- Nektar-users - imperial.ac.uk

NekMesh: msh to xml wrong space dimension
by Alexander Schukmann 14 Nov '24

14 Nov '24

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Hi all, when converting the attached 2D Gmsh mesh from msh to xml using NekMesh, I receive the following error: [InputGmsh] Reading Gmsh file: 'mesh.msh' [InputGmsh] Finished reading mesh. [InputGmsh] - Element dimension : 2 [InputGmsh] - Space dimension : 1 [InputGmsh] - No. of nodes : 2211 [InputGmsh] - No. of 2D elements : 2000 [InputGmsh] - No. of boundary elements : 420 [InputGmsh] - Elapsed time: 0.00769036s. [OutputNekpp] Writing Nektar++ file 'mesh.xml' Fatal : Level 0 assertion violation Coordinate dimension should be at least 2 for a 2D geometry terminate called after throwing an instance of 'Nektar::ErrorUtil::NekError' what(): Level 0 assertion violation Coordinate dimension should be at least 2 for a 2D geometry Aborted (core dumped) NekMesh does not read the correct space dimension for some reason. The mesh was created with ICEM, exported as cgns and converted to msh format with Gmsh version 4.5.2. I use this approach regularly and this is the first time this issue showed up. Any help will be very much appreciated! All the best Alex Sicher versendet mit [Proton Mail](https://proton.me/mail/home).

1 0

Openmp slow in Nektar
by Gao 11 Nov '24

11 Nov '24

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Hello everyone I run the Compressible solver in parallel, when I used 120 cores, it was slower than 68 cores. I used the commond that "mpirun -n 64 CmpressibleSolver Cyl.xml ". Could someone give me some advice? The computer has 128 cores, 2 NUMA. With best regards

2 1

Get source from CFD
by Gao 10 Nov '24

10 Nov '24

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Hello, I want to simulate the 2D cylinder in a crossflow in CAA. First, I need to compute the cylinder case by Compressible solver, then use the result of Compressible solver as the source of Acoustic solver. But how can I get the Lamb vector in CFD simulation? I think the Lamb vector in 2D is zero, am I right？ Could someone give me some advice? With best regards

2 1

Dimension of geometric curvature to include for a 3D simulation
by Debbahi Saad 10 Nov '24

10 Nov '24

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear Nektar users, I hope you're all doing fine, i'm using the incompressible navier-stokes solver, in fully 3D case with modified basis, and I have a question about the level at which adding curved geometry information, under <CURVED>....</CURVED> xml node, should be done. *Question:* is it necessary to add surface curvature information or should one be just fine with adding curved edges information only? Regarding this, I have noticed that exporting gmsh higher order mesh produces curved edges and curved faces under <CURVED>....</CURVED> section, which typically reads: > <E ID="45953" EDGEID="26270" NUMPOINTS="3" > TYPE="PolyEvenlySpaced">2.54571429e-01 -7.50000000e-03 -6.42857143e-04 > 2.50785714e-01 -7.50000000e-03 -6.96428571e-04 2.47000000e-01 > -7.50000000e-03 -7.50000000e-04 </E> > <F ID="45954" FACEID="26260" NUMPOINTS="9" > TYPE="PolyEvenlySpaced">-1.46714286e-01 7.50000000e-03 3.00000000e-02 > -1.50500000e-01 7.50000000e-03 3.00000000e-02 -1.54285714e-01 > 7.50000000e-03 3.00000000e-02 -1.46714286e-01 1.12500000e-02 > 3.00000000e-02 -1.50500000e-01 1.12500000e-02 3.00000000e-02 > -1.54285714e-01 1.12500000e-02 3.00000000e-02 -1.46714286e-01 > 1.50000000e-02 3.00000000e-02 -1.50500000e-01 1.50000000e-02 > 3.00000000e-02 -1.54285714e-01 1.50000000e-02 3.00000000e-02 </F> > Since I'm adding curvature information using an external python script, should I be worried about explicitly adding the face curvature as produced inside an auto-generated session file through NekMesh? A first naïve thought I had was that the Edges curvature information would propagate the curvature to higher dimensions through the tensor product !? However, I'm not even close to being formal on this statement. Thank you all in advance, Kind regards, Saad

2 2

Failed to calculate the z-direction force in pipe flow
by 小已 05 Nov '24

05 Nov '24

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear Nektar++ developers and users I am a phd student from BIT(Beijing institute of technology), and I am using nektar++ to simulat turbulent pipe flow recently. my simulation is a 3DH1D.The mesh is a circular. The flow direction is z-direction and expand with Fourier transform.The caculation only outputs the forces in the x and y directions and the moment of the z-direction. However,I want to get the force in the z-direction on the wall and the program is unable to output. Is there any methods to solve it? The mesh and parameter files are attached.

2 1

Defining a Restart File
by Selman BAYSAL 05 Nov '24

05 Nov '24

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear All,I am modeling laminar-to-turbulent transition in oscillatory boundary layers using IncNavierStokesSolver. I run this model for an half-period and intend to set the output obtained (say&nbsp;test.chk) as the "Initial Condition" in the subsequent model. To achieve this, I execute the following command:mpirun -np 28 FieldConvert test.xml test.chk test.fldThen, in the XML file (say&nbsp;new.xml), I define&lt;FUNCTION NAME="InitialConditions"&gt;&lt;F FILE="test.fld"&gt;&lt;/FUNCTION&gt;However, when I checked the wall shear stress (wss) outputs, I noticed that the second run may ignore the initial condition (test.fld). What might be causing this issue? What is the proper way to define a restart file? What are the possible solutions (e.g., if there is any other file format that I can use as an initial condition to restart the model, would you please explain how to obtain this format)?&nbsp;Attached is an example of the wall shear stress signal (x-component) signal. This curve should be sinusoidal when the flow is laminar. One can see in the red-circled area that the flow behaves as the model starts from the very beginning (as in the blue-circled area) like there is no initial condition file.&nbsp;Thank you very much for your time and intention.Best regards,Selman BaysalR. A. in HydraulicsDept. of Civil Eng'gÇanakkale Onsekiz Mart University--ITU Civil Eng., BSc.ITU Coastal Sci. &amp; Eng., MSc.

2 1

Question about Hybird CAA
by feigao2024 05 Nov '24

05 Nov '24

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Hello, everyone I saw some posts regarding the CAA in Nektar, I have some questions. Also, I want to couple the Compressible solver and Acoustic solver. First, how can I get the Lamb vector in Compressible solver? Second, if I export the result average file from the thirdparty CFD solver, then how can I transform the result file to .fld or .pts? Is there anything else I should be aware of about CAA in Nektar? With best regards

3 3

Failed to read nek5000 data
by Wei He 04 Nov '24

04 Nov '24

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Dear Nektar++ developers and users, I am planing to use 3D flow datas generated by nek5000. When I run the command on a centos machine, the process fails with errors, as shown below: FieldConvert roCyl_half0.f00100 cyl.fld -------------------------------------------------------------------------- WARNING: There was an error initializing an OpenFabrics device. Local host: master Local device: mlx5_0 -------------------------------------------------------------------------- Field contains unknown variable: X [master:42932] *** Process received signal *** [master:42932] Signal: Aborted (6) [master:42932] Signal code: (-6) [master:42932] [ 0] /lib64/libpthread.so.0(+0xf630)[0x7f85cfcf4630] [master:42932] [ 1] /lib64/libc.so.6(gsignal+0x37)[0x7f85cf94d387] [master:42932] [ 2] /lib64/libc.so.6(abort+0x148)[0x7f85cf94ea78] [master:42932] [ 3] /share/home/user1/nektar++/build/dist/lib64/libFieldUtils.so.5.6.0(_ZN6Nektar10FieldUtils12InputNek50009v_ProcessERN5boost15program_options13variables_mapE+0xe70)[0x7f85d3db2590] [master:42932] [ 4] FieldConvert[0x41e615] [master:42932] [ 5] FieldConvert[0x41b47e] [master:42932] [ 6] /lib64/libc.so.6(__libc_start_main+0xf5)[0x7f85cf939555] [master:42932] [ 7] FieldConvert[0x41dad2] [master:42932] *** End of error message *** Aborted (core dumped) While I also tried a 2D case on my laptop, it neither doesn't pass the procession. I wonder if I did the correct things, and thank you in advance! regards, Wei FieldConvert cylinder2D0.f00100 cyl.fld Field contains unknown variable: X [Harqin:55513] *** Process received signal *** [Harqin:55513] Signal: Abort trap: 6 (6) [Harqin:55513] Signal code: (0) [Harqin:55513] [ 0] 0 libsystem_platform.dylib 0x00000001878fe584 _sigtramp + 56 [Harqin:55513] [ 1] 0 libsystem_pthread.dylib 0x00000001878cdc20 pthread_kill + 288 [Harqin:55513] [ 2] 0 libsystem_c.dylib 0x00000001877daa30 abort + 180 [Harqin:55513] [ 3] 0 libFieldUtils.5.5.0.dylib 0x0000000100a784dc _ZN6Nektar10FieldUtils12InputNek50009v_ProcessERN5boost15program_options13variables_mapE + 5284 [Harqin:55513] [ 4] 0 FieldConvert 0x00000001002f7b5c _Z9RunModuleNSt3__110shared_ptrIN6Nektar10FieldUtils6ModuleEEERN5boost15program_options13variables_mapEb + 820 [Harqin:55513] [ 5] 0 FieldConvert 0x00000001002f24e8 main + 18272 [Harqin:55513] [ 6] 0 dyld 0x0000000187543154 s tart + 2476 [Harqin:55513] *** End of error message *** zsh: abort FieldConvert cylinder2D0.f00100 cyl.fld

