,'diss',ifverbose) call planar_average_s(diss_sqr_pl,dissavg,w1,w2) do i=1,ny1*nely/2 uavg_pl(i) = 0.5 * (uavg_pl(i) + uavg_pl(m-i+1)) vavg_pl(i) = 0.5 * (vavg_pl(i) + vavg_pl(m-i+1)) wavg_pl(i) = 0.5 * (wavg_pl(i) + wavg_pl(m-i+1)) diss_sqr_pl(i) = 0.5 * (diss_sqr_pl(i) + diss_sqr_pl(m-i+1)) urms_pl(i) = 0.5 * (urms_pl(i) + urms_pl(m-i+1)) vrms_pl(i) = 0.5 * (vrms_pl(i) + vrms_pl(m-i+1)) wrms_pl(i) = 0.5 * (wrms_pl(i) + wrms_pl(m-i+1)) enddo Many thanks. -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Jun 23 18:06:02 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 23 Jun 2017 23:06:02 +0000 Subject: [Nek5000-users] Help: compute dissipation In-Reply-To: References: Message-ID: Hi Paul, No, I haven?t tried with Poiseiulle flow. Any reason as to why I should test it on that flow? Thank, Kiran > On 23 Jun 2017, at 16:21, nek5000-users-request at lists.mcs.anl.gov wrote: > > Send Nek5000-users mailing list submissions to > nek5000-users at lists.mcs.anl.gov > > To subscribe or unsubscribe via the World Wide Web, visit > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > or, via email, send a message with subject or body 'help' to > nek5000-users-request at lists.mcs.anl.gov > > You can reach the person managing the list at > nek5000-users-owner at lists.mcs.anl.gov > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Nek5000-users digest..." > > > Today's Topics: > > 1. Re: Help: compute dissipation (nek5000-users at lists.mcs.anl.gov) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Fri, 23 Jun 2017 15:17:51 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > Subject: Re: [Nek5000-users] Help: compute dissipation > Message-ID: > > Content-Type: text/plain; charset="windows-1252" > > > Hi, > > > Have you tested your routine on plane Poiseiulle flow, > > u=1-y^2 ? > > > Paul > > ________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov > Sent: Friday, June 23, 2017 8:12:37 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] Help: compute dissipation > > Hi NEK users, > > I am trying to calculate dissipation for fully developed turbulent channel flow. I tried the code below, but result I got from this totally disagrees with Vreman?s 2014 paper. x-axis is streamwise, y-axis is wall-normal and z-axis is spanwise. Any errors in the code? Any thoughts? > > real diss_sqr(lx1,ly1,lz1, lelt) > real s11 > real s12 > real s13 > real s22 > real s23 > real s33 > > > common /plane/ uavg_pl(ly1*lely) > $ , vavg_pl(ly1*lely) > $ , wavg_pl(ly1*lely) > $ , urms_pl(ly1*lely) > $ , vrms_pl(ly1*lely) > $ , wrms_pl(ly1*lely) > $ , uvms_pl(ly1*lely) > $ , diss_sqr_pl(ly1*lely) > $ , yy(ly1*lely) > $ , w1(ly1*lely),w2(ly1*lely) > $ , ffx_avg, dragx_avg > > > parameter (lr=lx1*ly1*lz1) > common /scruz/ ur(lr),us(lr),ut(lr) > $ , vr(lr),vs(lr),vt(lr) > $ , wr(lr),ws(lr),wt(lr) > > > common /avg/ uavg(lx1,ly1,lz1,lelv) > & , vavg(lx1,ly1,lz1,lelv) > & , wavg(lx1,ly1,lz1,lelv) > & , urms(lx1,ly1,lz1,lelv) > & , vrms(lx1,ly1,lz1,lelv) > & , wrms(lx1,ly1,lz1,lelv) > & , uvms(lx1,ly1,lz1,lelv) > & , dissavg(lx1,ly1,lz1,lelv) > > > common /scrns/ sij (lx1*ly1*lz1,6,lelv) > > if(icalld.eq.0) then > call rzero(uavg,n) > call rzero(urms,n) > call rzero(vrms,n) > call rzero(wrms,n) > call rzero(uvms,n) > call rzero(dissavg,n) > call rzero(wavg,n) > call rzero(vavg,n) > call rzero(diss_sqr,n) > > > call comp_sij(sij,6,vx,vy,vz,ur,us,ut,vr,vs,vt,wr,ws,wt) > > do e=1,nelt > do i = 1,nx1*ny1*nz1 > s11=sij(i,1,e) > s12=sij(i,4,e) > s22=sij(i,2,e) > s13=sij(i,6,e) > s23=sij(i,5,e) > s33=sij(i,3,e) > > diss_sqr(i,1,1,e) = (s11*s11) + > $ (s22*s22) + > $ (s33*s33) + > $ (s12*s12)*2 + (s23*s23)*2 + > $ (s13*s13)*2 > > enddo > enddo > > call avg1(dissavg,diss_sqr,alpha,beta,n,'diss',ifverbose) > > call planar_average_s(diss_sqr_pl,dissavg,w1,w2) > > do i=1,ny1*nely/2 > uavg_pl(i) = 0.5 * (uavg_pl(i) + uavg_pl(m-i+1)) > vavg_pl(i) = 0.5 * (vavg_pl(i) + vavg_pl(m-i+1)) > wavg_pl(i) = 0.5 * (wavg_pl(i) + wavg_pl(m-i+1)) > diss_sqr_pl(i) = 0.5 * (diss_sqr_pl(i) + diss_sqr_pl(m-i+1)) > urms_pl(i) = 0.5 * (urms_pl(i) + urms_pl(m-i+1)) > vrms_pl(i) = 0.5 * (vrms_pl(i) + vrms_pl(m-i+1)) > wrms_pl(i) = 0.5 * (wrms_pl(i) + wrms_pl(m-i+1)) > > enddo > > > > Many thanks. > > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: > > ------------------------------ > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > End of Nek5000-users Digest, Vol 100, Issue 27 > ********************************************** From nek5000-users at lists.mcs.anl.gov Sat Jun 24 21:16:07 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sun, 25 Jun 2017 02:16:07 +0000 Subject: [Nek5000-users] Help: compute dissipation In-Reply-To: References: , Message-ID: Hi Kiran, I always like to test code on something I know the exact answer to... so this would be one such case. Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov Sent: Friday, June 23, 2017 6:06:02 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Help: compute dissipation Hi Paul, No, I haven?t tried with Poiseiulle flow. Any reason as to why I should test it on that flow? Thank, Kiran > On 23 Jun 2017, at 16:21, nek5000-users-request at lists.mcs.anl.gov wrote: > > Send Nek5000-users mailing list submissions to > nek5000-users at lists.mcs.anl.gov > > To subscribe or unsubscribe via the World Wide Web, visit > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > or, via email, send a message with subject or body 'help' to > nek5000-users-request at lists.mcs.anl.gov > > You can reach the person managing the list at > nek5000-users-owner at lists.mcs.anl.gov > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Nek5000-users digest..." > > > Today's Topics: > > 1. Re: Help: compute dissipation (nek5000-users at lists.mcs.anl.gov) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Fri, 23 Jun 2017 15:17:51 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > Subject: Re: [Nek5000-users] Help: compute dissipation > Message-ID: > > Content-Type: text/plain; charset="windows-1252" > > > Hi, > > > Have you tested your routine on plane Poiseiulle flow, > > u=1-y^2 ? > > > Paul > > ________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov > Sent: Friday, June 23, 2017 8:12:37 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] Help: compute dissipation > > Hi NEK users, > > I am trying to calculate dissipation for fully developed turbulent channel flow. I tried the code below, but result I got from this totally disagrees with Vreman?s 2014 paper. x-axis is streamwise, y-axis is wall-normal and z-axis is spanwise. Any errors in the code? Any thoughts? > > real diss_sqr(lx1,ly1,lz1, lelt) > real s11 > real s12 > real s13 > real s22 > real s23 > real s33 > > > common /plane/ uavg_pl(ly1*lely) > $ , vavg_pl(ly1*lely) > $ , wavg_pl(ly1*lely) > $ , urms_pl(ly1*lely) > $ , vrms_pl(ly1*lely) > $ , wrms_pl(ly1*lely) > $ , uvms_pl(ly1*lely) > $ , diss_sqr_pl(ly1*lely) > $ , yy(ly1*lely) > $ , w1(ly1*lely),w2(ly1*lely) > $ , ffx_avg, dragx_avg > > > parameter (lr=lx1*ly1*lz1) > common /scruz/ ur(lr),us(lr),ut(lr) > $ , vr(lr),vs(lr),vt(lr) > $ , wr(lr),ws(lr),wt(lr) > > > common /avg/ uavg(lx1,ly1,lz1,lelv) > & , vavg(lx1,ly1,lz1,lelv) > & , wavg(lx1,ly1,lz1,lelv) > & , urms(lx1,ly1,lz1,lelv) > & , vrms(lx1,ly1,lz1,lelv) > & , wrms(lx1,ly1,lz1,lelv) > & , uvms(lx1,ly1,lz1,lelv) > & , dissavg(lx1,ly1,lz1,lelv) > > > common /scrns/ sij (lx1*ly1*lz1,6,lelv) > > if(icalld.eq.0) then > call rzero(uavg,n) > call rzero(urms,n) > call rzero(vrms,n) > call rzero(wrms,n) > call rzero(uvms,n) > call rzero(dissavg,n) > call rzero(wavg,n) > call rzero(vavg,n) > call rzero(diss_sqr,n) > > > call comp_sij(sij,6,vx,vy,vz,ur,us,ut,vr,vs,vt,wr,ws,wt) > > do e=1,nelt > do i = 1,nx1*ny1*nz1 > s11=sij(i,1,e) > s12=sij(i,4,e) > s22=sij(i,2,e) > s13=sij(i,6,e) > s23=sij(i,5,e) > s33=sij(i,3,e) > > diss_sqr(i,1,1,e) = (s11*s11) + > $ (s22*s22) + > $ (s33*s33) + > $ (s12*s12)*2 + (s23*s23)*2 + > $ (s13*s13)*2 > > enddo > enddo > > call avg1(dissavg,diss_sqr,alpha,beta,n,'diss',ifverbose) > > call planar_average_s(diss_sqr_pl,dissavg,w1,w2) > > do i=1,ny1*nely/2 > uavg_pl(i) = 0.5 * (uavg_pl(i) + uavg_pl(m-i+1)) > vavg_pl(i) = 0.5 * (vavg_pl(i) + vavg_pl(m-i+1)) > wavg_pl(i) = 0.5 * (wavg_pl(i) + wavg_pl(m-i+1)) > diss_sqr_pl(i) = 0.5 * (diss_sqr_pl(i) + diss_sqr_pl(m-i+1)) > urms_pl(i) = 0.5 * (urms_pl(i) + urms_pl(m-i+1)) > vrms_pl(i) = 0.5 * (vrms_pl(i) + vrms_pl(m-i+1)) > wrms_pl(i) = 0.5 * (wrms_pl(i) + wrms_pl(m-i+1)) > > enddo > > > > Many thanks. > > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: > > ------------------------------ > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > End of Nek5000-users Digest, Vol 100, Issue 27 > ********************************************** _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Jun 26 08:49:06 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 26 Jun 2017 14:49:06 +0100 Subject: [Nek5000-users] genmap Message-ID: Hi Neks, Is there anyone know the reason for the following error, in which there are hundred millions of grid points in the geometry. ABORT: NELT>LELM, modify LELM in SIZE and recompile I changed the LELM to be big enough, even equals to 7000000000000, but it doesn't work though. I guess the reasons would be: 1. genmap cannot regonize large elements mesh; 2. I didn't get a recompilation after changing LELM parameter since I don't know how to compile and thought genmap commond would be changed automatically when the SIZE file in /home/bin/genmap was changed. Kind regards, Jian -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Jun 26 10:27:28 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 26 Jun 2017 15:27:28 +0000 Subject: [Nek5000-users] genmap In-Reply-To: References: Message-ID: Dear Jiang, I think you have too many elements. Keep in mind that with the SEM you need about 1/1000th of the number of elements that you would for FEM. What is the largest SEM run you have made so far? Best, Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov Sent: Monday, June 26, 2017 8:49:06 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] genmap Hi Neks, Is there anyone know the reason for the following error, in which there are hundred millions of grid points in the geometry. ABORT: NELT>LELM, modify LELM in SIZE and recompile I changed the LELM to be big enough, even equals to 7000000000000, but it doesn't work though. I guess the reasons would be: 1. genmap cannot regonize large elements mesh; 2. I didn't get a recompilation after changing LELM parameter since I don't know how to compile and thought genmap commond would be changed automatically when the SIZE file in /home/bin/genmap was changed. Kind regards, Jian -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Jun 26 12:06:06 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 26 Jun 2017 18:06:06 +0100 Subject: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 29 In-Reply-To: References: Message-ID: Hi Paul, Now the number of total elements is 576000 written in the rea file. By the way, would you please tell me the reason why only 1/1000 number of elements are needed indeed with SEM. Kind regards, Jian > Message: 2 > Date: Mon, 26 Jun 2017 15:27:28 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > Subject: Re: [Nek5000-users] genmap > Message-ID: > > Content-Type: text/plain; charset="us-ascii" > > > Dear Jiang, > > > I think you have too many elements. Keep in mind that with the SEM you > need about 1/1000th of the number of elements that you would for FEM. > > > What is the largest SEM run you have made so far? > > > Best, Paul > > > ________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov < > nek5000-users-bounces at lists.mcs.anl.gov> on behalf of > nek5000-users at lists.mcs.anl.gov > Sent: Monday, June 26, 2017 8:49:06 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] genmap > > Hi Neks, > > Is there anyone know the reason for the following error, in which there > are hundred millions of grid points in the geometry. > > ABORT: NELT>LELM, modify LELM in SIZE and recompile > > I changed the LELM to be big enough, even equals to 7000000000000, but it > doesn't work though. I guess the reasons would be: > > 1. genmap cannot regonize large elements mesh; > 2. I didn't get a recompilation after changing LELM parameter since I > don't know how to compile and thought genmap commond would be changed > automatically when the SIZE file in /home/bin/genmap was changed. > > Kind regards, > > Jian > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: attachments/20170626/74ae123f/attachment-0001.html> > > ------------------------------ > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > End of Nek5000-users Digest, Vol 100, Issue 29 > ********************************************** > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Jun 26 14:18:22 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 26 Jun 2017 19:18:22 +0000 Subject: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 29 In-Reply-To: References: , Message-ID: Dear Jiang, If you run with lx1=10 you have 1000 points per element. FEM never runs with 10th order - the SEM is meant to do this. What is the largest SEM problem you have run so far? I recommend starting with a sequence of small ones just to get used to the workflow. Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov Sent: Monday, June 26, 2017 12:06:06 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 29 Hi Paul, Now the number of total elements is 576000 written in the rea file. By the way, would you please tell me the reason why only 1/1000 number of elements are needed indeed with SEM. Kind regards, Jian Message: 2 Date: Mon, 26 Jun 2017 15:27:28 +0000 From: nek5000-users at lists.mcs.anl.gov To: "nek5000-users at lists.mcs.anl.gov" > Subject: Re: [Nek5000-users] genmap Message-ID: > Content-Type: text/plain; charset="us-ascii" Dear Jiang, I think you have too many elements. Keep in mind that with the SEM you need about 1/1000th of the number of elements that you would for FEM. What is the largest SEM run you have made so far? Best, Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov > on behalf of nek5000-users at lists.mcs.anl.gov > Sent: Monday, June 26, 2017 8:49:06 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] genmap Hi Neks, Is there anyone know the reason for the following error, in which there are hundred millions of grid points in the geometry. ABORT: NELT>LELM, modify LELM in SIZE and recompile I changed the LELM to be big enough, even equals to 7000000000000, but it doesn't work though. I guess the reasons would be: 1. genmap cannot regonize large elements mesh; 2. I didn't get a recompilation after changing LELM parameter since I don't know how to compile and thought genmap commond would be changed automatically when the SIZE file in /home/bin/genmap was changed. Kind regards, Jian -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users End of Nek5000-users Digest, Vol 100, Issue 29 ********************************************** -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Jun 27 19:06:09 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 28 Jun 2017 01:06:09 +0100 Subject: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 30 In-Reply-To: References: Message-ID: Hi Paul, I changed the elements in the straight pipe which is I am doing at the moment. I set the horizontal and vertical axis as 14 spectral elements in the quarter-section for the simulation at Re=5300 based on bulk velocity. And the pipe length is 25R. Do you think it's enough to run the simulation? By the way, I run the simulation for a short time, after which I used visit to check the mesh finding that there are several things inside the pipe which are likely to be walls. And I found the mesh in visit shows that the cross-sectional views are different in the outlet and inlet which is open or closed. I set the axis view in visit to not show the axis. Kind regards, Jian On 27 June 2017 at 18:00, wrote: > Send Nek5000-users mailing list submissions to > nek5000-users at lists.mcs.anl.gov > > To subscribe or unsubscribe via the World Wide Web, visit > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > or, via email, send a message with subject or body 'help' to > nek5000-users-request at lists.mcs.anl.gov > > You can reach the person managing the list at > nek5000-users-owner at lists.mcs.anl.gov > > When replying, please edit your Subject line so it is more specific > than "Re: Contents of Nek5000-users digest..." > > > Today's Topics: > > 1. Re: Nek5000-users Digest, Vol 100, Issue 29 > (nek5000-users at lists.mcs.anl.gov) > 2. Re: Nek5000-users Digest, Vol 100, Issue 29 > (nek5000-users at lists.mcs.anl.gov) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Mon, 26 Jun 2017 18:06:06 +0100 > From: nek5000-users at lists.mcs.anl.gov > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 29 > Message-ID: > > Content-Type: text/plain; charset="utf-8" > > Hi Paul, > > Now the number of total elements is 576000 written in the rea file. By the > way, would you please tell me the reason why only 1/1000 number of elements > are needed indeed with SEM. > > Kind regards, > > Jian > > > > Message: 2 > > Date: Mon, 26 Jun 2017 15:27:28 +0000 > > From: nek5000-users at lists.mcs.anl.gov > > To: "nek5000-users at lists.mcs.anl.gov" > > > > Subject: Re: [Nek5000-users] genmap > > Message-ID: > > > > Content-Type: text/plain; charset="us-ascii" > > > > > > Dear Jiang, > > > > > > I think you have too many elements. Keep in mind that with the SEM you > > need about 1/1000th of the number of elements that you would for FEM. > > > > > > What is the largest SEM run you have made so far? > > > > > > Best, Paul > > > > > > ________________________________ > > From: nek5000-users-bounces at lists.mcs.anl.gov < > > nek5000-users-bounces at lists.mcs.anl.gov> on behalf of > > nek5000-users at lists.mcs.anl.gov > > Sent: Monday, June 26, 2017 8:49:06 AM > > To: nek5000-users at lists.mcs.anl.gov > > Subject: [Nek5000-users] genmap > > > > Hi Neks, > > > > Is there anyone know the reason for the following error, in which there > > are hundred millions of grid points in the geometry. > > > > ABORT: NELT>LELM, modify LELM in SIZE and recompile > > > > I changed the LELM to be big enough, even equals to 7000000000000, but it > > doesn't work though. I guess the reasons would be: > > > > 1. genmap cannot regonize large elements mesh; > > 2. I didn't get a recompilation after changing LELM parameter since I > > don't know how to compile and thought genmap commond would be changed > > automatically when the SIZE file in /home/bin/genmap was changed. > > > > Kind regards, > > > > Jian > > -------------- next part -------------- > > An HTML attachment was scrubbed... > > URL: > attachments/20170626/74ae123f/attachment-0001.html> > > > > ------------------------------ > > > > _______________________________________________ > > Nek5000-users mailing list > > Nek5000-users at lists.mcs.anl.gov > > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > > > > End of Nek5000-users Digest, Vol 100, Issue 29 > > ********************************************** > > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: attachments/20170626/a406c1da/attachment-0001.html> > > ------------------------------ > > Message: 2 > Date: Mon, 26 Jun 2017 19:18:22 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > Subject: Re: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 29 > Message-ID: > > Content-Type: text/plain; charset="us-ascii" > > > Dear Jiang, > > > If you run with lx1=10 you have 1000 points per element. > > > FEM never runs with 10th order - the SEM is meant to do this. > > > What is the largest SEM problem you have run so far? I recommend starting > with a sequence of small ones just to get used to the workflow. > > > Paul > > > ________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov < > nek5000-users-bounces at lists.mcs.anl.gov> on behalf of > nek5000-users at lists.mcs.anl.gov > Sent: Monday, June 26, 2017 12:06:06 PM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 29 > > Hi Paul, > > Now the number of total elements is 576000 written in the rea file. By the > way, would you please tell me the reason why only 1/1000 number of elements > are needed indeed with SEM. > > Kind regards, > > Jian > > Message: 2 > Date: Mon, 26 Jun 2017 15:27:28 +0000 > From: nek5000-users at lists.mcs.anl.gov lists.mcs.anl.gov> > To: "nek5000-users at lists.mcs.anl.gov lists.mcs.anl.gov>" > lists.mcs.anl.gov>> > Subject: Re: [Nek5000-users] genmap > Message-ID: > > > Content-Type: text/plain; charset="us-ascii" > > > Dear Jiang, > > > I think you have too many elements. Keep in mind that with the SEM you > need about 1/1000th of the number of elements that you would for FEM. > > > What is the largest SEM run you have made so far? > > > Best, Paul > > > ________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov users-bounces at lists.mcs.anl.gov> > on behalf of > nek5000-users at lists.mcs.anl.gov < > nek5000-users at lists.mcs.anl.gov> > Sent: Monday, June 26, 2017 8:49:06 AM > To: nek5000-users at lists.mcs.anl.gov > > Subject: [Nek5000-users] genmap > > Hi Neks, > > Is there anyone know the reason for the following error, in which there > are hundred millions of grid points in the geometry. > > ABORT: NELT>LELM, modify LELM in SIZE and recompile > > I changed the LELM to be big enough, even equals to 7000000000000, but it > doesn't work though. I guess the reasons would be: > > 1. genmap cannot regonize large elements mesh; > 2. I didn't get a recompilation after changing LELM parameter since I > don't know how to compile and thought genmap commond would be changed > automatically when the SIZE file in /home/bin/genmap was changed. > > Kind regards, > > Jian > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: attachments/20170626/74ae123f/attachment-0001.html> > > ------------------------------ > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > End of Nek5000-users Digest, Vol 100, Issue 29 > ********************************************** > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: attachments/20170626/ccc11744/attachment-0001.html> > > ------------------------------ > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > End of Nek5000-users Digest, Vol 100, Issue 30 > ********************************************** > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Jun 28 00:08:07 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 28 Jun 2017 05:08:07 +0000 Subject: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 30 In-Reply-To: References: , Message-ID: Hi Jian, Have a look at the "pipe" example. I would expect the total element count to be about 10,000-30,000 for well resolved pipe flow at Re=5300. There is a README on how to construct the mesh. Please let me know if that helps... Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov on behalf of nek5000-users at lists.mcs.anl.gov Sent: Tuesday, June 27, 2017 7:06:09 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 30 Hi Paul, I changed the elements in the straight pipe which is I am doing at the moment. I set the horizontal and vertical axis as 14 spectral elements in the quarter-section for the simulation at Re=5300 based on bulk velocity. And the pipe length is 25R. Do you think it's enough to run the simulation? By the way, I run the simulation for a short time, after which I used visit to check the mesh finding that there are several things inside the pipe which are likely to be walls. And I found the mesh in visit shows that the cross-sectional views are different in the outlet and inlet which is open or closed. I set the axis view in visit to not show the axis. Kind regards, Jian On 27 June 2017 at 18:00, > wrote: Send Nek5000-users mailing list submissions to nek5000-users at lists.mcs.anl.gov To subscribe or unsubscribe via the World Wide Web, visit https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users or, via email, send a message with subject or body 'help' to nek5000-users-request at lists.mcs.anl.gov You can reach the person managing the list at nek5000-users-owner at lists.mcs.anl.gov When replying, please edit your Subject line so it is more specific than "Re: Contents of Nek5000-users digest..." Today's Topics: 1. Re: Nek5000-users Digest, Vol 100, Issue 29 (nek5000-users at lists.mcs.anl.gov) 2. Re: Nek5000-users Digest, Vol 100, Issue 29 (nek5000-users at lists.mcs.anl.gov) ---------------------------------------------------------------------- Message: 1 Date: Mon, 26 Jun 2017 18:06:06 +0100 From: nek5000-users at lists.mcs.anl.gov To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 29 Message-ID: > Content-Type: text/plain; charset="utf-8" Hi Paul, Now the number of total elements is 576000 written in the rea file. By the way, would you please tell me the reason why only 1/1000 number of elements are needed indeed with SEM. Kind regards, Jian > Message: 2 > Date: Mon, 26 Jun 2017 15:27:28 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > > Subject: Re: [Nek5000-users] genmap > Message-ID: > > > Content-Type: text/plain; charset="us-ascii" > > > Dear Jiang, > > > I think you have too many elements. Keep in mind that with the SEM you > need about 1/1000th of the number of elements that you would for FEM. > > > What is the largest SEM run you have made so far? > > > Best, Paul > > > ________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov < > nek5000-users-bounces at lists.mcs.anl.gov> on behalf of > nek5000-users at lists.mcs.anl.gov > > Sent: Monday, June 26, 2017 8:49:06 AM > To: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] genmap > > Hi Neks, > > Is there anyone know the reason for the following error, in which there > are hundred millions of grid points in the geometry. > > ABORT: NELT>LELM, modify LELM in SIZE and recompile > > I changed the LELM to be big enough, even equals to 7000000000000, but it > doesn't work though. I guess the reasons would be: > > 1. genmap cannot regonize large elements mesh; > 2. I didn't get a recompilation after changing LELM parameter since I > don't know how to compile and thought genmap commond would be changed > automatically when the SIZE file in /home/bin/genmap was changed. > > Kind regards, > > Jian > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: attachments/20170626/74ae123f/attachment-0001.html> > > ------------------------------ > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > > End of Nek5000-users Digest, Vol 100, Issue 29 > ********************************************** > -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ Message: 2 Date: Mon, 26 Jun 2017 19:18:22 +0000 From: nek5000-users at lists.mcs.anl.gov To: "nek5000-users at lists.mcs.anl.gov" > Subject: Re: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 29 Message-ID: > Content-Type: text/plain; charset="us-ascii" Dear Jiang, If you run with lx1=10 you have 1000 points per element. FEM never runs with 10th order - the SEM is meant to do this. What is the largest SEM problem you have run so far? I recommend starting with a sequence of small ones just to get used to the workflow. Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov > on behalf of nek5000-users at lists.mcs.anl.gov > Sent: Monday, June 26, 2017 12:06:06 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Nek5000-users Digest, Vol 100, Issue 29 Hi Paul, Now the number of total elements is 576000 written in the rea file. By the way, would you please tell me the reason why only 1/1000 number of elements are needed indeed with SEM. Kind regards, Jian Message: 2 Date: Mon, 26 Jun 2017 15:27:28 +0000 From: nek5000-users at lists.mcs.anl.gov> To: "nek5000-users at lists.mcs.anl.gov>" >> Subject: Re: [Nek5000-users] genmap Message-ID: >> Content-Type: text/plain; charset="us-ascii" Dear Jiang, I think you have too many elements. Keep in mind that with the SEM you need about 1/1000th of the number of elements that you would for FEM. What is the largest SEM run you have made so far? Best, Paul ________________________________ From: nek5000-users-bounces at lists.mcs.anl.gov> >> on behalf of nek5000-users at lists.mcs.anl.gov> >> Sent: Monday, June 26, 2017 8:49:06 AM To: nek5000-users at lists.mcs.anl.gov> Subject: [Nek5000-users] genmap Hi Neks, Is there anyone know the reason for the following error, in which there are hundred millions of grid points in the geometry. ABORT: NELT>LELM, modify LELM in SIZE and recompile I changed the LELM to be big enough, even equals to 7000000000000, but it doesn't work though. I guess the reasons would be: 1. genmap cannot regonize large elements mesh; 2. I didn't get a recompilation after changing LELM parameter since I don't know how to compile and thought genmap commond would be changed automatically when the SIZE file in /home/bin/genmap was changed. Kind regards, Jian -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users End of Nek5000-users Digest, Vol 100, Issue 29 ********************************************** -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users End of Nek5000-users Digest, Vol 100, Issue 30 ********************************************** -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Jun 28 17:12:31 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 28 Jun 2017 22:12:31 +0000 Subject: [Nek5000-users] calculate gradients of fluctuation Message-ID: Hi nek users, For u=U + u? where U= mean streamwise velocity and u?= fluctuation of streamwise velocity I know how to calculate du/dx , du/dy and du/dz. But how can I calculate gradients of the fluctuation of streamwise veloctity, i.e. du?/dx , du?/dy , du?/dz ? Thank you. -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Jun 29 10:26:18 2017 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 29 Jun 2017 08:26:18 -0700 Subject: [Nek5000-users] calculate gradients of fluctuation In-Reply-To: References: Message-ID: It depends, do you want it on the fly (while running nek5000) or in the post-processing. If you want it in the post-processing, first call avg_all() routine in usrchk, this way you can calculate the temporal average \bar{vx}, \bar{vy}, \bar{vz}, which will be dumped as separate field files in the running directory. Run nek5000 in the post-processing mode, load the snapshot as well as the average field files in usrchk() routine and subtract avg from the snapshots, vx - \bar{vx}. In this way you will get the fluctuations. Then you can use gradm1() to calculate the gradients of du'/dx. du'/dy and du'/dz. You can refer to the Nek5000 user guide to see how nek5000 is run in post-processing mode. If you want to calculate it on the fly, it is a little tricky, You cannot resort to time-average, but you can use some other meaningful average that would to some extent approximate the effects of temporal averaging. For example, in channel flow, you can use horizontal average, where streamwise-spanwise direction is periodic or in some other flows you can choose to use a volume average. The idea is to subtract, an average of (vx) from vx to get the fluctuations and then call the gradm1() functions. It really depends on the geometry and the type of fluid-problem you are solving, but the idea is the same. On Wed, Jun 28, 2017 at 3:12 PM, wrote: > Hi nek users, > > For u=U + u? where U= mean streamwise velocity and u?= fluctuation of > streamwise velocity > > I know how to calculate du/dx , du/dy and du/dz. But how can I calculate > gradients of the *fluctuation* of streamwise veloctity, i.e. du?/dx , > du?/dy , du?/dz ? > > Thank you. > > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > > -------------- next part -------------- An HTML attachment was scrubbed... URL: