From nek5000-users at lists.mcs.anl.gov Mon Oct 1 07:40:05 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Mon, 1 Oct 2018 13:40:05 +0100 Subject: [Nek5000-users] Problems in getting eigenvalues using ARNOLDI In-Reply-To: References: Message-ID: An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Oct 3 01:24:57 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 3 Oct 2018 11:54:57 +0530 Subject: [Nek5000-users] Problems in getting eigenvalues using ARNOLDI In-Reply-To: References: Message-ID: Hi Guillaume, Thanks for your reply. My reply to your queries is as below:- 1) I am getting unstable pair of Eigen values with growth rate around 0.02. 2) Solver tolerance that I used are ; 7.00000 p020 NORDER 0.100000E-09 p021 DIVERGENCE 0.100000E-09. p022 HELMHOLTZ 0.00000 p023 NPSCAL 0.100000E-09 p024 TOLREL 0.100000E-04 p025 TOLABS 1.00000E-05. p113 : ARNOLDI: arpack tolerance 3) Boundary conditions are : I am using sponge strength at inlet and outlet with following details in .rea file 1.00000 p119 : SPONGE STRENGTH 25.0000 p120 : SPONGE TOTAL WIDTH 3.00000 p121 : SPONGE DROP WIDTH (INFLOW) 5.00000 p122 : SPONGE RISE WIDTH (OUTFLOW) the sponge function and its cal is detailed in userf subroutine of .usr file Also following is defined in .usr file c----------------------------------------------------------------------- subroutine userbc (ix,iy,iz,iside,eg) include 'SIZE' include 'NEKUSE' ! UX, UY, UZ, TEMP, X, Y include 'PARALLEL' ! GLLEL integer e,eg c velocity c e = GLLEL(eg) UX = 0.0 UY = 0.0 UZ = 0.0 c t TEMP=0.0 return end c----------------------------------------------------------------------- subroutine useric (ix,iy,iz,ieg) include 'SIZE' include 'NEKUSE' ! UX, UY, UZ, TEMP, Z integer idum save idum data idum / 0 / real eps if (idum.eq.0) idum = 99 + nid eps = 0.1 c velocity c random distribution UX = eps*(ran1(idum)-0.50) UY = eps*(ran1(idum)-0.50) UZ = eps*(ran1(idum)-0.50) c t TEMP=0 return end 40.0000 p109 : ARNOLDI: frequency of calling arn_solve 4.00000 p110 : ARNOLDI: maximal number of arnoldi cycles 90.0000 p111 : ARNOLDI: size of Krylov space 30.0000 p112 : ARNOLDI: number of eigenvectors If I am increasing the Krylov subspace then will I get Eigen values at lesser restart ? Kindly help please as I am new to this. Thanks in advance. Alok Mishra Research Scholar Computational Propulsion Lab Aerospace Department IIT Kanpur +91-8795844555 -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Oct 3 04:24:05 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 3 Oct 2018 10:24:05 +0100 Subject: [Nek5000-users] Problems in getting eigenvalues using ARNOLDI In-Reply-To: References: Message-ID: An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Oct 4 09:49:14 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 4 Oct 2018 14:49:14 +0000 Subject: [Nek5000-users] Divergent issue in Accelerating BL Message-ID: Dear Nek users, I ran into an odd issue I do not quite understand. I am working on an accelerating BL with relatively high RE (Re_{\delta*}_inlet = 8000) and this is how I did it: I modified the available Blasius code in the NEK5000 such that it gives me the analytical profile for accelerating BL. Instead of using the pressure gradient to drive the flow, I used the analytical solutions as BCs at inlet and freestream. This approach works very well for low RE (Re_{\delta*}_inlet ~ < 1500), however it gives me issue when I increase the RE (by decreasing viscosity P02); CFL increased significantly and started to fluctuate. If delta_t is large (~0.1) then eventually it quits for CFL being too large. If delta_t is small enough (~0.01) then it kept fluctuating. I plotted instantaneous u and v contours (I am doing 2D case)and I noticed many vertical streaks of unphysically high velocities across the domain, which is clearly a numerical issue. I am not quite sure what is the reason contributing to this phenomenon. Could anyone help me on this? Many thanks. -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Oct 4 10:04:06 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 4 Oct 2018 15:04:06 +0000 Subject: [Nek5000-users] Divergent issue in Accelerating BL Message-ID: Dear NEK users, In the previous email, I forgot to mention that the variation and fluctuations of CFL did not really occur till after roughly 50,000 time steps (\delta_t = 0.0125); before that it works very well. Thanks in advance for the help -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Oct 4 11:39:47 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 04 Oct 2018 18:39:47 +0200 Subject: [Nek5000-users] Divergent issue in Accelerating BL In-Reply-To: References: Message-ID: Hi, Are you setting the timestep adaptively through a fixed CFL or fixing the timestep ? Also you may need to refine your mesh while going up in Re number specially if you are running a DNS. A well known scaling for the number of grid points in your mesh is N ~ Re**(9./4). Sincerely, Armand. Le 2018-10-04 17:04, nek5000-users at lists.mcs.anl.gov a ?crit : > Dear NEK users, > > In the previous email, I forgot to mention that the variation and fluctuations of CFL did not really occur till after roughly 50,000 time steps (\delta_t = 0.0125); before that it works very well. > > Thanks in advance for the help > > _______________________________________________ > 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 Oct 4 12:23:49 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 4 Oct 2018 17:23:49 +0000 Subject: [Nek5000-users] Divergent issue in Accelerating BL Message-ID: Hello Armand, Thank you for your reply. In the simulation, the time step is fixed. It is in fact a laminar case, not turbulent. This simulation is intended to generate a base flow for the instability analysis for the next step. So the divergence really bothers me. Thanks a lot -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Thu Oct 4 15:32:46 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 4 Oct 2018 20:32:46 +0000 Subject: [Nek5000-users] varvis example? Message-ID: Hi, I am looking for the ?varvis? example that used to be in the old nek5_svn. Can someone please upload the example in the website? Thank you, Hannah -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Oct 8 22:25:34 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 9 Oct 2018 03:25:34 +0000 Subject: [Nek5000-users] 'COMMON' block issue Message-ID: Dear Nek users, Currently, I am working on a simulation with a decent size (11760 * 12 elements) employing ?lpmin = 32?. When I compile, I received an error saying: ?relocation overflows omitted from the output?. I googled this error and it turned out this error comes from the ?COMMON? block; Static data, such as COMMON variables, and all of the code together is still limited to 2GB under any mcmodel. The suggested solution is to move the data out of a COMMON block and into a module and make the data allocatable. The most straightforward way to avoid this would be increase the ?lpmin? (for example if lpmin = 64, then it works), but this does not solve this issue, which is a big constraint. I am wondering has anyone has run into this and got any experience in solving it? Thanks a lot in advance -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Oct 9 01:29:12 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 9 Oct 2018 08:29:12 +0200 Subject: [Nek5000-users] 'COMMON' block issue In-Reply-To: References: Message-ID: you exceed 2 GB memory. Either compile with more processors or