2 2

Nektar++ 5.7.0 Released.
by Chris Cantwell 31 Oct '24

31 Oct '24

Dear all, We are pleased to announce the release of Nektar++ v5.7.0. As always, if you encounter any problems please feel free to contact us on the mailing list nektar-users(a)imperial.ac.uk. You can also report bugs and issues on our issue tracker (https://gitlab.nektar.info/nektar/nektar/-/issues) This release includes a range of new features and improvements over the 5.6.0 release: **Library** - Modified MatrixFreeOp library switch initialisation to use BOOST_PP (!1794) - Fix memory-leak with LowEnergyBlock preconditioner for time-updated matrices (!1627) - Fix Fourier expansion integration weights are related test (!1803) - Introduced the MatrixFree implementation for the PhysInterp1DScaled operator and tidying up MatrixFreeOps (!1812) - Separate MeshGraph input/output functions into a new class (!1778) - Added checkpoint file writing start time in the fieldconvert filter (!1789) - Fix fieldconvert filter incorrect boundary values (!1789) - Fix numerical precision issues with filters OutputStartTime (!1789) - Fix AdaptiveSFD for MPI (!1821) - Fix interpolation on manifold (!1840) - Fix IterativeStaticCond when using absolute tolerance (!1850) - Fix deadlock by scotch with multi-threading support (!1853) - Fixed L2norm for FilterError (!1871) - Fix variable p in tetrahedrons (!1881) - Fix BwdTrans for Pyr with var P (!1886) - Allow wrapper array around a existing raw pointer (!1848) - Tweaked some long tests to make them faster (!!1918) **IncNavierStokesSolver** - Fix initial and boundary conditions in the moving reference frame (!1692, ! 1820) - Fix memory-leak for the Mixed_CG_Discontinuous projection when initializing the traceMep (!1806) - Add synthetic turbulence generation for the incompressible solver (!1664) - Fix a uninitialized parameter in VCS (!1880) **ShallowWaterSolver** - Implement implicit time-discritization (!1784) **CompressibleSolver** - Add synthetic turbulence generator for the compressible solver (!1859) **NekMesh** - Added revolve module (!1825) - Fix Prism Reordering in Process PerAlign (!1899) - Extend quality measures in ProcessJac and add histogram generation(!1751) - Reducing run time of some tests in NekMesh(!1922) - Extend quality measures in ProcessJac and add histogram generation (!1751) - Reducing run time of some tests in NekMesh (!1922) - Added a reader for the CGNS input format (!1889) **FieldConvert** - Add vortexinducedvelocity module to compute the vortex-induced velocity (! 1824) - Add a module to transform coordinates and vectors for the moving reference frame method (!1830) **Miscellaneous** - Use std::stod instead of boost::lexical_cast<NekDouble> (!1819) **Documentation** - Add initial documentation for the IncNavierStokesSolver (!1822) - Updated the supported packages in Userguid (!1904) - Added a example for RayleighBenardConvection in the user-guide for IncNS (! 1919) - Fix some typos in tutorials (!1929) **CI and Packaging** - Debian 10 (BUSTER) is no longer supported (!1902) - Support is added for Ubuntu Noble Numbat and droped for Bionic Beaver (!1910) - Removed Fedora 35/36, added Fedora 39/40 (!1909) Cheers, Chris -- Chris Cantwell Reader in Computational Engineering Department of Aeronautics Imperial College London South Kensington Campus London SW7 2AZ Tel: +44 (0)20 759 45050 Email: c.cantwell(a)imperial.ac.uk www.imperial.ac.uk/people/c.cantwell

1 0

Error when restarting - ReynoldsStresses
by Daniel Garcia 24 Oct '24

24 Oct '24

******************* This email originates from outside Imperial. Do not click on links and attachments unless you recognise the sender. If you trust the sender, add them to your safe senders list https://spam.ic.ac.uk/SpamConsole/Senders.aspx to disable email stamping for this address. ******************* Hello everyone, how are you? I'm receiving an error (not specified) when restarting a 3D simulation which has the ReynoldsStresses filter activated. I don't get the error when the simulation is 2D nor when I use the AverageFields filter. So it just happens when I use the ReynoldsStresses filter in the Quasi-3D simulation. I'm also solving a scalar equation. Would it be a bug in this specific setup condition? I tried versions 5.4 and 5.5. Or does anyone know a workaround? I'm doing separated averages to combine later with the respective FieldConvert module for now. Thank you very much, Daniel.

2 2