From nek5000-users at lists.mcs.anl.gov Tue Dec 1 01:38:39 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 1 Dec 2015 08:38:39 +0100 Subject: [Nek5000-users] Sharing parameters between processors Message-ID: Hi Paul, Thanks for your quick answer! These are random numbers I use. They is not a very large number of them, I have about 100 elements in my array. However,I need to use the same on all my domain, but as they are random, I can only generate them once. Oscar On 01/12/2015 03:47, nek5000-users at lists.mcs.anl.gov wrote : Hi Oscar, Is there a reason why your quantity is modified only on one processor, and not on the others? How many values are you working with? 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:* Monday, November 30, 2015 4:31 PM *To:* nek5000-users at lists.mcs.anl.gov *Subject:* [Nek5000-users] Sharing parameters between processors Hi Nekers, In my current simulation, I am defining an array containing parameters that are shared between several processors. I have tried to do this in the usercheck subroutine, with a COMMON statement. However, the COMMON statement enables saving the parameters between two successive calls of the usercheck by the same processor, but does not enable to share a variable between two processor. In other words, if I modify the variable a on the processor 1, the variable a on the processor 2 will not be modified, which is not what I want. Any idea on how to do that? Thanks for your help, Oscar Valdorez -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Dec 1 02:44:07 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 1 Dec 2015 09:44:07 +0100 Subject: [Nek5000-users] Stress formulation Message-ID: Hi Neks, I have one question about stress formulation in NEK5000. I tried to make a simulation for Poiseuil flow with stress formulation. In inlet boundary condition use was made of parabolic profile. If IFSTRS=false the outlet profile is exactly the same as inlet one but in case of IFSTRS=true the outlet profile changes significantly (both the maximum value and profile shape). Could you tell me how I can get the parabolic profile at outlet, please? The same thing occurs with pressure values. The maximum pressure value with IFSTRS=true is three times higher then the one with IFSTRS=false. I found that the difference comes from subroutines where the stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for IFSTRS = true) but I don't undestand what is happening exactly. Thanks in advance for your help, Best regards, Andrew -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Dec 1 04:37:54 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 1 Dec 2015 11:37:54 +0100 Subject: [Nek5000-users] Sharing parameters between processors In-Reply-To: References: Message-ID: For that purpose you could also use a pseudo-random-numbmer generator that produces the sae numbers on each processor (i.e. uses the same seed). Otherwise, use what I wrote earlier, namely a bcast operation. That is one line in the code assuming the data is produced on processor 0. philipp On 2015-12-01 08:38, nek5000-users at lists.mcs.anl.gov wrote: > Hi Paul, > > Thanks for your quick answer! These are random numbers I use. They is > not a very large number of them, I have about 100 elements in my array. > However,I need to use the same on all my domain, but as they are random, > I can only generate them once. > > Oscar > > On 01/12/2015 03:47, nek5000-users at lists.mcs.anl.gov > wrote : >> >> Hi Oscar, >> >> Is there a reason why your quantity is modified only on one processor, >> and not on the others? >> >> How many values are you working with? >> >> 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:* Monday, November 30, 2015 4:31 PM >> *To:* nek5000-users at lists.mcs.anl.gov >> >> *Subject:* [Nek5000-users] Sharing parameters between processors >> >> Hi Nekers, >> >> In my current simulation, I am defining an array containing parameters >> that are shared between several processors. I have tried to do this in >> the usercheck subroutine, with a COMMON statement. However, the COMMON >> statement enables saving the parameters between two successive calls >> of the usercheck by the same processor, but does not enable to share a >> variable between two processor. In other words, if I modify the >> variable a on the processor 1, the variable a on the processor 2 will >> not be modified, which is not what I want. Any idea on how to do that? >> >> Thanks for your help, >> >> Oscar Valdorez >> > > > _______________________________________________ > 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 Dec 1 05:48:42 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 1 Dec 2015 12:48:42 +0100 Subject: [Nek5000-users] Sharing parameters between processors Message-ID: OK thanks, I will try that! Oscar On 01/12/2015 11:37, nek5000-users at lists.mcs.anl.gov wrote : For that purpose you could also use a pseudo-random-numbmer generator that produces the sae numbers on each processor (i.e. uses the same seed). Otherwise, use what I wrote earlier, namely a bcast operation. That is one line in the code assuming the data is produced on processor 0. philipp On 2015-12-01 08:38, nek5000-users at lists.mcs.anl.gov wrote: Hi Paul, Thanks for your quick answer! These are random numbers I use. They is not a very large number of them, I have about 100 elements in my array. However,I need to use the same on all my domain, but as they are random, I can only generate them once. Oscar On 01/12/2015 03:47, nek5000-users at lists.mcs.anl.gov wrote : Hi Oscar, Is there a reason why your quantity is modified only on one processor, and not on the others? How many values are you working with? 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:* Monday, November 30, 2015 4:31 PM *To:* nek5000-users at lists.mcs.anl.gov *Subject:* [Nek5000-users] Sharing parameters between processors Hi Nekers, In my current simulation, I am defining an array containing parameters that are shared between several processors. I have tried to do this in the usercheck subroutine, with a COMMON statement. However, the COMMON statement enables saving the parameters between two successive calls of the usercheck by the same processor, but does not enable to share a variable between two processor. In other words, if I modify the variable a on the processor 1, the variable a on the processor 2 will not be modified, which is not what I want. Any idea on how to do that? Thanks for your help, Oscar Valdorez _______________________________________________ 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 Dec 1 05:26:37 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 01 Dec 2015 13:26:37 +0200 Subject: [Nek5000-users] mpi output In-Reply-To: References: Message-ID: Yes there is. Check the mpi-io based wrappers in byte_mpi.f. Cheers, Stefan Von: Antworten an: Datum: Monday, November 30, 2015 at 2:36 PM An: Betreff: [Nek5000-users] mpi output Hi Neks, I should write a vector of real values in a file. I state that I do not know so much about mpi, Is there an mpi function in Nek that allow me to write it in parallel? sincerely Alessandro _______________________________________________ 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 Dec 1 08:49:15 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 1 Dec 2015 09:49:15 -0500 Subject: [Nek5000-users] Sharing parameters between processors In-Reply-To: References: Message-ID: Oscar, In general, you need to use MPI routines to do that sort of thing. There are a bunch of NEK subroutines which are "wrappers" for the underlying MPI routines, and properly set them up, pass the variables, etc... They are pretty easy to use, but you do have to read up a little about how the MPI routines work to understand the context. I have some .usr code where I am counting grid points during a specific number of time steps, holding them in an array at each rank, then adding them up at a certain time step. I am using the "gop()" subroutine to do this; it implements MPI_ALLREDUCE. There are other routines which find the min, max, etc... Look at comm_mpi.f for some ideas; sounds like you need some simple broadcast function, something using MPI_bcast... See Ya, Murph -- Murphy Leo O'Dea PhD Candidate Oakland University mlodea at oakland.edu --------------------------------------------------------------------------------------------------------------------------------------- On 12/1/2015 6:49 AM, nek5000-users-request at lists.mcs.anl.gov wrote: Indeed, common blocks only work with the local memory. The easiest way is, if it is only a few parameters, that you put them into the rea file in the unused section (say about index 120 or so). Then the values are immediately available on each processor. Otherwise you have to do explicit communication. The starting point would be the broadcast operation, bcast(buf,len), which shares len reals stored on process 0 with all other processes. bcast is a wrapper to the standard MPI function call mpi_bcast (buf,len,mpi_byte,0,nekcomm,ierr) Hope this helps, Philipp On 2015-11-30 23:31, nek5000-users at lists.mcs.anl.gov wrote: Hi Nekers, In my current simulation, I am defining an array containing parameters that are shared between several processors. I have tried to do this in the usercheck subroutine, with a COMMON statement. However, the COMMON statement enables saving the parameters between two successive calls of the usercheck by the same processor, but does not enable to share a variable between two processor. In other words, if I modify the variable a on the processor 1, the variable a on the processor 2 will not be modified, which is not what I want. Any idea on how to do that? Thanks for your help, Oscar Valdorez From nek5000-users at lists.mcs.anl.gov Tue Dec 1 11:00:52 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 1 Dec 2015 18:00:52 +0100 Subject: [Nek5000-users] mpi output In-Reply-To: References: Message-ID: thanks a lot Stefan! sincerely Alessandro 2015-12-01 12:26 GMT+01:00 : > Yes there is. Check the mpi-io based wrappers in byte_mpi.f. > > Cheers, > Stefan > > Von: > Antworten an: > Datum: Monday, November 30, 2015 at 2:36 PM > An: > Betreff: [Nek5000-users] mpi output > > Hi Neks, > > I should write a vector of real values in a file. I state that I do not > know so much about mpi, Is there an mpi function in Nek that allow me to > write it in parallel? > > sincerely > Alessandro > _______________________________________________ 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 Dec 1 11:02:13 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 1 Dec 2015 18:02:13 +0100 Subject: [Nek5000-users] find a point Message-ID: Hi Neks, Given a point P(x,y,z) (even not grid point), is there in Nek a fast way to found in which element it is located and also the corresponding processor? sincerely Alessandro -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Dec 1 10:35:43 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 01 Dec 2015 18:35:43 +0200 Subject: [Nek5000-users] find a point In-Reply-To: References: Message-ID: Yes, the routines in findpts.c are your friends. Have a look at hpts() to see how it works. Cheers, Stefan Von: Antworten an: Datum: Tuesday, December 1, 2015 at 7:02 PM An: Betreff: [Nek5000-users] find a point Hi Neks, Given a point P(x,y,z) (even not grid point), is there in Nek a fast way to found in which element it is located and also the corresponding processor? sincerely Alessandro _______________________________________________ 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 Dec 1 11:39:11 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 1 Dec 2015 18:39:11 +0100 Subject: [Nek5000-users] find a point In-Reply-To: References: Message-ID: in postpro.f there is a Fortran routine hpts which acts as a driver to the various hpts... routines. I guess from reading that subroutine you can figure out how to call the various operations. You can also use these routines to find out on what processor and in what element a certain point is located. philipp On 2015-12-01 17:35, nek5000-users at lists.mcs.anl.gov wrote: > Yes, the routines in findpts.c are your friends. Have a look at hpts() > to see how it works. > > Cheers, > Stefan > Von: > > Antworten an: > > Datum: Tuesday, December 1, 2015 at 7:02 PM > An: > > Betreff: [Nek5000-users] find a point > > Hi Neks, > > Given a point P(x,y,z) (even not grid point), is there in Nek a fast way > to found in which element it is located and also the corresponding > processor? > > sincerely > Alessandro > _______________________________________________ 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 Dec 1 17:14:00 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 2 Dec 2015 00:14:00 +0100 Subject: [Nek5000-users] find a point In-Reply-To: References: Message-ID: Thanks for your answers! I think I found the routine that is right for me: call findpts(h, code_base, code_stride, proc_base, proc_stride, el_base, el_stride, r_base, r_stride, dist2_base, dist2_stride, x_base, x_stride, y_base, y_stride, z_base, z_stride, npt) nevertheless I didn't understand the meaning of some variables. I think that I can set the *_stride parameters equal to 1 for my case because i should follow just one point. "code_base" return 0 if is in the element, 1 if it is on the border and 2 if it is out. "proc_base" the processor that have the element ("el_base") in which the point is located. "x_base", "y_base" and "z_base" are the coordinates of the input point. But is not clear for me the means of "r_base", "dist2_base" and "h". Moreover I should give in input the "h" value and I saw that "h" is the output of findpts_setup function or intpts_setup if I use the routine defined in postpro.f that wrap the findpts_setup routine. However in both cases it is not clear for me what "h" represents. Other question, are *_base values global variables or local? namely, does each processor know that the investigated point is contained in the el_base element? Alessandro 2015-12-01 18:39 GMT+01:00 : > in postpro.f there is a Fortran routine hpts which acts as a driver to the > various hpts... routines. I guess from reading that subroutine you can > figure out how to call the various operations. You can also use these > routines to find out on what processor and in what element a certain point > is located. > > philipp > > On 2015-12-01 17:35, nek5000-users at lists.mcs.anl.gov wrote: > >> Yes, the routines in findpts.c are your friends. Have a look at hpts() >> to see how it works. >> >> Cheers, >> Stefan >> Von: > > >> Antworten an: > > >> Datum: Tuesday, December 1, 2015 at 7:02 PM >> An: > > >> Betreff: [Nek5000-users] find a point >> >> Hi Neks, >> >> Given a point P(x,y,z) (even not grid point), is there in Nek a fast way >> to found in which element it is located and also the corresponding >> processor? >> >> sincerely >> Alessandro >> _______________________________________________ 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 Wed Dec 2 08:10:38 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 02 Dec 2015 16:10:38 +0200 Subject: [Nek5000-users] find a point In-Reply-To: References: Message-ID: Here you go: h ? handle to use generated by findpts_setup stride ? used to increment the pointer to the next grid point (e.g. proc_stride = 1, x_stride = mdim) r_base ? stores the coordinates in rst-space (needed for interpolation) There are no global variables. All MPI_ranks have to call findpts().However in your case only one MPI_rank (it is your choice and can be completely arbitrary) will do the call with npt = 1, all others will use npt = 0. It does not matter if this point is located on that MPI_rank or not! Cheers, Stefan Von: Antworten an: Datum: Wednesday, December 2, 2015 at 1:14 AM An: Betreff: Re: [Nek5000-users] find a point Thanks for your answers! I think I found the routine that is right for me: call findpts(h, code_base, code_stride, proc_base, proc_stride, el_base, el_stride, r_base, r_stride, dist2_base, dist2_stride, x_base, x_stride, y_base, y_stride, z_base, z_stride, npt) nevertheless I didn't understand the meaning of some variables. I think that I can set the *_stride parameters equal to 1 for my case because i should follow just one point. "code_base" return 0 if is in the element, 1 if it is on the border and 2 if it is out. "proc_base" the processor that have the element ("el_base") in which the point is located. "x_base", "y_base" and "z_base" are the coordinates of the input point. But is not clear for me the means of "r_base", "dist2_base" and "h". Moreover I should give in input the "h" value and I saw that "h" is the output of findpts_setup function or intpts_setup if I use the routine defined in postpro.f that wrap the findpts_setup routine. However in both cases it is not clear for me what "h" represents. Other question, are *_base values global variables or local? namely, does each processor know that the investigated point is contained in the el_base element? Alessandro 2015-12-01 18:39 GMT+01:00 : > in postpro.f there is a Fortran routine hpts which acts as a driver to the > various hpts... routines. I guess from reading that subroutine you can figure > out how to call the various operations. You can also use these routines to > find out on what processor and in what element a certain point is located. > > philipp > > On 2015-12-01 17:35, nek5000-users at lists.mcs.anl.gov wrote: >> Yes, the routines in findpts.c are your friends. Have a look at hpts() >> to see how it works. >> >> Cheers, >> Stefan >> Von: > > >> Antworten an: > > >> Datum: Tuesday, December 1, 2015 at 7:02 PM >> An: > > >> Betreff: [Nek5000-users] find a point >> >> Hi Neks, >> >> Given a point P(x,y,z) (even not grid point), is there in Nek a fast way >> to found in which element it is located and also the corresponding >> processor? >> >> sincerely >> Alessandro >> _______________________________________________ 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 Wed Dec 2 13:55:00 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 2 Dec 2015 19:55:00 +0000 Subject: [Nek5000-users] Failed in Helmholtz Message-ID: Dear Neks, I'm solving a jet problem with a very simple geometry but the simulation keeps crashing. The error in the logfile is "Failed in Helmholtz" and when I post process the results, it appears that the simulation crashes when the jet hits the outflow boundary. I'm using the turb_outflow routine and have verified it by running the turbJet example in the repository. However, when I use my own mesh, created in prex, the same case (or even a simpler case with a lower Reynolds number) always seems to fail. Laminar cases, however, work fine. I have included my usr, rea, SIZE and logfiles. Any help is appreciated. Thank you, Goran Marjanovic Graduate Research Assistant Center for Compressible Multiphase Turbulence Mechanical and Aerospace Engineering University of Florida -------------- next part -------------- An HTML attachment was scrubbed... URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: jet.log1 Type: application/octet-stream Size: 11214382 bytes Desc: jet.log1 URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: jet.rea Type: application/octet-stream Size: 4091 bytes Desc: jet.rea URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: jet.usr Type: application/octet-stream Size: 8261 bytes Desc: jet.usr URL: -------------- next part -------------- A non-text attachment was scrubbed... Name: SIZE Type: application/octet-stream Size: 3232 bytes Desc: SIZE URL: From nek5000-users at lists.mcs.anl.gov Fri Dec 4 20:37:58 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sat, 5 Dec 2015 02:37:58 +0000 Subject: [Nek5000-users] Stress formulation In-Reply-To: References: Message-ID: Hi Andrew, I checked into this... I think what is happening is that the 'O ' bc for the stress formulation means stress-free, which is not guaranteed to yield a parabolic profile at the outlet. If you increase your Reynolds number I'm guessing that you'll recover the parabolic profile because the viscous stresses will diminish -- this is what I observed. >From my perspective, the outlet boundary is not a region where I would count on accurate physics --- it is, after all, a truncated domain, so I don't generally worry too much about the behavior there. 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: Tuesday, December 01, 2015 2:44 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Stress formulation Hi Neks, I have one question about stress formulation in NEK5000. I tried to make a simulation for Poiseuil flow with stress formulation. In inlet boundary condition use was made of parabolic profile. If IFSTRS=false the outlet profile is exactly the same as inlet one but in case of IFSTRS=true the outlet profile changes significantly (both the maximum value and profile shape). Could you tell me how I can get the parabolic profile at outlet, please? The same thing occurs with pressure values. The maximum pressure value with IFSTRS=true is three times higher then the one with IFSTRS=false. I found that the difference comes from subroutines where the stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for IFSTRS = true) but I don't undestand what is happening exactly. Thanks in advance for your help, Best regards, Andrew -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Sat Dec 5 04:40:41 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sat, 5 Dec 2015 11:40:41 +0100 Subject: [Nek5000-users] help me! Message-ID: Hi Neks, i am still trying to run my first simulation, everytime, something is wrong so it becoming stressful. In my last run, into the output i have this errors: WARNING: It appears that your OpenFabrics subsystem is configured to only allow registering part of your physical memory. This can cause MPI jobs to run with erratic performance, hang, and/or crash. This may be caused by your OpenFabrics vendor limiting the amount of physical memory that can be registered. You should investigate the relevant Linux kernel module parameters that control how much physical memory can be registered, and increase them to allow registering all physical memory on your machine. See this Open MPI FAQ item for more information on these Linux kernel module parameters: http://www.open-mpi.org/faq/?category=openfabrics#ib-locked-pages Local host: node9 Registerable memory: 32768 MiB Total memory: 65457 MiB Your MPI job will continue, but may be behave poorly and/or hang. -------------------------------------------------------------------------- /----------------------------------------------------------\\ | _ __ ______ __ __ ______ ____ ____ ____ | | / | / // ____// //_/ / ____/ / __ \\ / __ \\ / __ \\ | | / |/ // __/ / ,< /___ \\ / / / // / / // / / / | | / /| // /___ / /| | ____/ / / /_/ // /_/ // /_/ / | | /_/ |_//_____//_/ |_|/_____/ \\____/ \\____/ \\____/ | | | |----------------------------------------------------------| | | | NEK5000: Open Source Spectral Element Solver | | COPYRIGHT (c) 2008-2010 UCHICAGO ARGONNE, LLC | | Version: 1.0rc1 / SVN r1050 | | Web: http://nek5000.mcs.anl.gov | | | \\----------------------------------------------------------/ Number of processors: 64 REAL wdsize : 8 INTEGER wdsize : 4 Beginning session: /datastor/eng/gualtieri/SIMULATIONS_NEK/RIB_3D_01/curve.rea timer accuracy: 0.0000000E+00 sec read .rea file -------------------------------------------------------------------------- An MPI process has executed an operation involving a call to the "fork()" system call to create a child process. Open MPI is currently operating in a condition that could result in memory corruption or other system errors; your MPI job may hang, crash, or produce silent data corruption. The use of fork() (or system() or other calls that create child processes) is strongly discouraged. The process that invoked fork was: Local host: node9 (PID 72129) MPI_COMM_WORLD rank: 0 If you are *absolutely sure* that your application will successfully and correctly survive a call to fork(), you may disable this warning by setting the mpi_warn_on_fork MCA parameter to 0. -------------------------------------------------------------------------- Thread 4 (Thread 0x2b167717b700 (LWP 72146)): #0 0x00000034136df343 in poll () from /lib64/libc.so.6 #1 0x00002b16754a2d46 in poll_dispatch () from /opt/shared/openmpi-1.8.1_gnu/lib/libopen-pal.so.6 #2 0x00002b167549a7db in opal_libevent2021_event_base_loop () from /opt/shared/openmpi-1.8.1_gnu/lib/libopen-pal.so.6 #3 0x00002b16751f830e in orte_progress_thread_engine () from /opt/shared/openmpi-1.8.1_gnu/lib/libopen-rte.so.7 #4 0x0000003413a079d1 in start_thread () from /lib64/libpthread.so.0 #5 0x00000034136e8b6d in clone () from /lib64/libc.so.6 Thread 3 (Thread 0x2b167af9e700 (LWP 72162)): #0 0x00000034136e15e3 in select () from /lib64/libc.so.6 #1 0x00002b16783e31c3 in service_thread_start () from /opt/shared/openmpi-1.8.1_gnu/lib/openmpi/mca_btl_openib.so #2 0x0000003413a079d1 in start_thread () from /lib64/libpthread.so.0 #3 0x00000034136e8b6d in clone () from /lib64/libc.so.6 Thread 2 (Thread 0x2b1683560700 (LWP 72190)): #0 0x00000034136df343 in poll () from /lib64/libc.so.6 #1 0x00002b16783e1902 in btl_openib_async_thread () from /opt/shared/openmpi-1.8.1_gnu/lib/openmpi/mca_btl_openib.so #2 0x0000003413a079d1 in start_thread () from /lib64/libpthread.so.0 #3 0x00000034136e8b6d in clone () from /lib64/libc.so.6 Thread 1 (Thread 0x2b1675708200 (LWP 72129)): #0 0x0000003413a0f203 in wait () from /lib64/libpthread.so.0 #1 0x00002b1674ee100d in ?? () from /usr/lib64/libgfortran.so.3 #2 0x00002b1674ee282e in ?? () from /usr/lib64/libgfortran.so.3 #3 0x00002b1674ee2a81 in _gfortran_generate_error () from /usr/lib64/libgfortran.so.3 #4 0x00002b1674f878a5 in ?? () from /usr/lib64/libgfortran.so.3 #5 0x00002b1674f87b53 in _gfortran_st_open () from /usr/lib64/libgfortran.so.3 #6 0x0000000000432446 in readat_ () #7 0x00000000004030b5 in nek_init_ () #8 0x0000000000402974 in main () ------------------------------------------------------- Primary job terminated normally, but 1 process returned a non-zero exit code.. Per user-direction, the job has been aborted. ------------------------------------------------------- -------------------------------------------------------------------------- mpirun detected that one or more processes exited with non-zero status, thus causing the job to be terminated. The first process to do so was: Process name: [[17614,1],0] Exit code: 2 -------------------------------------------------------------------------- My question is, are these related to the software and openmpi (how i suppose) or are they related to my setup? Thanks to everyone. From nek5000-users at lists.mcs.anl.gov Sat Dec 5 17:49:18 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sat, 5 Dec 2015 18:49:18 -0500 Subject: [Nek5000-users] help me! In-Reply-To: References: Message-ID: Would you mind sharing what kind of setup you have? How is your OpenMPI configured? On Dec 5, 2015 5:41 AM, wrote: > Hi Neks, > i am still trying to run my first simulation, everytime, something is > wrong so it becoming stressful. > In my last run, into the output i have this errors: > > > > WARNING: It appears that your OpenFabrics subsystem is configured to only > allow registering part of your physical memory. This can cause MPI jobs to > run with erratic performance, hang, and/or crash. > > This may be caused by your OpenFabrics vendor limiting the amount of > physical memory that can be registered. You should investigate the > relevant Linux kernel module parameters that control how much physical > memory can be registered, and increase them to allow registering all > physical memory on your machine. > > See this Open MPI FAQ item for more information on these Linux kernel > module > parameters: > > http://www.open-mpi.org/faq/?category=openfabrics#ib-locked-pages > > Local host: node9 > Registerable memory: 32768 MiB > Total memory: 65457 MiB > > Your MPI job will continue, but may be behave poorly and/or hang. > -------------------------------------------------------------------------- > /----------------------------------------------------------\\ > | _ __ ______ __ __ ______ ____ ____ ____ | > | / | / // ____// //_/ / ____/ / __ \\ / __ \\ / __ \\ | > | / |/ // __/ / ,< /___ \\ / / / // / / // / / / | > | / /| // /___ / /| | ____/ / / /_/ // /_/ // /_/ / | > | /_/ |_//_____//_/ |_|/_____/ \\____/ \\____/ \\____/ | > | | > |----------------------------------------------------------| > | | > | NEK5000: Open Source Spectral Element Solver | > | COPYRIGHT (c) 2008-2010 UCHICAGO ARGONNE, LLC | > | Version: 1.0rc1 / SVN r1050 | > | Web: http://nek5000.mcs.anl.gov | > | | > \\----------------------------------------------------------/ > > > Number of processors: 64 > REAL wdsize : 8 > INTEGER wdsize : 4 > > > Beginning session: > /datastor/eng/gualtieri/SIMULATIONS_NEK/RIB_3D_01/curve.rea > > > timer accuracy: 0.0000000E+00 sec > > read .rea file > -------------------------------------------------------------------------- > An MPI process has executed an operation involving a call to the > "fork()" system call to create a child process. Open MPI is currently > operating in a condition that could result in memory corruption or > other system errors; your MPI job may hang, crash, or produce silent > data corruption. The use of fork() (or system() or other calls that > create child processes) is strongly discouraged. > > The process that invoked fork was: > > Local host: node9 (PID 72129) > MPI_COMM_WORLD rank: 0 > > If you are *absolutely sure* that your application will successfully > and correctly survive a call to fork(), you may disable this warning > by setting the mpi_warn_on_fork MCA parameter to 0. > -------------------------------------------------------------------------- > Thread 4 (Thread 0x2b167717b700 (LWP 72146)): > #0 0x00000034136df343 in poll () from /lib64/libc.so.6 > #1 0x00002b16754a2d46 in poll_dispatch () from > /opt/shared/openmpi-1.8.1_gnu/lib/libopen-pal.so.6 > #2 0x00002b167549a7db in opal_libevent2021_event_base_loop () from > /opt/shared/openmpi-1.8.1_gnu/lib/libopen-pal.so.6 > #3 0x00002b16751f830e in orte_progress_thread_engine () from > /opt/shared/openmpi-1.8.1_gnu/lib/libopen-rte.so.7 > #4 0x0000003413a079d1 in start_thread () from /lib64/libpthread.so.0 > #5 0x00000034136e8b6d in clone () from /lib64/libc.so.6 > Thread 3 (Thread 0x2b167af9e700 (LWP 72162)): > #0 0x00000034136e15e3 in select () from /lib64/libc.so.6 > #1 0x00002b16783e31c3 in service_thread_start () from > /opt/shared/openmpi-1.8.1_gnu/lib/openmpi/mca_btl_openib.so > #2 0x0000003413a079d1 in start_thread () from /lib64/libpthread.so.0 > #3 0x00000034136e8b6d in clone () from /lib64/libc.so.6 > Thread 2 (Thread 0x2b1683560700 (LWP 72190)): > #0 0x00000034136df343 in poll () from /lib64/libc.so.6 > #1 0x00002b16783e1902 in btl_openib_async_thread () from > /opt/shared/openmpi-1.8.1_gnu/lib/openmpi/mca_btl_openib.so > #2 0x0000003413a079d1 in start_thread () from /lib64/libpthread.so.0 > #3 0x00000034136e8b6d in clone () from /lib64/libc.so.6 > Thread 1 (Thread 0x2b1675708200 (LWP 72129)): > #0 0x0000003413a0f203 in wait () from /lib64/libpthread.so.0 > #1 0x00002b1674ee100d in ?? () from /usr/lib64/libgfortran.so.3 > #2 0x00002b1674ee282e in ?? () from /usr/lib64/libgfortran.so.3 > #3 0x00002b1674ee2a81 in _gfortran_generate_error () from > /usr/lib64/libgfortran.so.3 > #4 0x00002b1674f878a5 in ?? () from /usr/lib64/libgfortran.so.3 > #5 0x00002b1674f87b53 in _gfortran_st_open () from > /usr/lib64/libgfortran.so.3 > #6 0x0000000000432446 in readat_ () > #7 0x00000000004030b5 in nek_init_ () > #8 0x0000000000402974 in main () > ------------------------------------------------------- > Primary job terminated normally, but 1 process returned > a non-zero exit code.. Per user-direction, the job has been aborted. > ------------------------------------------------------- > -------------------------------------------------------------------------- > mpirun detected that one or more processes exited with non-zero status, > thus causing > the job to be terminated. The first process to do so was: > > Process name: [[17614,1],0] > Exit code: 2 > -------------------------------------------------------------------------- > > My question is, are these related to the software and openmpi (how i > suppose) or are they related to my setup? > Thanks to everyone. > _______________________________________________ > 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 Dec 6 21:23:25 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sun, 6 Dec 2015 22:23:25 -0500 Subject: [Nek5000-users] How to obtain spectral coefficients and basis Message-ID: Hello, I'm trying to represent the solution, either velocity or temperature, in terms of spectral basis, i.e. $u=\Sigma a_i \phi_i $ with $\phi_i$ as the basis and $a_i$ as the coefficients. I have two questions: 1) (a naive one!) Which basis is being used in Nek5000? I assume it is Chebyshev polynomials. Is that true? If not, what is the basis? 2) How can I retrieve the coefficients $a_i$? Also, is there a way to access such coefficients during the solution, say by writing them to a file? I really appreciate the responses. As a side note, I'd like to use DNS data for model reduction. Many thanks, Saleh From nek5000-users at lists.mcs.anl.gov Sun Dec 6 21:39:02 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sun, 6 Dec 2015 21:39:02 -0600 Subject: [Nek5000-users] Local drag calculation Message-ID: Hello Neks, A question regarding the calculation of drag for y+ calculation at specific points in a domain: How would one go about altering the approach used in the example cases such as turbchannel, where the drag is integrated over larger surfaces, to only calculate the drag at a single point in the domain? My simulation pertains to flow through pebble beds, so the drag should be relatively anisotropic over the pebble surfaces, hence the point value approach. My thinking is to find the point in the domain where my line of interpolation begins and then calculating the drag for the element that contains said point, would that be appropriate? Finally, is it then appropriate to use the dU/dy value for the largest component (I?m assuming it?ll be for the stream-wise velocity component) as the basis for the calculation of the y+ value? Thanks for any assistance regarding this! Lambert From nek5000-users at lists.mcs.anl.gov Tue Dec 8 10:26:43 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 8 Dec 2015 17:26:43 +0100 Subject: [Nek5000-users] How to obtain spectral coefficients and basis In-Reply-To: References: Message-ID: Hello, If I've understood you propely, you can see all the collocation points and weights in the file coef.f (subroutine genwz). Andrew 2015-12-07 4:23 GMT+01:00 : > Hello, > > I'm trying to represent the solution, either velocity or temperature, in > terms of spectral basis, i.e. > > $u=\Sigma a_i \phi_i $ with $\phi_i$ as the basis and $a_i$ as the > coefficients. > > I have two questions: > > 1) (a naive one!) Which basis is being used in Nek5000? I assume it is > Chebyshev polynomials. Is that true? If not, what is the basis? > > 2) How can I retrieve the coefficients $a_i$? Also, is there a way to > access such coefficients during the solution, say by writing them to a file? > > I really appreciate the responses. As a side note, I'd like to use DNS > data for model reduction. > > Many thanks, > Saleh > _______________________________________________ > 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 Dec 8 12:59:55 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 8 Dec 2015 19:59:55 +0100 Subject: [Nek5000-users] Derivative matrix in 2D axisymmetric case Message-ID: Hello Nek's, I have a couple of questions regarding the simulations in 2D axisymmetric case. The first one concerns the variational problem. If IFAXIS=T, what Navier Stockes equation is utilized- the one in cartesian coordinate system or the one in cylindrical coordinate system (and what gradient of test function is taken)? The second one concerns stiffness matrix. It seems to me that we add the suplement term if an element is placed at the axe (ifrzer(e)=true). Why do we do this? Plus, different derivative matrix are utilized in differents collocation points (Gauss-Lobato-Jacobi) for theses elements but added term isn't utilized with stress-formulation. Thanks a lot, ANDREY -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Dec 9 06:29:50 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 9 Dec 2015 12:29:50 +0000 Subject: [Nek5000-users] Full restart with a different time step In-Reply-To: References: , Message-ID: Dear Neks, Sorry for bringing up this topic again. I just found something with the varying time step. In my simulation, I'm using fixed mass flow rate with PnPn-2 method. I'm also using the full restart option. As suggested by Philipp last time, once I stop reading restart files from the disk, I could go ahead with a different time step size. However, the simulation crashed with NaN values since the step I changed my DT. Interestingly, varying DT works when I use (fixed mass flow + PnPn) combination. Just wonder why varying DT doesn't work for (fixed mass flow + PnPn-2)? Does anyone have any idea about this? Thank you very much in advance. Best regards, Tony From nek5000-users at lists.mcs.anl.gov Wed Dec 9 14:23:27 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 9 Dec 2015 20:23:27 +0000 Subject: [Nek5000-users] Additional MHD Examples? Message-ID: Are there any other MHD examples available beyond the one provided in /examples/mhd/? I would be most interested in an example with a constant background magnetic field applied to the computational domain. Also, the example included in the nek source says that the magnetic field boundary conditions must match the velocity boundary conditions, but there is traffic on nek-users which implies that this may have been updated. Is this update already in the nek repository? Thanks, Joel From nek5000-users at lists.mcs.anl.gov Thu Dec 10 10:31:42 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 10 Dec 2015 16:31:42 +0000 Subject: [Nek5000-users] Additional MHD Examples? In-Reply-To: References: Message-ID: Hi Joel, We did implement changes to the code when boundary conditions for velocity and magnetic field could be different. Let's discuss offline what are the ones you need. Thanks. Aleks ________________________________________ 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: Wednesday, December 09, 2015 2:23 PM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Additional MHD Examples? Are there any other MHD examples available beyond the one provided in /examples/mhd/? I would be most interested in an example with a constant background magnetic field applied to the computational domain. Also, the example included in the nek source says that the magnetic field boundary conditions must match the velocity boundary conditions, but there is traffic on nek-users which implies that this may have been updated. Is this update already in the nek repository? Thanks, Joel _______________________________________________ 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 Dec 13 17:19:30 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 00:19:30 +0100 Subject: [Nek5000-users] Stress formulation In-Reply-To: References: Message-ID: Hi Paul, Thank you for your reply. Yes, if i increase Reynolds number, the outlet profile become much close to the inlet one. But in my simulations the Reynolds number is between 300 and 1200. Can you recommend me what kind of boundary conditions i can utilise for ensure for ensure the flow physics ? I would implement these one in code. Thanks, Andrew 2015-12-05 3:37 GMT+01:00 : > > Hi Andrew, > > I checked into this... I think what is happening is that the 'O ' bc for > the stress formulation > means stress-free, which is not guaranteed to yield a parabolic profile > at the outlet. > > If you increase your Reynolds number I'm guessing that you'll recover the > parabolic > profile because the viscous stresses will diminish -- this is what I > observed. > > From my perspective, the outlet boundary is not a region where I would > count on accurate > physics --- it is, after all, a truncated domain, so I don't generally > worry too much about > the behavior there. > > 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:* Tuesday, December 01, 2015 2:44 AM > *To:* nek5000-users at lists.mcs.anl.gov > *Subject:* [Nek5000-users] Stress formulation > > Hi Neks, > > I have one question about stress formulation in NEK5000. > I tried to make a simulation for Poiseuil flow with stress > formulation. In inlet boundary condition use was made of parabolic profile. > If IFSTRS=false the outlet profile is exactly the same as inlet one but in > case of IFSTRS=true the outlet profile changes significantly (both the > maximum value and profile shape). Could you tell me how I can get the > parabolic profile at outlet, please? > The same thing occurs with pressure values. The maximum pressure > value with IFSTRS=true is three times higher then the one with IFSTRS=false. > > I found that the difference comes from subroutines where the > stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for > IFSTRS = true) but I don't undestand what is happening exactly. > > Thanks in advance for your help, > Best regards, > Andrew > > _______________________________________________ > 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 Dec 13 17:55:04 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sun, 13 Dec 2015 23:55:04 +0000 Subject: [Nek5000-users] Stress formulation In-Reply-To: References: , Message-ID: Hi Andrew, In my view, you are already getting the correct flow physics. What happens at outflow, especially at moderate Reynolds number, is not necessarily parallel flow. The stress formulation is giving the stress-free result. 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, December 13, 2015 5:19 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Stress formulation Hi Paul, Thank you for your reply. Yes, if i increase Reynolds number, the outlet profile become much close to the inlet one. But in my simulations the Reynolds number is between 300 and 1200. Can you recommend me what kind of boundary conditions i can utilise for ensure for ensure the flow physics ? I would implement these one in code. Thanks, Andrew 2015-12-05 3:37 GMT+01:00 >: Hi Andrew, I checked into this... I think what is happening is that the 'O ' bc for the stress formulation means stress-free, which is not guaranteed to yield a parabolic profile at the outlet. If you increase your Reynolds number I'm guessing that you'll recover the parabolic profile because the viscous stresses will diminish -- this is what I observed. >From my perspective, the outlet boundary is not a region where I would count on accurate physics --- it is, after all, a truncated domain, so I don't generally worry too much about the behavior there. 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: Tuesday, December 01, 2015 2:44 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Stress formulation Hi Neks, I have one question about stress formulation in NEK5000. I tried to make a simulation for Poiseuil flow with stress formulation. In inlet boundary condition use was made of parabolic profile. If IFSTRS=false the outlet profile is exactly the same as inlet one but in case of IFSTRS=true the outlet profile changes significantly (both the maximum value and profile shape). Could you tell me how I can get the parabolic profile at outlet, please? The same thing occurs with pressure values. The maximum pressure value with IFSTRS=true is three times higher then the one with IFSTRS=false. I found that the difference comes from subroutines where the stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for IFSTRS = true) but I don't undestand what is happening exactly. Thanks in advance for your help, Best regards, Andrew _______________________________________________ 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 Dec 13 21:48:09 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 11:48:09 +0800 Subject: [Nek5000-users] freestream turbulence at INLET Message-ID: Hi neks, I looked at this nek5000 simulation on youtube of a flow over an airfoil. https://www.youtube.com/watch?v=aR-hehP1pTk In the video, a Dirichlet condition was imposed at the inlet from a RANS simulation. My question is how do you this? I have a RANS simulation from Fluent which prescribes freestream turbulence. I wish to use a plane of the Fluent simulations as my inlet in the nek solver. Can you help with this? Cheers Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Dec 14 00:39:19 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 07:39:19 +0100 Subject: [Nek5000-users] freestream turbulence at INLET Message-ID: Dear Rif, In our wing simulation the boundary layers are tripped using a volume force in the direction normal to the surface. This emulates the approach in wind tunnel experiments, where the models are tripped using sand paper, V disturbances, etc. In such wind tunnel tests the transition location is prescribed, and the freestream turbulence intensity is very low. The idea is to have a ?canonical? turbulent boundary layer, developing in the most controlled way in order to study the effects of turbulence. Therefore, in our wing simulation turbulence is also zero at our boundaries. Regarding the use of Fluent data in Nek, could you give us some more information about your setup? Which RANS model are you using, and is it time-dependent? What kind of geometry are you simulating? Best regards, Ricardo Vinuesa. Sent from Mail for Windows 10 -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Dec 14 01:08:00 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 00:08:00 -0700 Subject: [Nek5000-users] Nek5000-users Digest, Vol 82, Issue 11 In-Reply-To: References: Message-ID: Hi Ricardo, I am using the k-omega SST turbulence model in Fluent to simulate a 3D flow over the S805 airfoil. The wind tunnel measurements measured turbulence intensity at 0.4% with wind speed of 2.8 m/s. It is a time dependent simulation. What I plan to do is emulate the RANS inlet velocity field in the NEK5000 DNS simulations at the inlet. Can you help with this? Many thanks. Cheers Rif On Sun, Dec 13, 2015 at 11:39 PM, 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: Stress formulation (nek5000-users at lists.mcs.anl.gov) > 2. Re: Stress formulation (nek5000-users at lists.mcs.anl.gov) > 3. freestream turbulence at INLET (nek5000-users at lists.mcs.anl.gov) > 4. Re: freestream turbulence at INLET > (nek5000-users at lists.mcs.anl.gov) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Mon, 14 Dec 2015 00:19:30 +0100 > From: nek5000-users at lists.mcs.anl.gov > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Stress formulation > Message-ID: > > Content-Type: text/plain; charset="utf-8" > > Hi Paul, > > Thank you for your reply. > Yes, if i increase Reynolds number, the outlet profile become much close to > the inlet one. But in my simulations the Reynolds number is between 300 and > 1200. > Can you recommend me what kind of boundary conditions i can utilise for > ensure for ensure the flow physics ? I would implement these one in code. > > Thanks, > Andrew > > > 2015-12-05 3:37 GMT+01:00 : > > > > > Hi Andrew, > > > > I checked into this... I think what is happening is that the 'O ' bc for > > the stress formulation > > means stress-free, which is not guaranteed to yield a parabolic profile > > at the outlet. > > > > If you increase your Reynolds number I'm guessing that you'll recover the > > parabolic > > profile because the viscous stresses will diminish -- this is what I > > observed. > > > > From my perspective, the outlet boundary is not a region where I would > > count on accurate > > physics --- it is, after all, a truncated domain, so I don't generally > > worry too much about > > the behavior there. > > > > 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:* Tuesday, December 01, 2015 2:44 AM > > *To:* nek5000-users at lists.mcs.anl.gov > > *Subject:* [Nek5000-users] Stress formulation > > > > Hi Neks, > > > > I have one question about stress formulation in NEK5000. > > I tried to make a simulation for Poiseuil flow with stress > > formulation. In inlet boundary condition use was made of parabolic > profile. > > If IFSTRS=false the outlet profile is exactly the same as inlet one but > in > > case of IFSTRS=true the outlet profile changes significantly (both the > > maximum value and profile shape). Could you tell me how I can get the > > parabolic profile at outlet, please? > > The same thing occurs with pressure values. The maximum pressure > > value with IFSTRS=true is three times higher then the one with > IFSTRS=false. > > > > I found that the difference comes from subroutines where the > > stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for > > IFSTRS = true) but I don't undestand what is happening exactly. > > > > Thanks in advance for your help, > > Best regards, > > Andrew > > > > _______________________________________________ > > 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: < > http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151214/abcab862/attachment-0001.html > > > > ------------------------------ > > Message: 2 > Date: Sun, 13 Dec 2015 23:55:04 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > Subject: Re: [Nek5000-users] Stress formulation > Message-ID: > > Content-Type: text/plain; charset="iso-8859-1" > > > Hi Andrew, > > In my view, you are already getting the correct flow physics. > > What happens at outflow, especially at moderate Reynolds number, is not > necessarily parallel flow. > > The stress formulation is giving the stress-free result. > > 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, December 13, 2015 5:19 PM > To: nek5000-users at lists.mcs.anl.gov > Subject: Re: [Nek5000-users] Stress formulation > > Hi Paul, > > Thank you for your reply. > Yes, if i increase Reynolds number, the outlet profile become much close > to the inlet one. But in my simulations the Reynolds number is between 300 > and 1200. > Can you recommend me what kind of boundary conditions i can utilise for > ensure for ensure the flow physics ? I would implement these one in code. > > Thanks, > Andrew > > > 2015-12-05 3:37 GMT+01:00 nek5000-users at lists.mcs.anl.gov>>: > > Hi Andrew, > > I checked into this... I think what is happening is that the 'O ' bc for > the stress formulation > means stress-free, which is not guaranteed to yield a parabolic profile > at the outlet. > > If you increase your Reynolds number I'm guessing that you'll recover the > parabolic > profile because the viscous stresses will diminish -- this is what I > observed. > > >From my perspective, the outlet boundary is not a region where I would > count on accurate > physics --- it is, after all, a truncated domain, so I don't generally > worry too much about > the behavior there. > > Paul > > ________________________________ > From: nek5000-users-bounces at lists.mcs.anl.gov nek5000-users-bounces at lists.mcs.anl.gov> [ > 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: Tuesday, December 01, 2015 2:44 AM > To: nek5000-users at lists.mcs.anl.gov > > Subject: [Nek5000-users] Stress formulation > > Hi Neks, > > I have one question about stress formulation in NEK5000. > I tried to make a simulation for Poiseuil flow with stress > formulation. In inlet boundary condition use was made of parabolic profile. > If IFSTRS=false the outlet profile is exactly the same as inlet one but in > case of IFSTRS=true the outlet profile changes significantly (both the > maximum value and profile shape). Could you tell me how I can get the > parabolic profile at outlet, please? > The same thing occurs with pressure values. The maximum pressure > value with IFSTRS=true is three times higher then the one with IFSTRS=false. > > I found that the difference comes from subroutines where the > stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for > IFSTRS = true) but I don't undestand what is happening exactly. > > Thanks in advance for your help, > Best regards, > Andrew > > _______________________________________________ > 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: < > http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151213/31d8ae52/attachment-0001.html > > > > ------------------------------ > > Message: 3 > Date: Mon, 14 Dec 2015 11:48:09 +0800 > From: nek5000-users at lists.mcs.anl.gov > To: nek5000-users at lists.mcs.anl.gov > Subject: [Nek5000-users] freestream turbulence at INLET > Message-ID: > > Content-Type: text/plain; charset="utf-8" > > Hi neks, > I looked at this nek5000 simulation on youtube of a flow over an airfoil. > https://www.youtube.com/watch?v=aR-hehP1pTk > In the video, a Dirichlet condition was imposed at the inlet from a RANS > simulation. My question is how do you this? > I have a RANS simulation from Fluent which prescribes freestream > turbulence. I wish to use a plane of the Fluent simulations as my inlet in > the nek solver. Can you help with this? > > Cheers > Rif > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151214/a41bf16f/attachment-0001.html > > > > ------------------------------ > > Message: 4 > Date: Mon, 14 Dec 2015 07:39:19 +0100 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > > Subject: Re: [Nek5000-users] freestream turbulence at INLET > Message-ID: > > Content-Type: text/plain; charset="utf-8" > > Dear Rif, > > In our wing simulation the boundary layers are tripped using a volume > force in the direction normal to the surface. This emulates the approach in > wind tunnel experiments, where the models are tripped using sand paper, V > disturbances, etc. In such wind tunnel tests the transition location is > prescribed, and the freestream turbulence intensity is very low. The idea > is to have a ?canonical? turbulent boundary layer, developing in the most > controlled way in order to study the effects of turbulence. Therefore, in > our wing simulation turbulence is also zero at our boundaries. > > Regarding the use of Fluent data in Nek, could you give us some more > information about your setup? Which RANS model are you using, and is it > time-dependent? What kind of geometry are you simulating? > > Best regards, > > Ricardo Vinuesa. > > Sent from Mail for Windows 10 > > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: < > http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151214/f909c493/attachment.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 82, Issue 11 > ********************************************* > -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Dec 14 03:04:12 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 02:04:12 -0700 Subject: [Nek5000-users] freestream turbulence at INLET Message-ID: Hi Ricardo, I am using the k-omega SST turbulence model in Fluent to simulate a 3D flow over the S805 airfoil. The wind tunnel measurements measured turbulence intensity at 0.4% with wind speed of 2.8 m/s. It is a time dependent simulation. What I plan to do is emulate the RANS inlet velocity field in the NEK5000 DNS simulations at the inlet. Can you help with this? Many thanks. Cheers Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Dec 14 00:39:07 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 07:39:07 +0100 Subject: [Nek5000-users] freestream turbulence at INLET Message-ID: Dear Rif, In our wing simulation the boundary layers are tripped using a volume force in the direction normal to the surface, emulating the approach used in wind tunnel experiments. Such tripping devices include sand paper and "V" disturbances, and in wind tunnel measurements the transition location is prescribed, whereas the turbulence level is very low. This is done to obtain a "canonical" developing turbulent boundary layer, to ease the study of turbulence effects. This is also done in our case, where the turbulence intensity is zero at the boundaries (since the velocity is prescribed at all times). Regarding your simulation, could you gives us more details about your setup? What RANS models are using in Fluent, and is it a time-dependent simulation? What geometry are you simulating in Nek? Best regards, Ricardo Vinuesa. From nek5000-users at lists.mcs.anl.gov Mon Dec 14 02:16:34 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 09:16:34 +0100 Subject: [Nek5000-users] Nek5000-users Digest, Vol 82, Issue 11 In-Reply-To: References: Message-ID: Dear Rif, I think that a good starting point would be to use the time history of the RANS field as a time-varying inflow condition of your Nek simulation. We did that in one of our cases, where we imposed a time-varying inflow from a DNS to run another DNS, and the results were excellent. See attached paper for further details. Let us know how it goes, and if you need any additional help! Cheers, Ricardo. On Mon, Dec 14, 2015 at 8:08 AM, wrote: > Hi Ricardo, > I am using the k-omega SST turbulence model in Fluent to simulate a 3D > flow over the S805 airfoil. The wind tunnel measurements measured > turbulence intensity at 0.4% with wind speed of 2.8 m/s. It is a time > dependent simulation. What I plan to do is emulate the RANS inlet velocity > field in the NEK5000 DNS simulations at the inlet. Can you help with this? > Many thanks. > > Cheers > Rif > > On Sun, Dec 13, 2015 at 11:39 PM, > 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: Stress formulation (nek5000-users at lists.mcs.anl.gov) >> 2. Re: Stress formulation (nek5000-users at lists.mcs.anl.gov) >> 3. freestream turbulence at INLET (nek5000-users at lists.mcs.anl.gov) >> 4. Re: freestream turbulence at INLET >> (nek5000-users at lists.mcs.anl.gov) >> >> >> ---------------------------------------------------------------------- >> >> Message: 1 >> Date: Mon, 14 Dec 2015 00:19:30 +0100 >> From: nek5000-users at lists.mcs.anl.gov >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Stress formulation >> Message-ID: >> >> Content-Type: text/plain; charset="utf-8" >> >> Hi Paul, >> >> Thank you for your reply. >> Yes, if i increase Reynolds number, the outlet profile become much close >> to >> the inlet one. But in my simulations the Reynolds number is between 300 >> and >> 1200. >> Can you recommend me what kind of boundary conditions i can utilise for >> ensure for ensure the flow physics ? I would implement these one in code. >> >> Thanks, >> Andrew >> >> >> 2015-12-05 3:37 GMT+01:00 : >> >> > >> > Hi Andrew, >> > >> > I checked into this... I think what is happening is that the 'O ' bc >> for >> > the stress formulation >> > means stress-free, which is not guaranteed to yield a parabolic profile >> > at the outlet. >> > >> > If you increase your Reynolds number I'm guessing that you'll recover >> the >> > parabolic >> > profile because the viscous stresses will diminish -- this is what I >> > observed. >> > >> > From my perspective, the outlet boundary is not a region where I would >> > count on accurate >> > physics --- it is, after all, a truncated domain, so I don't generally >> > worry too much about >> > the behavior there. >> > >> > 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:* Tuesday, December 01, 2015 2:44 AM >> > *To:* nek5000-users at lists.mcs.anl.gov >> > *Subject:* [Nek5000-users] Stress formulation >> > >> > Hi Neks, >> > >> > I have one question about stress formulation in NEK5000. >> > I tried to make a simulation for Poiseuil flow with stress >> > formulation. In inlet boundary condition use was made of parabolic >> profile. >> > If IFSTRS=false the outlet profile is exactly the same as inlet one but >> in >> > case of IFSTRS=true the outlet profile changes significantly (both the >> > maximum value and profile shape). Could you tell me how I can get the >> > parabolic profile at outlet, please? >> > The same thing occurs with pressure values. The maximum pressure >> > value with IFSTRS=true is three times higher then the one with >> IFSTRS=false. >> > >> > I found that the difference comes from subroutines where the >> > stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for >> > IFSTRS = true) but I don't undestand what is happening exactly. >> > >> > Thanks in advance for your help, >> > Best regards, >> > Andrew >> > >> > _______________________________________________ >> > 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: < >> http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151214/abcab862/attachment-0001.html >> > >> >> ------------------------------ >> >> Message: 2 >> Date: Sun, 13 Dec 2015 23:55:04 +0000 >> From: nek5000-users at lists.mcs.anl.gov >> To: "nek5000-users at lists.mcs.anl.gov" >> >> Subject: Re: [Nek5000-users] Stress formulation >> Message-ID: >> >> Content-Type: text/plain; charset="iso-8859-1" >> >> >> Hi Andrew, >> >> In my view, you are already getting the correct flow physics. >> >> What happens at outflow, especially at moderate Reynolds number, is not >> necessarily parallel flow. >> >> The stress formulation is giving the stress-free result. >> >> 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, December 13, 2015 5:19 PM >> To: nek5000-users at lists.mcs.anl.gov >> Subject: Re: [Nek5000-users] Stress formulation >> >> Hi Paul, >> >> Thank you for your reply. >> Yes, if i increase Reynolds number, the outlet profile become much close >> to the inlet one. But in my simulations the Reynolds number is between 300 >> and 1200. >> Can you recommend me what kind of boundary conditions i can utilise for >> ensure for ensure the flow physics ? I would implement these one in code. >> >> Thanks, >> Andrew >> >> >> 2015-12-05 3:37 GMT+01:00 > nek5000-users at lists.mcs.anl.gov>>: >> >> Hi Andrew, >> >> I checked into this... I think what is happening is that the 'O ' bc for >> the stress formulation >> means stress-free, which is not guaranteed to yield a parabolic profile >> at the outlet. >> >> If you increase your Reynolds number I'm guessing that you'll recover the >> parabolic >> profile because the viscous stresses will diminish -- this is what I >> observed. >> >> >From my perspective, the outlet boundary is not a region where I would >> count on accurate >> physics --- it is, after all, a truncated domain, so I don't generally >> worry too much about >> the behavior there. >> >> Paul >> >> ________________________________ >> From: nek5000-users-bounces at lists.mcs.anl.gov> nek5000-users-bounces at lists.mcs.anl.gov> [ >> 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: Tuesday, December 01, 2015 2:44 AM >> To: nek5000-users at lists.mcs.anl.gov> nek5000-users at lists.mcs.anl.gov> >> Subject: [Nek5000-users] Stress formulation >> >> Hi Neks, >> >> I have one question about stress formulation in NEK5000. >> I tried to make a simulation for Poiseuil flow with stress >> formulation. In inlet boundary condition use was made of parabolic profile. >> If IFSTRS=false the outlet profile is exactly the same as inlet one but in >> case of IFSTRS=true the outlet profile changes significantly (both the >> maximum value and profile shape). Could you tell me how I can get the >> parabolic profile at outlet, please? >> The same thing occurs with pressure values. The maximum pressure >> value with IFSTRS=true is three times higher then the one with IFSTRS=false. >> >> I found that the difference comes from subroutines where the >> stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for >> IFSTRS = true) but I don't undestand what is happening exactly. >> >> Thanks in advance for your help, >> Best regards, >> Andrew >> >> _______________________________________________ >> 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: < >> http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151213/31d8ae52/attachment-0001.html >> > >> >> ------------------------------ >> >> Message: 3 >> Date: Mon, 14 Dec 2015 11:48:09 +0800 >> From: nek5000-users at lists.mcs.anl.gov >> To: nek5000-users at lists.mcs.anl.gov >> Subject: [Nek5000-users] freestream turbulence at INLET >> Message-ID: >> >> Content-Type: text/plain; charset="utf-8" >> >> Hi neks, >> I looked at this nek5000 simulation on youtube of a flow over an airfoil. >> https://www.youtube.com/watch?v=aR-hehP1pTk >> In the video, a Dirichlet condition was imposed at the inlet from a RANS >> simulation. My question is how do you this? >> I have a RANS simulation from Fluent which prescribes freestream >> turbulence. I wish to use a plane of the Fluent simulations as my inlet in >> the nek solver. Can you help with this? >> >> Cheers >> Rif >> -------------- next part -------------- >> An HTML attachment was scrubbed... >> URL: < >> http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151214/a41bf16f/attachment-0001.html >> > >> >> ------------------------------ >> >> Message: 4 >> Date: Mon, 14 Dec 2015 07:39:19 +0100 >> From: nek5000-users at lists.mcs.anl.gov >> To: "nek5000-users at lists.mcs.anl.gov" >> >> Subject: Re: [Nek5000-users] freestream turbulence at INLET >> Message-ID: >> >> Content-Type: text/plain; charset="utf-8" >> >> Dear Rif, >> >> In our wing simulation the boundary layers are tripped using a volume >> force in the direction normal to the surface. This emulates the approach in >> wind tunnel experiments, where the models are tripped using sand paper, V >> disturbances, etc. In such wind tunnel tests the transition location is >> prescribed, and the freestream turbulence intensity is very low. The idea >> is to have a ?canonical? turbulent boundary layer, developing in the most >> controlled way in order to study the effects of turbulence. Therefore, in >> our wing simulation turbulence is also zero at our boundaries. >> >> Regarding the use of Fluent data in Nek, could you give us some more >> information about your setup? Which RANS model are you using, and is it >> time-dependent? What kind of geometry are you simulating? >> >> Best regards, >> >> Ricardo Vinuesa. >> >> Sent from Mail for Windows 10 >> >> -------------- next part -------------- >> An HTML attachment was scrubbed... >> URL: < >> http://lists.mcs.anl.gov/pipermail/nek5000-users/attachments/20151214/f909c493/attachment.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 82, Issue 11 >> ********************************************* >> > > > _______________________________________________ > 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: Vinuesa_et_al_JoT_2015.pdf Type: application/pdf Size: 5354214 bytes Desc: not available URL: From nek5000-users at lists.mcs.anl.gov Mon Dec 14 02:29:25 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 09:29:25 +0100 Subject: [Nek5000-users] Nek5000-users Digest, Vol 82, Issue 11 Message-ID: Dear Rif, I think that a good starting point would be to use the time history of the RANS field as a time-varying inflow condition of your Nek simulation. We did that in one of our cases, where we imposed a time-varying inflow from a DNS to run another DNS, and the results were excellent. See attached paper for further details. Let us know how it goes, and if you need any additional help! Cheers, Ricardo. From nek5000-users at lists.mcs.anl.gov Mon Dec 14 12:35:16 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 13:35:16 -0500 Subject: [Nek5000-users] How to obtain spectral coefficients and basis In-Reply-To: References: Message-ID: Andrew, Thanks for the reply. And can you please confirm that the polynomials used are Chebyshev? On 12/8/2015 11:26 AM, nek5000-users at lists.mcs.anl.gov wrote: > Hello, > > If I've understood you propely, you can see all the collocation points > and weights in the file coef.f (subroutine genwz). > > Andrew > > 2015-12-07 4:23 GMT+01:00 >: > > Hello, > > I'm trying to represent the solution, either velocity or > temperature, in terms of spectral basis, i.e. > > $u=\Sigma a_i \phi_i $ with $\phi_i$ as the basis and $a_i$ as the > coefficients. > > I have two questions: > > 1) (a naive one!) Which basis is being used in Nek5000? I assume > it is Chebyshev polynomials. Is that true? If not, what is the basis? > > 2) How can I retrieve the coefficients $a_i$? Also, is there a way > to access such coefficients during the solution, say by writing > them to a file? > > I really appreciate the responses. As a side note, I'd like to use > DNS data for model reduction. > > Many thanks, > Saleh > _______________________________________________ > 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 Mon Dec 14 15:20:37 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 14 Dec 2015 21:20:37 +0000 Subject: [Nek5000-users] How to obtain spectral coefficients and basis In-Reply-To: References: , Message-ID: Nek is based on a Gauss-Legendre-Lobatto(GLL) quadrature, so the grids are GLL points. The basis polynomials are Lagrange polynomials. Oana ________________________________________basic 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: Monday, December 14, 2015 12:35 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] How to obtain spectral coefficients and basis Andrew, Thanks for the reply. And can you please confirm that the polynomials used are Chebyshev? On 12/8/2015 11:26 AM, nek5000-users at lists.mcs.anl.gov wrote: Hello, If I've understood you propely, you can see all the collocation points and weights in the file coef.f (subroutine genwz). Andrew 2015-12-07 4:23 GMT+01:00 >: Hello, I'm trying to represent the solution, either velocity or temperature, in terms of spectral basis, i.e. $u=\Sigma a_i \phi_i $ with $\phi_i$ as the basis and $a_i$ as the coefficients. I have two questions: 1) (a naive one!) Which basis is being used in Nek5000? I assume it is Chebyshev polynomials. Is that true? If not, what is the basis? 2) How can I retrieve the coefficients $a_i$? Also, is there a way to access such coefficients during the solution, say by writing them to a file? I really appreciate the responses. As a side note, I'd like to use DNS data for model reduction. Many thanks, Saleh _______________________________________________ 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 Wed Dec 16 00:11:35 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 16 Dec 2015 14:11:35 +0800 Subject: [Nek5000-users] freestream turbulence at INLET Message-ID: HI Ricardo, What is the title of the paper again? I could not see an attachment. Cheers Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Dec 16 03:19:37 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 16 Dec 2015 10:19:37 +0100 Subject: [Nek5000-users] freestream turbulence at INLET In-Reply-To: References: Message-ID: Sorry, the paper is: R. Vinuesa, P. Schlatter, J. Malm, C. Mavriplis and D. S. Henningson. Direct numerical simulation of the flow around a wall-mounted square cylinder under various inflow conditions. J. Turbul., 16, 555?587, 2015. I am also attaching it, hopefully it goes through. Best, Ricardo. On Wed, Dec 16, 2015 at 7:11 AM, wrote: > HI Ricardo, > What is the title of the paper again? I could not see an attachment. > > Cheers > Rif > > _______________________________________________ > 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: Vinuesa_et_al_JoT_2015.pdf Type: application/pdf Size: 5354214 bytes Desc: not available URL: From nek5000-users at lists.mcs.anl.gov Wed Dec 16 13:41:49 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 16 Dec 2015 20:41:49 +0100 (CET) Subject: [Nek5000-users] (no subject) Message-ID: Hi nek's I'm trying to perform a DNS of a turbulent channel flow ussing the tutorial "turbChannel.f". I have some troubles about the subroutine "planar_average_s", in particular when I generate a mesh using genbox in a single box, it works very well, such that the vector "yy" goes from -1 to 1. On the contrary when i split the box in genbox in two boxes, "yy" goes from -0.5 to 0.5 and it is repeated two rimes (-0.5, ... 0.5, -0.5, ..., 0.5), of course the averages have no sense. Could someone help me? Thanks --------------------------------------------------------------------------------------------------------- Box 1 -48 16 -48 nelx,nely,nelz for Box 0.0 6.28 1. x0,x1,gain -1. -.999 -.998 -.995 -.988 -.974 -.943 -.877 -.780 -.682 -.584 -.487 -.389 -.292 -.194 -.096 0.0 0.0 3.14 1. z0,z1,gain P ,P ,W ,E ,P ,P bc's (3 chars each!) Box 2 -48 16 -48 nelx,nely,nelz for Box 0.0 6.28 1. x0,x1,gain 0.0 .096 .194 .292 .389 .487 .584 .682 .780 .877 .943 .974 .988 .995 .998 .999 1. 0.0 3.14 1. z0,z1,gain P ,P ,E ,W ,P ,P bc's (3 chars each!) ---------------------------------------------------------------------------------------------------------- -- Mr. Mirko FaranoPh.D. student at Politecnico di BariTel.: +39-0883529289 / +39-0805963462Mob.: +39-3202342719Department of Mechanics Mathematics and Management - Fluid Machinery and Energy SystemsVia Re David, 200 - 70125 Bari (Italy)mirko.farano at poliba.itIn joint supervision with theEcole Nationale Sup?rieure d'Arts et M?tiers de Paris - ParistechTel.: +33-144246436Mob.: +33-789822134Dynfluid Laboratory151 Boulevard de l'H?pital - 75013 Paris (France)mirko.farano at ensam.eu -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Dec 16 13:46:04 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 16 Dec 2015 20:46:04 +0100 (CET) Subject: [Nek5000-users] problems genbox Message-ID: Sorry, I forgot the object -- Mr. Mirko FaranoPh.D. student at Politecnico di BariTel.: +39-0883529289 / +39-0805963462Mob.: +39-3202342719Department of Mechanics Mathematics and Management - Fluid Machinery and Energy SystemsVia Re David, 200 - 70125 Bari (Italy)mirko.farano at poliba.itIn joint supervision with theEcole Nationale Sup?rieure d'Arts et M?tiers de Paris - ParistechTel.: +33-144246436Mob.: +33-789822134Dynfluid Laboratory151 Boulevard de l'H?pital - 75013 Paris (France)mirko.farano at ensam.eu ----Messaggio originale---- Da: nek5000-users at lists.mcs.anl.gov Data: 16-dic-2015 20.41 A: Ogg: [Nek5000-users] (no subject) Hi nek's I'm trying to perform a DNS of a turbulent channel flow ussing the tutorial "turbChannel.f". I have some troubles about the subroutine "planar_average_s", in particular when I generate a mesh using genbox in a single box, it works very well, such that the vector "yy" goes from -1 to 1. On the contrary when i split the box in genbox in two boxes, "yy" goes from -0.5 to 0.5 and it is repeated two rimes (-0.5, ... 0.5, -0.5, ..., 0.5), of course the averages have no sense. Could someone help me? Thanks --------------------------------------------------------------------------------------------------------- Box 1 -48 16 -48 nelx,nely,nelz for Box 0.0 6.28 1. x0,x1,gain -1. -.999 -.998 -.995 -.988 -.974 -.943 -.877 -.780 -.682 -.584 -.487 -.389 -.292 -.194 -.096 0.0 0.0 3.14 1. z0,z1,gain P ,P ,W ,E ,P ,P bc's (3 chars each!) Box 2 -48 16 -48 nelx,nely,nelz for Box 0.0 6.28 1. x0,x1,gain 0.0 .096 .194 .292 .389 .487 .584 .682 .780 .877 .943 .974 .988 .995 .998 .999 1. 0.0 3.14 1. z0,z1,gain P ,P ,E ,W ,P ,P bc's (3 chars each!) ---------------------------------------------------------------------------------------------------------- -- Mr. Mirko FaranoPh.D. student at Politecnico di BariTel.: +39-0883529289 / +39-0805963462Mob.: +39-3202342719Department of Mechanics Mathematics and Management - Fluid Machinery and Energy SystemsVia Re David, 200 - 70125 Bari (Italy)mirko.farano at poliba.itIn joint supervision with theEcole Nationale Sup?rieure d'Arts et M?tiers de Paris - ParistechTel.: +33-144246436Mob.: +33-789822134Dynfluid Laboratory151 Boulevard de l'H?pital - 75013 Paris (France)mirko.farano at ensam.eu -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Dec 16 22:53:51 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 17 Dec 2015 12:53:51 +0800 Subject: [Nek5000-users] freestream turbulence at INLET Message-ID: Hi Ricardo, Thanks for the paper. Were you in Canmore for the CFD challenge? I was there as well! My prof organized that event. Small world. Cheers Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Dec 17 01:42:42 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 17 Dec 2015 08:42:42 +0100 Subject: [Nek5000-users] freestream turbulence at INLET In-Reply-To: References: Message-ID: Actually my colleague Johan Malm attended the meeting. Small world indeed!! Cheers, Ricardo. Enviado desde mi iPhone > El 17 dec 2015, a las 05:53, nek5000-users at lists.mcs.anl.gov escribi?: > > Hi Ricardo, > Thanks for the paper. Were you in Canmore for the CFD challenge? I was there as well! My prof organized that event. Small world. > > > Cheers > Rif > _______________________________________________ > 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 Dec 17 21:03:27 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 18 Dec 2015 03:03:27 +0000 Subject: [Nek5000-users] Stress formulation In-Reply-To: References: , Message-ID: Hi Andrew Are you running the "axi" case in the examples? This case started with u=0 as an initial condition. A quick glance at the output it produces in f00004, after 200 steps, shows plots similar to yours. If you run longer (change nsteps in the .rea file), the viscous boundary layer will diffuse inward from the walls and you should see a parabolic profile. This is what I observed just now. Please let me know if this resolves the issue --- if not, I'm happy to check further. Best, Paul ________________________________ From: Andrey Souchkov [andrey.souchkov at gmail.com] Sent: Thursday, December 17, 2015 7:07 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Stress formulation Hi Paul, I think I have misspoken. The thing is that I haven't got the profile corresponding to the inlet one anywhere in pipe. I add some clarifications. You can find attached file with velocity curves for my simulaltions. I have done the simulations of Poiseuille flow in the cylindrical pipe. With IFAXIS = true, IFSTRS = false, param (30) = 0 the input profile doesn't chage. With IFAXIS = true, IFSTSR = true, param (30) = 0 the input velocity profile changes considerably along the pipe and on the axis there is strange pressure peak, the nature of which I can not understand. Can I have clarification on this question? Thanks, Andrew 2015-12-14 0:55 GMT+01:00 >: Hi Andrew, In my view, you are already getting the correct flow physics. What happens at outflow, especially at moderate Reynolds number, is not necessarily parallel flow. The stress formulation is giving the stress-free result. 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, December 13, 2015 5:19 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Stress formulation Hi Paul, Thank you for your reply. Yes, if i increase Reynolds number, the outlet profile become much close to the inlet one. But in my simulations the Reynolds number is between 300 and 1200. Can you recommend me what kind of boundary conditions i can utilise for ensure for ensure the flow physics ? I would implement these one in code. Thanks, Andrew 2015-12-05 3:37 GMT+01:00 >: Hi Andrew, I checked into this... I think what is happening is that the 'O ' bc for the stress formulation means stress-free, which is not guaranteed to yield a parabolic profile at the outlet. If you increase your Reynolds number I'm guessing that you'll recover the parabolic profile because the viscous stresses will diminish -- this is what I observed. >From my perspective, the outlet boundary is not a region where I would count on accurate physics --- it is, after all, a truncated domain, so I don't generally worry too much about the behavior there. 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: Tuesday, December 01, 2015 2:44 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Stress formulation Hi Neks, I have one question about stress formulation in NEK5000. I tried to make a simulation for Poiseuil flow with stress formulation. In inlet boundary condition use was made of parabolic profile. If IFSTRS=false the outlet profile is exactly the same as inlet one but in case of IFSTRS=true the outlet profile changes significantly (both the maximum value and profile shape). Could you tell me how I can get the parabolic profile at outlet, please? The same thing occurs with pressure values. The maximum pressure value with IFSTRS=true is three times higher then the one with IFSTRS=false. I found that the difference comes from subroutines where the stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for IFSTRS = true) but I don't undestand what is happening exactly. Thanks in advance for your help, Best regards, Andrew _______________________________________________ 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 Dec 17 19:07:47 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 18 Dec 2015 02:07:47 +0100 Subject: [Nek5000-users] Stress formulation In-Reply-To: References: Message-ID: Hi Paul, I think I have misspoken. The thing is that I haven't got the profile corresponding to the inlet one anywhere in pipe. I add some clarifications. You can find attached file with velocity curves for my simulaltions. I have done the simulations of Poiseuille flow in the cylindrical pipe. With IFAXIS = true, IFSTRS = false, param (30) = 0 the input profile doesn't chage. With IFAXIS = true, IFSTSR = true, param (30) = 0 the input velocity profile changes considerably along the pipe and on the axis there is strange pressure peak, the nature of which I can not understand. Can I have clarification on this question? Thanks, Andrew 2015-12-14 0:55 GMT+01:00 : > > Hi Andrew, > > In my view, you are already getting the correct flow physics. > > What happens at outflow, especially at moderate Reynolds number, is not > necessarily parallel flow. > > The stress formulation is giving the stress-free result. > > 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, December 13, 2015 5:19 PM > *To:* nek5000-users at lists.mcs.anl.gov > *Subject:* Re: [Nek5000-users] Stress formulation > > Hi Paul, > > Thank you for your reply. > Yes, if i increase Reynolds number, the outlet profile become much close > to the inlet one. But in my simulations the Reynolds number is between 300 > and 1200. > Can you recommend me what kind of boundary conditions i can utilise for > ensure for ensure the flow physics ? I would implement these one in code. > > Thanks, > Andrew > > > 2015-12-05 3:37 GMT+01:00 : > >> >> Hi Andrew, >> >> I checked into this... I think what is happening is that the 'O ' bc for >> the stress formulation >> means stress-free, which is not guaranteed to yield a parabolic profile >> at the outlet. >> >> If you increase your Reynolds number I'm guessing that you'll recover the >> parabolic >> profile because the viscous stresses will diminish -- this is what I >> observed. >> >> From my perspective, the outlet boundary is not a region where I would >> count on accurate >> physics --- it is, after all, a truncated domain, so I don't generally >> worry too much about >> the behavior there. >> >> 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:* Tuesday, December 01, 2015 2:44 AM >> *To:* nek5000-users at lists.mcs.anl.gov >> *Subject:* [Nek5000-users] Stress formulation >> >> Hi Neks, >> >> I have one question about stress formulation in NEK5000. >> I tried to make a simulation for Poiseuil flow with stress >> formulation. In inlet boundary condition use was made of parabolic profile. >> If IFSTRS=false the outlet profile is exactly the same as inlet one but in >> case of IFSTRS=true the outlet profile changes significantly (both the >> maximum value and profile shape). Could you tell me how I can get the >> parabolic profile at outlet, please? >> The same thing occurs with pressure values. The maximum pressure >> value with IFSTRS=true is three times higher then the one with IFSTRS=false. >> >> I found that the difference comes from subroutines where the >> stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for >> IFSTRS = true) but I don't undestand what is happening exactly. >> >> Thanks in advance for your help, >> Best regards, >> Andrew >> >> _______________________________________________ >> 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: curves.pdf Type: application/pdf Size: 67619 bytes Desc: not available URL: From nek5000-users at lists.mcs.anl.gov Fri Dec 18 10:19:59 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 18 Dec 2015 16:19:59 +0000 Subject: [Nek5000-users] Stress formulation In-Reply-To: References: , Message-ID: Hi Andrew Are you running the "axi" case in the examples? This case started with u=0 as an initial condition. A quick glance at the output it produces in f00004, after 200 steps, shows plots similar to yours. If you run longer (change nsteps in the .rea file), the viscous boundary layer will diffuse inward from the walls and you should see a parabolic profile. This is what I observed just now. Please let me know if this resolves the issue --- if not, I'm happy to check further. 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, December 17, 2015 7:07 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Stress formulation Hi Paul, I think I have misspoken. The thing is that I haven't got the profile corresponding to the inlet one anywhere in pipe. I add some clarifications. You can find attached file with velocity curves for my simulaltions. I have done the simulations of Poiseuille flow in the cylindrical pipe. With IFAXIS = true, IFSTRS = false, param (30) = 0 the input profile doesn't chage. With IFAXIS = true, IFSTSR = true, param (30) = 0 the input velocity profile changes considerably along the pipe and on the axis there is strange pressure peak, the nature of which I can not understand. Can I have clarification on this question? Thanks, Andrew 2015-12-14 0:55 GMT+01:00 >: Hi Andrew, In my view, you are already getting the correct flow physics. What happens at outflow, especially at moderate Reynolds number, is not necessarily parallel flow. The stress formulation is giving the stress-free result. 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, December 13, 2015 5:19 PM To: nek5000-users at lists.mcs.anl.gov Subject: Re: [Nek5000-users] Stress formulation Hi Paul, Thank you for your reply. Yes, if i increase Reynolds number, the outlet profile become much close to the inlet one. But in my simulations the Reynolds number is between 300 and 1200. Can you recommend me what kind of boundary conditions i can utilise for ensure for ensure the flow physics ? I would implement these one in code. Thanks, Andrew 2015-12-05 3:37 GMT+01:00 >: Hi Andrew, I checked into this... I think what is happening is that the 'O ' bc for the stress formulation means stress-free, which is not guaranteed to yield a parabolic profile at the outlet. If you increase your Reynolds number I'm guessing that you'll recover the parabolic profile because the viscous stresses will diminish -- this is what I observed. >From my perspective, the outlet boundary is not a region where I would count on accurate physics --- it is, after all, a truncated domain, so I don't generally worry too much about the behavior there. 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: Tuesday, December 01, 2015 2:44 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Stress formulation Hi Neks, I have one question about stress formulation in NEK5000. I tried to make a simulation for Poiseuil flow with stress formulation. In inlet boundary condition use was made of parabolic profile. If IFSTRS=false the outlet profile is exactly the same as inlet one but in case of IFSTRS=true the outlet profile changes significantly (both the maximum value and profile shape). Could you tell me how I can get the parabolic profile at outlet, please? The same thing occurs with pressure values. The maximum pressure value with IFSTRS=true is three times higher then the one with IFSTRS=false. I found that the difference comes from subroutines where the stiffness matrix is calculated (axehlm for IFSTRS = false and axhmsf for IFSTRS = true) but I don't undestand what is happening exactly. Thanks in advance for your help, Best regards, Andrew _______________________________________________ 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 Dec 24 09:35:02 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 24 Dec 2015 16:35:02 +0100 Subject: [Nek5000-users] Grid 2 Grid Interpolation Message-ID: Hi Neks, I am using the routine " Grid 2 Grid Interpolation " , it does interpolate data between two meshes, but there is another problem: Both the "base.fld" and "new_geo.fld" are double-precision, however, the interpolated "new.fld" is single precision. show the head here: #std 4 8 8 8 31200 31200 0.1969999999996E+03 0 0 1 XUPT - I have already active the parameter(63) to 8, trying to save the double-precision result. Any one have a solution? thanks! Wei - -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Dec 29 23:28:37 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 30 Dec 2015 13:28:37 +0800 Subject: [Nek5000-users] Coriolis Force Message-ID: Hi Neks, I tried to include a rotation in my simulation for a flat plate by introducing a Coriolis term in the subroutine *userf* like so : --------------------------------------------------------------------------------------- subroutine userf (ix,iy,iz,eg) include 'SIZE' include 'TOTAL' include 'NEKUSE' integer e,f,eg e = gllel(eg) Omega= 2 ffx = +2.0*Omega*vy ffy = -2.0*Omega*vx ffz = 0.0 return end ----------------------------------------------------------------------------------------------- PROBLEM: When i tried compiling, it gives me an error ERROR: Incompatible ranks 0 and 4 in assignment at (1) Is there a way out of this? Does this have something to do with assigning a scalar to an array? if so how do I get past this? Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Dec 29 23:51:59 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 30 Dec 2015 05:51:59 +0000 Subject: [Nek5000-users] Coriolis Force In-Reply-To: References: Message-ID: Hi Rif, You should use "ux" and "uy" ... which are the point wise values for the velocity that are passed in via NEKUSE. 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: Tuesday, December 29, 2015 11:28 PM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Coriolis Force Hi Neks, I tried to include a rotation in my simulation for a flat plate by introducing a Coriolis term in the subroutine userf like so : --------------------------------------------------------------------------------------- subroutine userf (ix,iy,iz,eg) include 'SIZE' include 'TOTAL' include 'NEKUSE' integer e,f,eg e = gllel(eg) Omega= 2 ffx = +2.0*Omega*vy ffy = -2.0*Omega*vx ffz = 0.0 return end ----------------------------------------------------------------------------------------------- PROBLEM: When i tried compiling, it gives me an error ERROR: Incompatible ranks 0 and 4 in assignment at (1) Is there a way out of this? Does this have something to do with assigning a scalar to an array? if so how do I get past this? Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Dec 31 01:27:55 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 31 Dec 2015 15:27:55 +0800 Subject: [Nek5000-users] Sym boundary condition in mshconvert file Message-ID: Hi Neks, I have downloaded the program mshconvert file to convert my Fluent mesh to nek5000 format. It works well with normal boundary conditions like W,v and O where W is wall, v is inlet velocity and O is outlet. However, the I am unable to prescribe symmetry, SYM as the code only takes the last letter (for this case, M). Any of you facing this problem and is there a way around this? Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Dec 31 01:43:41 2015 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 31 Dec 2015 07:43:41 +0000 Subject: [Nek5000-users] Sym boundary condition in mshconvert file In-Reply-To: References: Message-ID: Rif, I would try in your .usr file: subroutine usrdat : : integer e,f nface = 2*ndim do e=1,nelt do f=1,nface if (cbc(e,f,1).eq.'M ') cbc(e,f,1) = 'SYM' enddo enddo (this is assuming that the "M" gets loaded into the first position of the 3-character string cbc; you could add: if (cbc(e,f,1).eq.' M ') cbc(e,f,1)='SYM' if (cbc(e,f,1).eq.' M') cbc(e,f,1)='SYM' if you want to cover all bases. One important point --- "SYM" requires that the face in question to be planar with a normal pointing in either X Y or Z direction... If it it pointing somewhere in between you would have to run with the stress formulation. 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, December 31, 2015 1:27 AM To: nek5000-users at lists.mcs.anl.gov Subject: [Nek5000-users] Sym boundary condition in mshconvert file Hi Neks, I have downloaded the program mshconvert file to convert my Fluent mesh to nek5000 format. It works well with normal boundary conditions like W,v and O where W is wall, v is inlet velocity and O is outlet. However, the I am unable to prescribe symmetry, SYM as the code only takes the last letter (for this case, M). Any of you facing this problem and is there a way around this? Rif -------------- next part -------------- An HTML attachment was scrubbed... URL: