- Nektar-users - imperial.ac.uk

implementation details of the adjoint linearised NSE
by Simon Schmidt 26 Sep '19

26 Sep '19

Hi everyone, This might be a trivial question but I have some trouble understanding the implementation of the linearised adjoint (incompressible) navier stokes equations in nektar++. From Barkley, Blackburn and Sherwin (2008) it seems to me that the only differences between the direct and adjoint equations, are switched signs for the time derivative (du/dt -> -du/dt) and one of the advective terms ((U nabla) u -> -(U nabla) u). From my understanding the necessary modifications to the existing direct solver are then to modify this advection term and to choose a negative time step to "reverse" the time integration. The modification to the advection term are taken into account in AdjointAdvection.cpp. However, I cannot find any reference in the code where the negative time derivative is considered. In the parameter files for the adjoint cylinder example the time step is as well positive. Am I just missing the respective line of code or are my assumptions wrong? I would really appreciate any insight to help me understand the procedure :) Cheers Simon

2 1

Convergence issues with compressible flow solver
by Thibaut Appel 20 Sep '19

20 Sep '19

Dear Nektar++ developers/users, We're encountering some difficulties with the compressible flow solver of Nektar++. We're trying to replicate a laminar flat plate case from SU2 (https://su2code.github.io/tutorials/Laminar_Flat_Plate/) where the free-stream Mach is 0.2 and with a rather high unit Reynolds number of 4.2 million. The structured mesh was generated with Gmsh, we impose a Blasius profile at the inlet and a similarity solution as the initial condition. This results in a transient wavepacket advected downstream with the iterations. To get the simulation running, we had to resort to using a very small time step (1e-8) but it still blows up after a certain number of iterations. You can find attached a screenshot of the last flow field (u,v) saved as well as our XML parameter file. We read carefully the user guide and went through the different threads of the mailing list, with some questions: - is it hopeless to try and solve such a high Reynolds number, regardless of the mesh resolution? Is it likely we're suffering from a under-resolution issue? - are there any other parameters implemented recently in the master branch that we could change? The compressible flow solver section of the user guide is not really verbose and for example, we found out about the PressureOutflowNonReflective b.c. Our XML file was set up with some inspiration from the 3D hump of De Grazia. - are there any sponge layers implemented for the compressible flow solver? - we noticed there is a CompressibleBL.cpp source file in the Utilities folder, could it be useful to compute straight away a similarity solution? What was the original purpose of that source file? We'd appreciate if you could notify us of anything suspicious in our XML file. Thank you for your support, Thibaut Appel PhD Student Virtual Product Engineering, Airbus CR&T Department of Mathematics, Imperial College London

1 0

Sponge-layer
by 蒋树杰 18 Sep '19

18 Sep '19

Hi all, First, I want to say thanks to everyone for helping me so much! Now I want to use a time-depentdent sponge-layer. Is there anyone using this function now? The simeluation will diverge very soon. In the simualtion, It is an empty domain and gust moves with flow. The boundary condition is the same with the reference fields. I use the Compressible Flow Solver with Euler euqations. My reference like this: <FUNCTION NAME="referencefields"> <E VAR="rho" VALUE="rhoInf" /> <E VAR="rhou" VALUE="rhoInf*uInf*(1-0.02*cos(k*x+k*y-k*uInf*t)*sqrt(2)/2)" /> <E VAR="rhov" VALUE="rhoInf*uInf*(0.02*cos(k*x+k*y-k*uInf*t)*sqrt(2)/2)" /> <E VAR="E" VALUE="pInf/(Gamma-1)+0.5*rhoInf*uInf*(1-0.02*cos(k*x+k*y-k*uInf*t)*sqrt(2)/2)*uInf*(1-0.02*cos(k*x+k*y-k*uInf*t)*sqrt(2)/2)+0.5*rhoInf*uInf*(0.02*cos(k*x+k*y-k*uInf*t)*sqrt(2)/2)*uInf*(0.02*cos(k*x+k*y-k*uInf*t)*sqrt(2)/2)" /> </FUNCTION> Of course with the <REFFLOWTIME> tag: <FORCING> <FORCE TYPE="Absorption"> <COEFF> SpongeCoefficient </COEFF> <REFFLOWTIME> referencefields </REFFLOWTIME> </FORCE> </FORCING> The reference field is the function of x, y and t (time-dependent). Is there any thing wrong with my reference fields? I test the reference fields without x,y and t. It works well. It looks like: <FUNCTION NAME="referencefields"> <E VAR="rho" VALUE="rhoInf" /> <E VAR="rhou" VALUE="rhoInf*uInf" /> <E VAR="rhov" VALUE="rhoInf*vInf" /> <E VAR="E" VALUE="pInf/(Gamma-1)+0.5*rhoInf*uInf^2+0.5*rhoInf*vInf^2" /> </FUNCTION> So I think maybe my sponge-layer coffecient is fine. If anyone has some idea please let me. I can send my case file and mesh file to you. Besides, I am using the Compressible Flow Solver to do some areoacoustic research. I found there is some reflection on the boundary when I use far-field boundary conditon. At first I think the far-field boundary condition is a non-reflection boundary conditions when relate with solving the Euler euqaitons.(in the paper: A guide to the implementation of boundary conditons in compact high-order methods for compressible aerodynamics). I do not know if there is any misunderstanding in that paper. So I want to use the sponge-layer to reduce the reflection. Kind regards, Shujie University of Southampton

1 1

Coordinate transformations
by 蒋树杰 15 Sep '19

15 Sep '19

Hi All, I am using the Coordinate transformations for generating the wavy wing.  <FUNCTION NAME="Mapping"> <E VAR="x" VALUE="x - wavyAmplitude/2 * cos(2*PI*z/wavyLambda)" /> </FUNCTION> <MAPPING TYPE="XofZ"> <COORDS>Mapping</COORDS> </MAPPING> When the computation is finished, I use FieldConvert -m mapping wavyWing-aoa12.xml wavyWing-aoa12.fld wavyWing-aoa12.plt to get the result. But in tecplot, I can not see the deformed geometry. The first picture is the slices on y=0. If someone is familiar with this, please no hesitate to contact me. Many thanks for your guys! Shujie University of Southampton

2 1

3DH1D interppoints module
by Guillermo 15 Sep '19

15 Sep '19

Dear all, I have performed a quasi 3D simulation, being the z-axis the homogeneous direction, with an homogeneous length equal to 0.1 (LZ=0.1) and up to 3 Fourier modes (HomModesZ = 4). In that sense, my planes are located at: 0, 0.025, 0.05 and 0.075. I want to interpolate the solution to a series of points, thus I am using the interppoints module. These points are located along the z direction and between the Fourier planes, e.g., z=0.033, 0.066 etc. The output file from the FieldConvert -m interppoints command is quite strange, since each variable is constant along the z direction, and of course, it is not the actual case. I have also tested it for another set of points. In this second test I have changed the z-coordinates and fixed it to the Fourier planes. However, the result is the same as before, constant measure for each variable and with exactly the same value as for the first set of points. So, is the interppoints module implemented for the 3DH1D simulations? I would like to mention that on the user guide, the commands shown for the usage of the interppoints module, except for the first example, they are missing the output "file2.dat". Thank you. Regards, Guillermo

2 1

Error - tutorial NACA
by Felipe Silva Maffei 12 Sep '19

12 Sep '19

Hi, I am master student in universidade de São Paulo. As part of a discipline, I need to desenvolve a project using nektar++. I am following the tutorial of NACA, but when I try to generate the .xml file by tapping $NEK/NekMesh, I get the following message: No such module: input: mcf. What would I be doing wrong? Best regards, Felipe Silva Maffei <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campai…> Livre de vírus. www.avast.com <https://www.avast.com/sig-email?utm_medium=email&utm_source=link&utm_campai…>. <#DAB4FAD8-2DD7-40BB-A1B8-4E2AA1F9FDF2>

2 2

Unable to use Compressible flow solver for transonic/supersonic flow problems
by Gaurav Kumar 01 Sep '19

01 Sep '19

Hi! I am trying to simulate a hypersonic flow over wedge configuration. I have successfully simulated the flow problem at a subsonic Mach number (M=0.3). But increasing Mach number to a supersonic range, Let's say M=3, the solution diverges even with a decrease in time step size. I think my problem could be with the use of artificial viscosity or aliasing errors. Can somebody share the setup files for a supersonic case, maybe flow past NACA0012 airfoil test case from the user manual? And, any other suggestion or documentation relating to such a problem will be very helpful to me. Thank you Regards Gaurav Kumar Ph.D Student, Department of Aerospace Engineering, Indian Institute of Technology, Kanpur

1 0

Invitation to International Symposium on High-Fidelity Computational Methods & Applications 2019
by Xu, Hui 30 Aug '19

30 Aug '19

Dear Colleagues, We are very pleased to invite you to attend “International Symposium on High-Fidelity Computational Methods & Applications 2019 (ISHFCMA 2019)”, which will take place on 14-16 Dec. 2019 in Shanghai. Honorary Chairs: Jialing Le (Member of CAE), Wenquan Tao (Member of CAS) Chairs: Shiyi Chen (Member of CAS), Spencer J Sherwin (Fellow of the Royal Academy of Engineering) The symposium (International Symposium on High-Fidelity Computational Methods & Applications 2019, ISHFCMA 2019), which will be held in Shanghai, is to enhance deployment and applications of the high-fidelity methods in complex industrial fluid flows. Due to the increasingly stringent precision requirements for industrial designs, these methods have recently attracted great attention. The objective of this symposium is to bring together experts in computational science and experts in engineering application to exchange new ideas and discuss development perspectives of high-fidelity methods. It also offers a platform to show the exciting scientific and engineering study undertaken in this area. The symposium will also end with a Nektar++ workshop to provide a more opportunity to understand how to apply high order spectral/hp element software. The selected papers are recommended to be published in Advances in Aerodynamics, Journal of Hydrodynamics or Acta Aerodynamica Sinica. You can find the exact dates and further information at the webpage https://www.ishfcma.org Your attendance will be greatly appreciated, and your presence will do us great honour. Please let us know as soon as possible but no later than Sept 30 (extended) if you will be able to come to Shanghai. If you need further help for your trip, please feel free to contact us. Sincerely, Hui Xu, Secretary of ISHFCMA 2019 On behalf of ISHFCMA 2019 Committee Email: ishfcma(a)hotmail.com<mailto:ishfcma@hotmail.com> ---------------------------------------------------------------------------- School of Aeronautics and Astronautics Shanghai Jiao Tong University, Shanghai, China

1 0

NekMesh compilation error
by Oliviu Sugar 15 Aug '19

15 Aug '19

Dear all, I get the following error when compiling://usr/lib/x86_64-linux-gnu/libpng16.so.16: undefined reference to `inflateValidate(a)ZLIB_1.2.9' collect2: error: ld returned 1 exit status utilities/NekMesh/CMakeFiles/NekMesh.dir/build.make:1083: recipe for target 'utilities/NekMesh/NekMesh' failed make[2]: *** [utilities/NekMesh/NekMesh] Error 1 CMakeFiles/Makefile2:618: recipe for target 'utilities/NekMesh/CMakeFiles/NekMesh.dir/all' failed make[1]: *** [utilities/NekMesh/CMakeFiles/NekMesh.dir/all] Error 2 Can anyone point me in the right direction about what could be causing this issue? Many thanks, Oliviu

2 6

Mesh Generation
by 蒋树杰 12 Aug '19

12 Aug '19

Hi All, May I ask which software can generate high order quadrilateral mesh for Nektar++ easily? Now I want to use Quasi-3D approach. It seems that only quadrilateral mesh can be used in Quasi-3D approach. I tried to use Gmsh to generate mesh for NACA0002 airfoil. I have an example mesh. But I do not know which software can generate mesh like that. I used to use Gmsh for mesh generation. But for NACA0002 airfoil, it seems it is difficult to generate a good quality quadrilateral mesh. If someone has good ideas, please let me know! Many thanks! Kind regards, Shujie

2 1