use mcmodel=big flags (something like big-memory for Nek, I forgot). On 2018-10-09 05:25, nek5000-users at lists.mcs.anl.gov wrote: > Dear Nek users, > > Currently, I am working on a simulation with a decent size (11760 * 12 > elements) employing ?lpmin = 32?. When I compile, I received an error > saying: ?relocation overflows omitted from the output?. ?I googled this > error and it turned out this error comes from the ?COMMON? block; Static > data, such as COMMON variables, and all of the code together is still > limited to 2GB under any mcmodel.? The suggested solution is to ?move > the data out of a COMMON block and into a module and make the data > allocatable. > > The most straightforward way to avoid this would be increase the ?lpmin? > (for example if lpmin = 64, then it works), but this does not solve this > issue, which is a big constraint. I am wondering has anyone has run into > this and got any experience in solving it? > > Thanks a lot in advance > > > > _______________________________________________ > 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 Oct 9 01:56:07 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 9 Oct 2018 08:56:07 +0200 Subject: [Nek5000-users] 'COMMON' block issue In-Reply-To: References: Message-ID: Check our FAQs: https://nek5000.github.io/NekDoc/faq.html -----Original message----- > From:nek5000-users at lists.mcs.anl.gov > Sent: Tuesday 9th October 2018 5:25 > To: NEK5000 > Subject: [Nek5000-users] 'COMMON' block issue > > Dear Nek users, > Currently, I am working on a simulation with a decent size (11760 * 12 elements) employing ?lpmin = 32?. When I compile, I received an error saying: ?relocation overflows omitted from the output?. ?I googled this > error and it turned out this error comes from the ?COMMON? block; ?Static data, such as COMMON variables, and all of the code together is still limited to 2GB under any mcmodel.? The suggested solution is to ?move > the data out of a COMMON block and into a module and make the data allocatable. > The most straightforward way to avoid this would be increase the ?lpmin? (for example if lpmin = 64, then it works), but this does not solve this issue, which is a > big constraint. I am wondering has anyone has run into this and got any experience in solving it? > Thanks a lot in advance > _______________________________________________ > 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 Oct 10 09:39:47 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Oct 2018 14:39:47 +0000 Subject: [Nek5000-users] Sub-grid turbulent kinetic energy Message-ID: Hi, I'm trying to compute the efficacy of my LES and would like to do this by considering the ratio of turbulence kinetic energy both on the grid and sub-grid. Is there an option within NEK (maybe in .par) to include the amount of sub-grid TKE as an output from the simulation itself, so that I can directly compare it to the computed on-grid TKE? Cheers, Daniel Ward -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Oct 10 10:27:29 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Oct 2018 17:27:29 +0200 Subject: [Nek5000-users] Sub-grid turbulent kinetic energy In-Reply-To: References: Message-ID: Hi, what SGS model are you using? Just the filtering? Philipp On October 10, 2018 4:39:47 PM GMT+02:00, nek5000-users at lists.mcs.anl.gov wrote: >Hi, > > >I'm trying to compute the efficacy of my LES and would like to do this >by considering the ratio of turbulence kinetic energy both on the grid >and sub-grid. Is there an option within NEK (maybe in .par) to include >the amount of sub-grid TKE as an output from the simulation itself, so >that I can directly compare it to the computed on-grid TKE? > > >Cheers, > > >Daniel Ward -- Sent from my Android device with K-9 Mail. Please excuse my brevity. -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Sat Oct 13 10:38:31 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Sat, 13 Oct 2018 11:38:31 -0400 Subject: [Nek5000-users] Rotating computational domain In-Reply-To: References: Message-ID: Hi Paul and Phillip, I have validated my model at constant AOA and am now returning to my pitch and hold scenario to do some work. I definitely think the ALE formulation applied to the entire domain is the way to go, but I'm feeling a little confused about the "mv" boundary condition. In the mv_wall example, the "mv" boundary condition is applied to the section of the wall that is moving, so in my mind it follows logically that to have the entire domain pitch about the z axis, the entire domain would need the "mv" condition. This doesn't seem possible while preserving my inlet and outlet BCs. I understand that for ALE to work appropriately, we need to match the normal component of fluid velocity to that of the mesh velocity at any boundary. Does this apply also to the inlet and outlet, or only internal fluid boundaries such as my airfoil? I am also looking at the mv_cyl and ocyl/ocyl2 examples, but these are in rea/re2 format and I cannot look at the element boundary conditions (forgive me, I'm still learning terminology so some of what I say is not 100% accurate). While looking through the example files, I see that the logical switch "ifusermv" is assigned in "usrchk" in some of the examples, and in one of the "usrdat" subroutines in others. I don't really appreciate the difference, is there one? I am building my mesh in ANSYS Fluent, and then converting the mesh to ASCII .rea format with mshconvert2, and then extruding to 3D and converting to rea/re2 format with n2to3. Before, I would scroll to the bottom of the .msh file and add Nek5000 BC flags to each boundary (i.e. adding an _W to wall, _O to outlet, and _v to inlet) before converting to .rea format, and this would successfully apply boundary conditions. However, when attempting to insert the _mv boundary condition on my wall, the boundary conditions in the .rea file are unchanged. Is there a better/more appropriate way to assign the "mv" BC when converting a mesh? On Tue, Aug 21, 2018 at 11:30 AM wrote: > > Hi Harry, > > > One simple approach is to use ALE but to have the entire domain pitch, > > thus preserving your mesh structure. > > > That would properly account for all the accelerations in your non inertial > > domain. > > > The slightly tricky part would be to determine what BCs you need. You > > need to be careful about changing the character of the BCs (e.g., from > inflow > > to outflow) in the middle of a computation because that changes which > quantities > > are prescribed and which are unknown. Perhaps one approach would be > > to have a circular domain with 3/4 of it Dirichlet and 1/4 being outflow > (i.e., > > Neumann for velocity). > > > hth, > > > Paul > > > ------------------------------ > *From:* Nek5000-users on behalf > of nek5000-users at lists.mcs.anl.gov > *Sent:* Tuesday, August 21, 2018 9:43:56 AM > *To:* nek5000-users at lists.mcs.anl.gov > *Subject:* [Nek5000-users] Rotating computational domain > > Hello Neks, > > I am an undergraduate student new to Nek and I am working on a simulation > studying dynamic stall in a dynamically pitching airfoil. I want to do a > pitch and hold simulation at low Reynold's number (Re = 12000) where I > rotate the entire computational domain up to about 50 degrees AOA and then > hold at 50 degrees, with steady inflow ux = 1. I would like to rotate the > domain about the z axis, which passes through the quarter chord location of > the airfoil profile (modified NACA 0012). > > Because of the steep AOA, I would like to stay away from pitching the > airfoil within the stationary domain with ALE to avoid too much mesh > deformation. To this end, I have been scouring documentation to try and > find a method of applying a rotational mesh velocity to the entire domain > about the z axis while keeping the flow within the domain at ux = 1. I am > feeling very stuck. Others have suggested applying a Coriolis force to the > flow, but I don't really see how this would simulate rotation of the entire > domain. It seems to me that simulation of a Coriolis force would be > acceptable to correct the flow field to always be ux = 1 inside a rotating > domain, but I don't see how that would apply a mesh velocity anywhere. > > Are there any examples where an entire computational domain is rotated > that I can look at, or has anybody done similar work who can point me in > the right direction? > > Regards, > -Harry > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > -- Harry E. Werner IV Secretary Clarkson University SCUBA Club *Cell:* (716) 570-4023 -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Oct 16 09:01:44 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 16:01:44 +0200 (CEST) Subject: [Nek5000-users] Boundary nodes cordinates and random generator Message-ID: Hello Neks, Its been a long time that I posted something in NEK5000. Here I came up with something which is simple. First thing is, I was trying to access the co ordinates of the inlet boundary elements for my free stream turbulence model(Von karman). I have an idea to get the using cbc(f,e,1) but I just would like to know is there any quick way to get the inlet bounday nodes coordinate. Second thing is about the making of random numbers. I know that in fortran there is an inbuilt function to generate random numbers (call random_generator(iseed)) but it seems it is an expensive process since my number of modes of turbulence model is 150. So just I have a question that how you usually generate random numbers when you model free stream turbulence? By the way I have completed the FSI calculation of NACA0012 at Re 64000 in NEK5000. Thanks for your help. Regards Sijo GEORGE -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Oct 16 09:38:12 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 16:38:12 +0200 Subject: [Nek5000-users] Boundary nodes cordinates and random generator In-Reply-To: References: Message-ID: Hi, how would your inflow turbulence model work? Do you need to create 150 random numbers every time step? Philipp On 2018-10-16 16:01, nek5000-users at lists.mcs.anl.gov wrote: > Hello Neks, > > Its been a long time that I posted something in NEK5000. Here I came up > with something which is simple. > First thing is, I was trying to access the co ordinates of the inlet > boundary elements for my free stream turbulence model(Von karman). > > I have an idea to get the using cbc(f,e,1) but I just would like to know > is there any quick way to get the inlet bounday nodes coordinate. > > Second thing is about the making of random numbers. > I know that in fortran there is an inbuilt function to generate random > numbers (call random_generator(iseed)) but it seems it is an expensive > process since my number of modes of turbulence model is 150. > > So just I have a question that how you usually generate random numbers > when you model free stream turbulence? > > By the way I have completed the FSI calculation of NACA0012 at Re 64000 > in NEK5000. Thanks for your help. > > Regards > > Sijo GEORGE > > > > > > > > _______________________________________________ > 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 Oct 16 09:49:37 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 16:49:37 +0200 (CEST) Subject: [Nek5000-users] Boundary nodes cordinates and random generator In-Reply-To: References: Message-ID: Hello Philip, The turbulent velocity field is generated by random Fourier modes. So, yes I have to generate 150 random numbers for 4 different variables(Angles) at each time step. Or do you have any other simple way to suggest to make it computationally cheaper? Thanks Sijo ----- Mail original ----- De: "nek5000-users" ?: "nek5000-users" Envoy?: Mardi 16 Octobre 2018 16:38:12 Objet: Re: [Nek5000-users] Boundary nodes cordinates and random generator Hi, how would your inflow turbulence model work? Do you need to create 150 random numbers every time step? Philipp On 2018-10-16 16:01, nek5000-users at lists.mcs.anl.gov wrote: > Hello Neks, > > Its been a long time that I posted something in NEK5000. Here I came up > with something which is simple. > First thing is, I was trying to access the co ordinates of the inlet > boundary elements for my free stream turbulence model(Von karman). > > I have an idea to get the using cbc(f,e,1) but I just would like to know > is there any quick way to get the inlet bounday nodes coordinate. > > Second thing is about the making of random numbers. > I know that in fortran there is an inbuilt function to generate random > numbers (call random_generator(iseed)) but it seems it is an expensive > process since my number of modes of turbulence model is 150. > > So just I have a question that how you usually generate random numbers > when you model free stream turbulence? > > By the way I have completed the FSI calculation of NACA0012 at Re 64000 > in NEK5000. Thanks for your help. > > Regards > > Sijo GEORGE > > > > > > > > _______________________________________________ > 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 Oct 16 11:15:33 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 18:15:33 +0200 Subject: [Nek5000-users] Boundary nodes cordinates and random generator In-Reply-To: References: Message-ID: But I guess you want some temporal correlation as well, i.e. your modes will evolve as a harmonic signal also in time? Typically, we chose a few modes (say 150) with random wavenumbers and phases, but then we just evolve them in space and time. So the problem is then rather to evaluate the complex exponential (i.e. cosine and sine signals) than random numbers. So in that case, one needs to have a RNG that works well in parallel and you can restart a simulation without "kinks". But if you do not need these properties, I would say that computing 150 random numbers is not very expensive. I typically use a quite "simple" generator from Numerical recipes, just because it is portable and gives the same numbers on all machines. Philipp On 2018-10-16 16:49, nek5000-users at lists.mcs.anl.gov wrote: > Hello Philip, > > The turbulent velocity field is generated by random Fourier modes. So, yes I have to generate 150 random numbers for 4 different variables(Angles) at each time step. Or do you have any other simple way to suggest to make it computationally cheaper? > > Thanks > > Sijo > > ----- Mail original ----- > De: "nek5000-users" > ?: "nek5000-users" > Envoy?: Mardi 16 Octobre 2018 16:38:12 > Objet: Re: [Nek5000-users] Boundary nodes cordinates and random generator > > Hi, how would your inflow turbulence model work? Do you need to create > 150 random numbers every time step? > > Philipp > > On 2018-10-16 16:01, nek5000-users at lists.mcs.anl.gov wrote: >> Hello Neks, >> >> Its been a long time that I posted something in NEK5000. Here I came up >> with something which is simple. >> First thing is, I was trying to access the co ordinates of the inlet >> boundary elements for my free stream turbulence model(Von karman). >> >> I have an idea to get the using cbc(f,e,1) but I just would like to know >> is there any quick way to get the inlet bounday nodes coordinate. >> >> Second thing is about the making of random numbers. >> I know that in fortran there is an inbuilt function to generate random >> numbers (call random_generator(iseed)) but it seems it is an expensive >> process since my number of modes of turbulence model is 150. >> >> So just I have a question that how you usually generate random numbers >> when you model free stream turbulence? >> >> By the way I have completed the FSI calculation of NACA0012 at Re 64000 >> in NEK5000. Thanks for your help. >> >> Regards >> >> Sijo GEORGE >> >> >> >> >> >> >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > From nek5000-users at lists.mcs.anl.gov Tue Oct 16 13:31:31 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 20:31:31 +0200 (CEST) Subject: [Nek5000-users] Boundary nodes cordinates and random generator In-Reply-To: References: Message-ID: Yes I need some temporal correlation as well. In fact for me evaluating the complex exponential part is okay, the problem was computational expense for generating random number for 150 modes for theta,Phi,Psi and Alfa. I used the built in function of fortran and I experience Segmentation fault in my computer and when I use the code from Numerical recipes (subroutine ran1) it works well. But I just want to know why this happens and what are the other possible solutions. By using the ran1 code from Numerical recipe, 150X4 = 600 random numbers are generated at each time step for the random Fourier mode velocities. Or is there anyway solutions to cut down the number of random numbers by slightly any change in the Fourier mode? And finally regarding the co ordinates of the inlet boundary is there any quick way to get it? Regards Sijo GEORGE Your replay for my last question: But I guess you want some temporal correlation as well, i.e. your modes will evolve as a harmonic signal also in time? Typically, we chose a few modes (say 150) with random wavenumbers and phases, but then we just evolve them in space and time. So the problem is then rather to evaluate the complex exponential (i.e. cosine and sine signals) than random numbers. So in that case, one needs to have a RNG that works well in parallel and you can restart a simulation without "kinks". But if you do not need these properties, I would say that computing 150 random numbers is not very expensive. I typically use a quite "simple" generator from Numerical recipes, just because it is portable and gives the same numbers on all machines. Philipp From nek5000-users at lists.mcs.anl.gov Tue Oct 16 13:34:55 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 20:34:55 +0200 Subject: [Nek5000-users] Boundary nodes cordinates and random generator In-Reply-To: References: Message-ID: Hi, I typically use ran1 in fact. It is simple, and portable. Of course, not too many different numbers are created, but should be enough for the purposes of inflow conditions. But, perhaps I am not sure I understand fully, you would need to create these random numbers just once when you start the simulation, and then just evaluate the complex exponentials? At least in my implementation I do not re-create random numbers. For the coordinates, I guess you get them in userbc, where you can check, depending on your geometry, on what boundary you are. Philipp On 2018-10-16 20:31, nek5000-users at lists.mcs.anl.gov wrote: > > Yes I need some temporal correlation as well. In fact for me evaluating the complex exponential part is okay, the problem was computational expense for generating random number for 150 modes for theta,Phi,Psi and Alfa. I used the built in function of fortran and I experience Segmentation fault in my computer and when I use the code from Numerical recipes (subroutine ran1) it works well. But I just want to know why this happens and what are the other possible solutions. > > By using the ran1 code from Numerical recipe, 150X4 = 600 random numbers are generated at each time step for the random Fourier mode velocities. > > Or is there anyway solutions to cut down the number of random numbers by slightly any change in the Fourier mode? > > And finally regarding the co ordinates of the inlet boundary is there any quick way to get it? > > > Regards > > Sijo GEORGE > > > Your replay for my last question: > > But I guess you want some temporal correlation as well, i.e. your modes > will evolve as a harmonic signal also in time? Typically, we chose a few > modes (say 150) with random wavenumbers and phases, but then we just > evolve them in space and time. So the problem is then rather to evaluate > the complex exponential (i.e. cosine and sine signals) than random numbers. > > So in that case, one needs to have a RNG that works well in parallel and > you can restart a simulation without "kinks". > > But if you do not need these properties, I would say that computing 150 > random numbers is not very expensive. I typically use a quite "simple" > generator from Numerical recipes, just because it is portable and gives > the same numbers on all machines. > > Philipp > > _______________________________________________ > 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 Oct 16 13:43:57 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 20:43:57 +0200 (CEST) Subject: [Nek5000-users] Boundary nodes cordinates and random generator In-Reply-To: References: Message-ID: Ohh really? But if you don't re-create new random numbers at each time step then it will give you the same velocity filed at all time step right? Or in that case, the time correlation helps to avoid this problem? I feel we have to generate different random numbers at each time step to maintain the turbulence nature of the flow. Sijo ----- Mail original ----- De: "nek5000-users" ?: "nek5000-users" Envoy?: Mardi 16 Octobre 2018 20:34:55 Objet: Re: [Nek5000-users] Boundary nodes cordinates and random generator Hi, I typically use ran1 in fact. It is simple, and portable. Of course, not too many different numbers are created, but should be enough for the purposes of inflow conditions. But, perhaps I am not sure I understand fully, you would need to create these random numbers just once when you start the simulation, and then just evaluate the complex exponentials? At least in my implementation I do not re-create random numbers. For the coordinates, I guess you get them in userbc, where you can check, depending on your geometry, on what boundary you are. Philipp On 2018-10-16 20:31, nek5000-users at lists.mcs.anl.gov wrote: > > Yes I need some temporal correlation as well. In fact for me evaluating the complex exponential part is okay, the problem was computational expense for generating random number for 150 modes for theta,Phi,Psi and Alfa. I used the built in function of fortran and I experience Segmentation fault in my computer and when I use the code from Numerical recipes (subroutine ran1) it works well. But I just want to know why this happens and what are the other possible solutions. > > By using the ran1 code from Numerical recipe, 150X4 = 600 random numbers are generated at each time step for the random Fourier mode velocities. > > Or is there anyway solutions to cut down the number of random numbers by slightly any change in the Fourier mode? > > And finally regarding the co ordinates of the inlet boundary is there any quick way to get it? > > > Regards > > Sijo GEORGE > > > Your replay for my last question: > > But I guess you want some temporal correlation as well, i.e. your modes > will evolve as a harmonic signal also in time? Typically, we chose a few > modes (say 150) with random wavenumbers and phases, but then we just > evolve them in space and time. So the problem is then rather to evaluate > the complex exponential (i.e. cosine and sine signals) than random numbers. > > So in that case, one needs to have a RNG that works well in parallel and > you can restart a simulation without "kinks". > > But if you do not need these properties, I would say that computing 150 > random numbers is not very expensive. I typically use a quite "simple" > generator from Numerical recipes, just because it is portable and gives > the same numbers on all machines. > > Philipp > > _______________________________________________ > 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 Oct 16 13:53:54 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 20:53:54 +0200 Subject: [Nek5000-users] Boundary nodes cordinates and random generator In-Reply-To: References: Message-ID: If I understand well, then you would destroy the temporal coherency by creating new phases at every time step. The way that we have used is described (to some detail) in my very old report (https://www.kth.se/social/files/5bc6330556be5b1206650f12/schlatter_2001a.pdf), in particular eq. (2.71). The time dependence goes in via an omega. But of course, you might be using another way, in that disregard my comments. Philipp On 2018-10-16 20:43, nek5000-users at lists.mcs.anl.gov wrote: > Ohh really? But if you don't re-create new random numbers at each time step then it will give you the same velocity filed at all time step right? Or in that case, the time correlation helps to avoid this problem? > > I feel we have to generate different random numbers at each time step to maintain the turbulence nature of the flow. > > > Sijo > > ----- Mail original ----- > De: "nek5000-users" > ?: "nek5000-users" > Envoy?: Mardi 16 Octobre 2018 20:34:55 > Objet: Re: [Nek5000-users] Boundary nodes cordinates and random generator > > Hi, > I typically use ran1 in fact. It is simple, and portable. Of course, not > too many different numbers are created, but should be enough for the > purposes of inflow conditions. But, perhaps I am not sure I understand > fully, you would need to create these random numbers just once when you > start the simulation, and then just evaluate the complex exponentials? > At least in my implementation I do not re-create random numbers. > > For the coordinates, I guess you get them in userbc, where you can > check, depending on your geometry, on what boundary you are. > > Philipp > > On 2018-10-16 20:31, nek5000-users at lists.mcs.anl.gov wrote: >> >> Yes I need some temporal correlation as well. In fact for me evaluating the complex exponential part is okay, the problem was computational expense for generating random number for 150 modes for theta,Phi,Psi and Alfa. I used the built in function of fortran and I experience Segmentation fault in my computer and when I use the code from Numerical recipes (subroutine ran1) it works well. But I just want to know why this happens and what are the other possible solutions. >> >> By using the ran1 code from Numerical recipe, 150X4 = 600 random numbers are generated at each time step for the random Fourier mode velocities. >> >> Or is there anyway solutions to cut down the number of random numbers by slightly any change in the Fourier mode? >> >> And finally regarding the co ordinates of the inlet boundary is there any quick way to get it? >> >> >> Regards >> >> Sijo GEORGE >> >> >> Your replay for my last question: >> >> But I guess you want some temporal correlation as well, i.e. your modes >> will evolve as a harmonic signal also in time? Typically, we chose a few >> modes (say 150) with random wavenumbers and phases, but then we just >> evolve them in space and time. So the problem is then rather to evaluate >> the complex exponential (i.e. cosine and sine signals) than random numbers. >> >> So in that case, one needs to have a RNG that works well in parallel and >> you can restart a simulation without "kinks". >> >> But if you do not need these properties, I would say that computing 150 >> random numbers is not very expensive. I typically use a quite "simple" >> generator from Numerical recipes, just because it is portable and gives >> the same numbers on all machines. >> >> Philipp >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > From nek5000-users at lists.mcs.anl.gov Tue Oct 16 14:07:43 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 21:07:43 +0200 (CEST) Subject: [Nek5000-users] Boundary nodes cordinates and random generator In-Reply-To: References: Message-ID: That great. It is good to have a look on your work. But I was using a different method as I said using Fourier modes to generate the velocity field so the time correlation is also described in a different way. Even though I will have a deep look on your work. Thank you Sijo ----- Mail original ----- De: "nek5000-users" ?: "nek5000-users" Envoy?: Mardi 16 Octobre 2018 20:53:54 Objet: Re: [Nek5000-users] Boundary nodes cordinates and random generator If I understand well, then you would destroy the temporal coherency by creating new phases at every time step. The way that we have used is described (to some detail) in my very old report (https://www.kth.se/social/files/5bc6330556be5b1206650f12/schlatter_2001a.pdf), in particular eq. (2.71). The time dependence goes in via an omega. But of course, you might be using another way, in that disregard my comments. Philipp On 2018-10-16 20:43, nek5000-users at lists.mcs.anl.gov wrote: > Ohh really? But if you don't re-create new random numbers at each time step then it will give you the same velocity filed at all time step right? Or in that case, the time correlation helps to avoid this problem? > > I feel we have to generate different random numbers at each time step to maintain the turbulence nature of the flow. > > > Sijo > > ----- Mail original ----- > De: "nek5000-users" > ?: "nek5000-users" > Envoy?: Mardi 16 Octobre 2018 20:34:55 > Objet: Re: [Nek5000-users] Boundary nodes cordinates and random generator > > Hi, > I typically use ran1 in fact. It is simple, and portable. Of course, not > too many different numbers are created, but should be enough for the > purposes of inflow conditions. But, perhaps I am not sure I understand > fully, you would need to create these random numbers just once when you > start the simulation, and then just evaluate the complex exponentials? > At least in my implementation I do not re-create random numbers. > > For the coordinates, I guess you get them in userbc, where you can > check, depending on your geometry, on what boundary you are. > > Philipp > > On 2018-10-16 20:31, nek5000-users at lists.mcs.anl.gov wrote: >> >> Yes I need some temporal correlation as well. In fact for me evaluating the complex exponential part is okay, the problem was computational expense for generating random number for 150 modes for theta,Phi,Psi and Alfa. I used the built in function of fortran and I experience Segmentation fault in my computer and when I use the code from Numerical recipes (subroutine ran1) it works well. But I just want to know why this happens and what are the other possible solutions. >> >> By using the ran1 code from Numerical recipe, 150X4 = 600 random numbers are generated at each time step for the random Fourier mode velocities. >> >> Or is there anyway solutions to cut down the number of random numbers by slightly any change in the Fourier mode? >> >> And finally regarding the co ordinates of the inlet boundary is there any quick way to get it? >> >> >> Regards >> >> Sijo GEORGE >> >> >> Your replay for my last question: >> >> But I guess you want some temporal correlation as well, i.e. your modes >> will evolve as a harmonic signal also in time? Typically, we chose a few >> modes (say 150) with random wavenumbers and phases, but then we just >> evolve them in space and time. So the problem is then rather to evaluate >> the complex exponential (i.e. cosine and sine signals) than random numbers. >> >> So in that case, one needs to have a RNG that works well in parallel and >> you can restart a simulation without "kinks". >> >> But if you do not need these properties, I would say that computing 150 >> random numbers is not very expensive. I typically use a quite "simple" >> generator from Numerical recipes, just because it is portable and gives >> the same numbers on all machines. >> >> Philipp >> >> _______________________________________________ >> Nek5000-users mailing list >> Nek5000-users at lists.mcs.anl.gov >> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users >> > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ > Nek5000-users mailing list > Nek5000-users at lists.mcs.anl.gov > https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users > _______________________________________________ Nek5000-users mailing list Nek5000-users at lists.mcs.anl.gov https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users From nek5000-users at lists.mcs.anl.gov Fri Oct 19 06:47:44 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 19 Oct 2018 11:47:44 +0000 Subject: [Nek5000-users] Sub-grid turbulent kinetic energy In-Reply-To: References: Message-ID: I'm currently using the hpfrt model but am planning to try out a couple (such as dynamic Smagorinsky) in the future. Cheers, Daniel ________________________________ From: Nek5000-users on behalf of nek5000-users-request at lists.mcs.anl.gov Sent: 10 October 2018 18:00 To: nek5000-users at lists.mcs.anl.gov Subject: Nek5000-users Digest, Vol 116, Issue 6 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. Sub-grid turbulent kinetic energy (nek5000-users at lists.mcs.anl.gov) 2. Re: Sub-grid turbulent kinetic energy (nek5000-users at lists.mcs.anl.gov) ---------------------------------------------------------------------- Message: 1 Date: Wed, 10 Oct 2018 14:39:47 +0000 From: nek5000-users at lists.mcs.anl.gov To: "nek5000-users at lists.mcs.anl.gov" Subject: [Nek5000-users] Sub-grid turbulent kinetic energy Message-ID: Content-Type: text/plain; charset="iso-8859-1" Hi, I'm trying to compute the efficacy of my LES and would like to do this by considering the ratio of turbulence kinetic energy both on the grid and sub-grid. Is there an option within NEK (maybe in .par) to include the amount of sub-grid TKE as an output from the simulation itself, so that I can directly compare it to the computed on-grid TKE? Cheers, Daniel Ward -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ Message: 2 Date: Wed, 10 Oct 2018 17:27:29 +0200 From: nek5000-users at lists.mcs.anl.gov To: nek5000-users at lists.mcs.anl.gov, "nek5000-users at lists.mcs.anl.gov" Subject: Re: [Nek5000-users] Sub-grid turbulent kinetic energy Message-ID: Content-Type: text/plain; charset="utf-8" Hi, what SGS model are you using? Just the filtering? Philipp On October 10, 2018 4:39:47 PM GMT+02:00, nek5000-users at lists.mcs.anl.gov wrote: >Hi, > > >I'm trying to compute the efficacy of my LES and would like to do this >by considering the ratio of turbulence kinetic energy both on the grid >and sub-grid. Is there an option within NEK (maybe in .par) to include >the amount of sub-grid TKE as an output from the simulation itself, so >that I can directly compare it to the computed on-grid TKE? > > >Cheers, > > >Daniel Ward -- Sent from my Android device with K-9 Mail. Please excuse my brevity. -------------- next part -------------- An HTML attachment was scrubbed... URL: ------------------------------ Subject: Digest Footer _______________________________________________ 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 116, Issue 6 ********************************************* -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Fri Oct 19 07:33:15 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Fri, 19 Oct 2018 14:33:15 +0200 Subject: [Nek5000-users] Sub-grid turbulent kinetic energy In-Reply-To: References: Message-ID: ok, for the hpf model the only way to get the SGS dissipation would be to calculate the TKE budget, and look for the missing contribution which is then due to the SGS term. For eddy viscosity models, you can calculate the SGS dissipation using eps_sgs=tau_ij * S_ij with tau_ij being the SGS tensor. Does this help? Philipp On 2018-10-19 13:47, nek5000-users at lists.mcs.anl.gov wrote: > I'm currently using the hpfrt model but am planning to try out a couple > (such as dynamic Smagorinsky) in the future.? Cheers, Daniel > > > ------------------------------------------------------------------------ > *From:* Nek5000-users on > behalf of nek5000-users-request at lists.mcs.anl.gov > > *Sent:* 10 October 2018 18:00 > *To:* nek5000-users at lists.mcs.anl.gov > *Subject:* Nek5000-users Digest, Vol 116, Issue 6 > 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. Sub-grid turbulent kinetic energy > ????? (nek5000-users at lists.mcs.anl.gov) > ?? 2. Re: Sub-grid turbulent kinetic energy > ????? (nek5000-users at lists.mcs.anl.gov) > > > ---------------------------------------------------------------------- > > Message: 1 > Date: Wed, 10 Oct 2018 14:39:47 +0000 > From: nek5000-users at lists.mcs.anl.gov > To: "nek5000-users at lists.mcs.anl.gov" > ??????? > Subject: [Nek5000-users] Sub-grid turbulent kinetic energy > Message-ID: > ??????? > Content-Type: text/plain; charset="iso-8859-1" > > Hi, > > > I'm trying to compute the efficacy of my LES and would like to do this > by considering the ratio of turbulence kinetic energy both on the grid > and sub-grid.? Is there an option within NEK (maybe in .par) to include > the amount of sub-grid TKE as an output from the simulation itself, so > that I can directly compare it to the computed on-grid TKE? > > > Cheers, > > > Daniel Ward > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: > > > ------------------------------ > > Message: 2 > Date: Wed, 10 Oct 2018 17:27:29 +0200 > From: nek5000-users at lists.mcs.anl.gov > To: nek5000-users at lists.mcs.anl.gov,??? "nek5000-users at lists.mcs.anl.gov" > ??????? > Subject: Re: [Nek5000-users] Sub-grid turbulent kinetic energy > Message-ID: > ??????? > Content-Type: text/plain; charset="utf-8" > > Hi, what SGS model are you using? Just the filtering? > Philipp > > On October 10, 2018 4:39:47 PM GMT+02:00, > nek5000-users at lists.mcs.anl.gov wrote: >>Hi, >> >> >>I'm trying to compute the efficacy of my LES and would like to do this >>by considering the ratio of turbulence kinetic energy both on the grid >>and sub-grid.? Is there an option within NEK (maybe in .par) to include >>the amount of sub-grid TKE as an output from the simulation itself, so >>that I can directly compare it to the computed on-grid TKE? >> >> >>Cheers, >> >> >>Daniel Ward > > -- > Sent from my Android device with K-9 Mail. Please excuse my brevity. > -------------- next part -------------- > An HTML attachment was scrubbed... > URL: > > > ------------------------------ > > Subject: Digest Footer > > _______________________________________________ > 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 116, Issue 6 > ********************************************* > > > _______________________________________________ > 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 Oct 23 23:35:48 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 24 Oct 2018 04:35:48 +0000 Subject: [Nek5000-users] Different P-order across domain Message-ID: Dear Nek users I have two quick questions: Q1) Is it possible to change the order of polynomial for different sections of the geometry? For example a pipe with an expansion in the downstream. _______________ /| | | / | | | /| | | | ________________/ | | | | | | | | | | | | | | | | | | | | +-----+-----+---+--+---+------+------+----> z <--- Pn=6 --><------ Pn=8 -----------> Q2)Would it be feasible to restart using a different P-order? Thanks for the help in advance -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Oct 23 23:44:22 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 24 Oct 2018 06:44:22 +0200 Subject: [Nek5000-users] Different P-order across domain In-Reply-To: References: Message-ID: No and yes. On October 24, 2018 6:35:48 AM GMT+02:00, nek5000-users at lists.mcs.anl.gov wrote: >Dear Nek users > > >I have two quick questions: > > >Q1) Is it possible to change the order of polynomial for different >sections of the geometry? For example a pipe with an expansion in the >downstream. > > > _______________ > /| | | > / | | | > /| | | | > ________________/ | | | | > | | | | | | | | > | | | | | | | | > +-----+-----+---+--+---+------+------+----> z > <--- Pn=6 --><------ Pn=8 -----------> > > >Q2)Would it be feasible to restart using a different P-order? > > >Thanks for the help in advance -- Sent from my Android device with K-9 Mail. Please excuse my brevity. -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Oct 30 00:40:19 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 30 Oct 2018 11:10:19 +0530 Subject: [Nek5000-users] Problems in getting eigenvalues using ARNOLDI In-Reply-To: References: Message-ID: Hi Guillaume, I tried with the help of your valuable suggestions. I am getting dominant Eigenvalue with time period 8.34 which is very much close to the literature. However, I require 30 Eigen values by setting P112=30 and P111=90. Below is the eigenvalues I am getting after many restarts. re(RITZ); im(RITZ) ln|RITZ|; arg(RITZ) 1 0.991128702238537 0.150881402754817 1.272083634754888E-002 0.755360071443281 2 0.991128702238537 -0.150881402754817 1.272083634754888E-002 -0.755360071443281 3 0.991728721496092 1.806383192432841E-002 -4.069909517841635E-002 9.106237555496137E-002 4 0.991728721496092 -1.806383192432841E-002 -4.069909517841635E-002 -9.106237555496137E-002 5 0.989791831070271 3.446186469202143E-002 -4.827437507304033E-002 0.174016133198715 6 0.989791831070271 -3.446186469202143E-002 -4.827437507304033E-002 -0.174016133198715 7 0.987601103156919 5.049392478862426E-000 -5.585540570484418E-002 0.255416865677410 8 0.987601103156919 -5.049392478862426E-002 -5.585540570484418E-002 -0.255416865677410 9 0.995627135254111 0.000000000000000E+000 -2.191226841613384E-002 0.000000000000000E+000 10 0.995700191328997 0.000000000000000E+000 -2.154539716426180E-002 0.000000000000000E+000 11 0.995229254767405 0.000000000000000E+000 -2.391080780851450E-002 0.000000000000000E+000 12 0.995434205850176 0.000000000000000E+00 -2.288124611941097E-002 0.000000000000000E+000 13 0.994877066906734 0.000000000000000E+000 -2.568050152041379E-002 0.000000000000000E+000 14 0.994526533813701 0.000000000000000E+000 -2.744250243633435E-002. 0.000000000000000E+000 15 0.994024234181460 0.000000000000000E+000 -2.996846079257938E-002. 0.000000000000000E+000 16 0.992886352948494 0.000000000000000E+000 -3.569534837742794E-002 0.000000000000000E+000 17 0.993531725861019 0.000000000000000E+000 -3.244642035911912E-002 0.000000000000000E+000 18 0.992269436061858 0.000000000000000E+000 -3.880299821483946E-002 0.000000000000000E+000 19 0.991568951835935 0.000000000000000E+000 -4.233395244633451E-002 0.000000000000000E+000 20 0.991491977086931 0.000000000000000E+000 -4.272211373739676E-002 0.000000000000000E+000 21 0.991513431383682 0.000000000000000E+000 -4.261392292430112E-002 0.000000000000000E+000 22 0.991180149347166 0.000000000000000E+00 -4.429487878603097E-002 0.000000000000000E+000 23 0.991284115416627 0.000000000000000E+000 -4.377045031887505E-002 0.000000000000000E+000 24 0.990874009258556 0.000000000000000E+000 -4.583943845265666E-002 0.000000000000000E+000 25 0.990632850108536 0.000000000000000E+000 -4.705648774052894E-002 0.000000000000000E+000 26 0.990852910254318 0.000000000000000E+000 -4.594590622089855E-002. 0.000000000000000E+000 27 0.990206173286647 0.000000000000000E+000 -4.921050845320660E-002 0.000000000000000E+000 28 0.990051445648789 0.000000000000000E+000 -4.999185950921724E-002 0.000000000000000E+000 29 0.989332783273003 0.000000000000000E+000 -5.362259677015029E-002 0.000000000000000E+000 30 0.989848247701740 0.000000000000000E+000 -5.101816376314156E-002 0.000000000000000E+000 31 0.989278105749528 0.000000000000000E+000 -5.389893975295782E-002 0.000000000000000E+000 I am unable to understand the reason behind getting so many Eigenvalues with zero imaginary part/rotation. Please help. Regards, On Wed, Oct 3, 2018 at 2:54 PM wrote: > Hi, > > What is your time step? I calculated the spectrum of the cylinder flow a > while ago at Re=44 and I got an angular frequency of 0.7278, which > corresponds to a period of 8.63. > It is possible that 40 time steps per Arnoldi iteration is too small; a > good start is to integrate for 1/10th of the period, so dt*p109 = 0.8 > approximately. I used 120 time steps at dt=4.2e-3 but it may be faster with > more steps. > It depends on which frequencies you want to resolve; if you integrate for > a short time your operator is close to identity and all eigenvalues are > close to 1, but if you integrate for too long then the frequencies of the > eigenvalues you are interested in will be aliased. > > I think you could try decreasing p021 and p022 to maybe 1e-10 or 1e-12 and > see if helps, too. > > Best, > Guillaume > > On 03/10/2018 07:24, nek5000-users at lists.mcs.anl.gov wrote: > > Hi Guillaume, > > Thanks for your reply. My reply to your queries is as below:- > > 1) I am getting unstable pair of Eigen values with growth rate around 0.02. > 2) Solver tolerance that I used are ; > 7.00000 p020 NORDER > 0.100000E-09 p021 DIVERGENCE > 0.100000E-09. p022 HELMHOLTZ > 0.00000 p023 NPSCAL > 0.100000E-09 p024 TOLREL > 0.100000E-04 p025 TOLABS > 1.00000E-05. p113 : ARNOLDI: arpack tolerance > > > 3) Boundary conditions are : > > I am using sponge strength at inlet and outlet with following details in > .rea file > 1.00000 p119 : SPONGE STRENGTH > 25.0000 p120 : SPONGE TOTAL WIDTH > 3.00000 p121 : SPONGE DROP WIDTH (INFLOW) > 5.00000 p122 : SPONGE RISE WIDTH (OUTFLOW) > the sponge function and its cal is detailed in userf subroutine of .usr > file > > Also following is defined in .usr file > c----------------------------------------------------------------------- > subroutine userbc (ix,iy,iz,iside,eg) > > include 'SIZE' > include 'NEKUSE' ! UX, UY, UZ, TEMP, X, Y > include 'PARALLEL' ! GLLEL > > > integer e,eg > > c velocity > c e = GLLEL(eg) > UX = 0.0 > UY = 0.0 > UZ = 0.0 > > c t > TEMP=0.0 > > return > end > c----------------------------------------------------------------------- > subroutine useric (ix,iy,iz,ieg) > > include 'SIZE' > include 'NEKUSE' ! UX, UY, UZ, TEMP, Z > > integer idum > save idum > data idum / 0 / > > real eps > > if (idum.eq.0) idum = 99 + nid > eps = 0.1 > > c velocity > c random distribution > UX = eps*(ran1(idum)-0.50) > UY = eps*(ran1(idum)-0.50) > UZ = eps*(ran1(idum)-0.50) > > c t > TEMP=0 > > return > end > > > 40.0000 p109 : ARNOLDI: frequency of calling arn_solve > 4.00000 p110 : ARNOLDI: maximal number of arnoldi cycles > 90.0000 p111 : ARNOLDI: size of Krylov space > 30.0000 p112 : ARNOLDI: number of eigenvectors > > > If I am increasing the Krylov subspace then will I get Eigen values at > lesser restart ? > > Kindly help please as I am new to this. Thanks in advance. > > Alok Mishra > Research Scholar > Computational Propulsion Lab > Aerospace Department IIT Kanpur > +91-8795844555 > > _______________________________________________ > Nek5000-users mailing listNek5000-users at lists.mcs.anl.govhttps://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 > -- Alok Mishra Research Scholar Computational Propulsion Lab Aerospace Department IIT Kanpur +91-8795844555 -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Wed Oct 31 02:37:48 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 31 Oct 2018 10:37:48 +0300 Subject: [Nek5000-users] =?utf-8?q?Variable_properties_and_Reynolds_numbe?= =?utf-8?q?r?= Message-ID: Hi, Neks! I have a question about setting Reynolds number and using variable properties in uservp (density, viscousity and etc.). I want to solve an dimensionless equation (1D case, for example): rho*Cp*(dT/dt + ux*(dT/dx))=(1/Pe)*d(k(dT/dx))/dx, where rho*Cp I set in uservp (utrans for ifield = 2) and it's variable, k I set in uservp (udiff for ifield = 2) and it's variable. According to the documentation for using uservp I must use "variableProperties = yes" in .par file, and for the Peclet number? I must use "conductivity = -3536.69" (minus for Pe), but I was looking at the code and noticed that both of "uservp" and "conductivity" use vtrans array. Does it mean that in case of variable properties I could solve only rho*Cp*(dT/dt + ux*(dT/dx))=d(k(dT/dx))/dx, without my Peclet number and I sould to include it in uservp in k, for example? Best regards, John. -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Mon Oct 8 22:37:09 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 09 Oct 2018 03:37:09 -0000 Subject: [Nek5000-users] 'COMMON' block issue In-Reply-To: References: Message-ID: When you define your object in the common block, do you define it so that it stays the same size regardless of how many threads you use? Presumably, it should look something like this: common /mydata/ data(lx1*ly1*lz1*lelt) So that when you double the number of threads, you halve the value of lelt On Mon, Oct 8, 2018 at 8:25 PM wrote: > Dear Nek users, > > > > Currently, I am working on a simulation with a decent size (11760 * 12 > elements) employing ?lpmin = 32?. When I compile, I received an error > saying: ?relocation overflows omitted from the output?. I googled this > error and it turned out this error comes from the ?COMMON? block; Static > data, such as COMMON variables, and all of the code together is still > limited to 2GB under any mcmodel. The suggested solution is to move the > data out of a COMMON block and into a module and make the data allocatable. > > The most straightforward way to avoid this would be increase the ?lpmin? > (for example if lpmin = 64, then it works), but this does not solve this > issue, which is a big constraint. I am wondering has anyone has run into > this and got any experience in solving it? > > > > Thanks a lot in advance > _______________________________________________ > 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 Oct 10 12:06:23 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Wed, 10 Oct 2018 17:06:23 -0000 Subject: [Nek5000-users] Genmap Error? Message-ID: Hello, I created a ".re2" mesh by converting an Exodus (".exo") mesh from Cubit 15.0 using the "exo2nek" script. It's a relatively large mesh (780k elements) and I need to run it using parallel processing, and so I wanted to use the "genmap" script. However, I keep getting these errors, no matter what input mesh tolerance value I use, from 0.2 to 0.00001): genmap Input .rea / .re2 name: RCF_3D_FullSimple reading RCF_3D_FullSimple.re2 Input mesh tolerance (default 0.2): NOTE: smaller is better, but generous is more forgiving for bad meshes. 0.00001 reading mesh data ... start locglob_lexico: 8 775728 6205824 1.0000000000000001E-005 locglob: 1 1 6205824 locglob: 2 11290 6205824 locglob: 3 15413 6205824 locglob: 1 632909 6205824 locglob: 2 632909 6205824 locglob: 3 632909 6205824 locglob: 1 632909 6205824 locglob: 2 632909 6205824 locglob: 3 632909 6205824 593741 2 4 Matrix: SELF!! 1 SELF!! 313837 313837 313837 313837 2 SELF!! 313837 313837 313837 313837 cont: SELF!! 593741 ABORT: SELF-CHK 1 2 593741 0 Try to tighten the mesh tolerance! 0 quit Do you have any idea what this means? Just from the output, it looks like it's saying there are coincident vertices or elements? Is it a problem with my mesh or with running the script? Thanks, Matt From nek5000-users at lists.mcs.anl.gov Thu Oct 11 02:31:34 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Thu, 11 Oct 2018 07:31:34 -0000 Subject: [Nek5000-users] Calculate the inverse laplacian Message-ID: Hi all, I am having some problems in calculating the inverse Laplacian within Nek5000. The equation I wanna solve looks like "laplacian(phi) = Q", here Q is known, the solutions of phi need to be obtained. I learnt that the Helmholtz equation includes laplacian operator. And in Nek5000, there is a subroutine named "hmholtz(name,u,rhs,h1,h2,mask,mult,imsh,tli,maxit,isd)", but I did not find any explanation on its usage. I am not sure whether it can be used for the solutions of inverse laplacian. Does anyone have an idea on computing the inverse laplacian? Any discussion would be appreciated. Thank you very much for your time. feng -------------- next part -------------- A non-text attachment was scrubbed... Name: winmail.dat Type: application/ms-tnef Size: 14024 bytes Desc: not available URL: From nek5000-users at lists.mcs.anl.gov Tue Oct 16 09:10:24 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 16 Oct 2018 14:10:24 -0000 Subject: [Nek5000-users] gradient at every point Message-ID: Hi, I am a new user. Want to calculate gradient at every point so as to obtain TKE values. Is there a subroutine for this? Thanks, Samuel -------------- next part -------------- An HTML attachment was scrubbed... URL: From nek5000-users at lists.mcs.anl.gov Tue Oct 30 08:36:16 2018 From: nek5000-users at lists.mcs.anl.gov (nek5000-users at lists.mcs.anl.gov) Date: Tue, 30 Oct 2018 13:36:16 +0000 Subject: [Nek5000-users] Remesh a moving geometry Message-ID: Hi Neks We try to simulate a flow with a moving geometry. In order to avoid large geometry distortions, we have to remesh the geometry at regular intervals. It is possible to remesh the geometry during a simulation without a restart? Best regards, Philipp -------------- next part -------------- An HTML attachment was scrubbed... URL: