May 2020 - Nektar-users - imperial.ac.uk

Outflow boundary conditions with an AOA
by 岳泽华 08 Jul '20

08 Jul '20

******************* 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, I'm working on a stability and dynamics analysis of a high Re folw past a slender blunt-based axisymmetric body by the IncNavierStokesSolver, then I need to set an angle of attack for the flow.So I set the inflow surface boundary condition by defining its velocity component, and it works well.My question is about the corresponding outflow boundary conditions, if i remain it as the 0-AOA conditions, the streamlines near the outlet surface will become wrong curves. And I tried several definitions of the outflow surface but it always failed.So I'm wondering how to write the outflow boundary conditions with an AOA, I'm really confused。Thanks for your kind attention and help. Many thanks, Zehua Yue Beihang University

2 1

loading data of velocity components from a file(*.fld) or expression into user-defined arrays in Nektar++
by 王昭予 01 Jun '20

01 Jun '20

******************* 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 want to load data of velocity components(u,v,w) from a file(*.fld) or expression into a user-defined array just like that in boundary conditions: <D VAR="u" USERDEFINEDTYPE="TimeDependent" VALUE="sin(PI*(x-t))" /> <REGION REF="1"> <D VAR="u" USERDEFINEDTYPE="TimeDependent" VALUE="sin(PI*(x-t))" FILE="bcsfromfiles_u_1.bc" /> </REGION> I wonder how it works. My purpose is to load the data into some arrays defined by myself like U[i], V[i], W[i] with the mesh data. Then I can use the data on each mesh point for further computation. Could you please tell me how to realize it with codes in Nektar++? I have check the codes in Explist.cpp, SessionReader.cpp, Conditions.cpp and IncNavierStokes.cpp but still have no idea. Many thanks, Zhaoyu Wang Shanghai Jiao Tong University

3 2

Question about Nektar++ fiter: Interpolation from field to field for 3DH1D case
by Xiao, Chengnian 28 May '20

28 May '20

******************* 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++ community, I have been using version 5.0.0 of nektar++ to run unsteady incompressible flow simulations of the 3DH1D type, i.e. 2d spectral elements combined with Fourier expansion along one homogeneous direction. The problem I am encountering is interpolation of the resulting field data. I used the nektar++ example file for incompressible Navier Stokes solver ChanFlow_3DH1D_Flowrate_MVM.xml and ran the field interpolation as follows: FieldConvert -m interpfield:fromxml=ChanFlow_3DH1D_Flowrate_MVM.xml:fromfld=ChanFlow_3DH1D_Flowrate_MVM_1.chk \ rect2d.xml interp.chk where rect2d is the file containing the mesh information (Geometry and Expansion) only, the same information copy-pasted from the session .xml file used as input.( When I used the entire session file, same as input file, the module would immediately crash due to segmentation fault.) I received the following error message after the execution of the above: Fatal : Level 0 assertion violation Failed to type cast m_exp[0] terminate called after throwing an instance of 'Nektar::ErrorUtil::NekError' what(): Level 0 assertion violation Failed to type cast m_exp[0] Aborted (core dumped) I have checked the resulting field data specified in ChanFlow_3DH1D_Flowrate_MVM_1.chk , after converting it to .vtu, and it looks to be without obvious faults. I have also had no problems so far interpolating from 2D or 3D cases to the same geometry. Could this be a specific issue about interpolation for the 3DH1D case, or was there an error in the session file or my command execution? I would really appreciate if you could elucidate this matter for me. Best regards, Cheng-Nian Xiao

2 1

Aeroacoustics using Cwipi
by Nitish Arya 28 May '20

28 May '20

******************* 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 am working on an aeroacoustic problem where I am sending acoustic sources (lamb vector) from Openfoam to nektar. I have already validated the acoustic results from the DNS study. Now I want to see the effects of acoustics on the flow. I have two doubts regarding it - 1. When I am trying to the send the perturbation pressure (variable p) from nektar to openfoam, the quantity I am receiving is not the pressure perturbation but what seems like a crude representation of the source term I already sent from openfoam. (Also the SendMethod "Evaluate" does not work). When I tried sending velocity perturbation instead, I still received the same thing. I have attached the session file. 2. The perturbation pressure for APE-4 system is composed of hydrodynamic as well as acoustic parts. In order to study the effect of acoustics on the flow, probably I have to remove the hydrodynamic part. Is it possible to separate (to some extent) the hydrodynamic component from the acoustic component using the linear differential filter already employed in the code. Sincerely, Nitish Arya

1 0

CWIPI: passing just a region in the mesh
by Hoang-Tuan TRUONG 25 May '20

25 May '20

******************* 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 am just reading the manual and wonder if it is possible for CWIPI to just pass variable at a certain region not the whole computational domain. Let's say I have an ADR solver for heat conduction and want to pass the temperature field at the the solid-fluid interface to a Navier-Stoke solver + an energy transport solver. Is it possible with CWIPI? Regards, Tuan H Truong.

2 1

use of wss module in FieldConvert in 2.5D simulations
by 王昭予 21 May '20

21 May '20

******************* 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, I wonder if the wss module (Computing the wall shear stress) in FieldConvert can be use in 2.5 simulations. The command line written in user-guide is FieldConvert -m wss:bnd=0:addnormals=1 test.xml test.fld test-wss.fld. However, in a 2.5 simulation like BL, the wall boundary is usually a line not a surface. When I want to compute the wall shear stress along the center line of a wall surface, something wrong may happen. So what can I do to solve the problem in a 2.5D simulation. Thank you for your kind attention and help. Many thanks, Zhaoyu Wang Shanghai Jiao Tong University

2 1

Mesh/Nektar++
by Silvia Ceccacci (HDR) 20 May '20

20 May '20

******************* 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, Does Nektar++ have a threshold regarding the number of elements in the mesh (in particular in the 3D case)? Is there a maximum value below which I should work? Thank you. Kind regards, Silvia

2 1

Quasi-3D simulation of NACA0015
by 王昭予 18 May '20

18 May '20

******************* 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, I have a Quasi-3D simulation of NACA0015 at Rec=40000. My purpose is to verify the aerodynamic forces(lift coefficient, drag coefficient,pressure coefficient). The simulation is run at P=3 and the chord length is 0.12m(AOA=4 degree). In the spectral/hp planes(x-y plane), I generate Quadrilateral elements with ICEM CFD. 20 convective time (T=20C/U) is used to cover the transient phase.Time-averages are subsequently sampled for further 20 convective time. However, something may be wrong with the setting of parameters. The variation of aerodynamic forces in DragLift.fce(attached) become stagnant after few time steps. I also think the flow is not 'started' correctly, I cant see any vortex structures during post-processing, even the convective time is sufficient(attached,NACA0015_4degree_2.5D_0515_9.chk.tar.gz). I put the mesh file (NACA0015_4degree_2.5D_0515.xml, some mesh pictures in mesh_pictures.tar.gz) and setting file(session.xml) in my e-mail. Could you please tell me where the trouble is, the mesh or the setting file? I wish all of you are fine and healthy. Thank you for your help. Kind regards, Zhaoyu Wang Shanghai Jiao Tong University

3 2

IncNavierStokesSolver
by A.P. 13 May '20

13 May '20

******************* 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 need to calculate a steady flow which may be unstable. Is the option <I PROPERTY="EQTYPE" VALUE="SteadyLinearisedNS"/> can be used only for Re<<1 or it is possble to make an appropriate AdvectionVelocity section for large Re? Alex Proskurin.

4 8

terminate called after throwing an instance of 'std::bad_alloc'
by Silvia Ceccacci (HDR) 13 May '20

13 May '20

******************* 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 working on a server which has 48 threads and 376GB of RAM. I am running the IncNavierStokesSolver with an order 2 mesh that has 15353 nodes. This works fine. If I increase the number of nodes (say 179892, but also 34446), without changing the parameters, the simulation crashes with a std::bad_alloc error (see error.txt attached). Do you think I need a larger RAM? Or could it be something else? Thank you. Kind regards, Silvia

1 0