Question about a Steady Case
Dear Nektar++ Developers, We intend to do a blood flow simulation for a *steady* case by using the “velocity correction scheme”. In order to achieve the *steady* state, we have questions as follows: 1) What quantity(s) should be monitored? 2) Where should this quantity be tracked? 3) What criteria with which tolerance should be considered? Moreover, if we have a high order mesh produced by Nektar++ consisting of different types of elements with Pth order polynomials, what will be the equivalent number of linear elements of that mesh in 2D and 3D? We would be so pleased if you could help us about above-mentioned questions. Kind Regards, Mohammadali Alidoost
Hi, 1. What I do is track the energy using the Energy filter: <FILTERS> <FILTER TYPE="ModalEnergy"> <PARAM NAME="OutputFile">EnergyFile</PARAM> <PARAM NAME="OutputFrequency">100</PARAM> </FILTER> </FILTERS> This will produce a file EnergyFile.mdl with the time history of the kinetic energy. If the value does not change it could be indicative of a steady state. What should be the tolerance depends on the problem. Alternatively you could be comparing flow field snapshots (.chk) subtracting one from the other using FieldConvert utility. Than calculate an integral with e.g. the ParaView. What would be the tolerance, I do not know. Also it is possible to introduce points at which values are tracked (see the manual), and track their time evolution. Regarding the linear mesh equivalence. Are you asking about the mesh resulting in the same number of DOFs? So just splitting the HO mesh? The FieldConvert utility visualization does that - you can experiment with NUMMODES (<E COMPOSITE="C[10]" NUMMODES="4" TYPE="MODIFIED" FIELDS="u"/>), visualize the mesh to see the number of elements. If you ask what is the number of linear elements in the mesh that will allow for a solution of similar quality. It is not so easy. Might be the developers have some illustrative examples. Could be some of the publications listed on the Nektar web page? Best, Stan Gepner On 14.09.2016 09:59, mohammadali alidoost wrote:
Dear Nektar++ Developers,
We intend to do a blood flow simulation for a *steady* case by using the “velocity correction scheme”. In order to achieve the *steady* state, we have questions as follows:
1) What quantity(s) should be monitored?
2) Where should this quantity be tracked?
3) What criteria with which tolerance should be considered?
Moreover, if we have a high order mesh produced by Nektar++ consisting of different types of elements with Pth order polynomials, what will be the equivalent number of linear elements of that mesh in 2D and 3D?
We would be so pleased if you could help us about above-mentioned questions.
Kind Regards,
Mohammadali Alidoost
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Dear Nektar++ Developers, We really appreciate Dr. Gepner for his responses to our questions. However, we would be so pleased if the other Nektar++ developers could provide different suggestions about the questions related to the “Steady State” and “Mesh Issues”. Best Regards, Mohammadali Alidoost On Thu, Sep 15, 2016 at 1:51 PM, Stanisław Gepner <sgepner@meil.pw.edu.pl> wrote:
Hi,
1. What I do is track the energy using the Energy filter:
<FILTERS> <FILTER TYPE="ModalEnergy"> <PARAM NAME="OutputFile">EnergyFile</PARAM> <PARAM NAME="OutputFrequency">100</PARAM> </FILTER> </FILTERS>
This will produce a file EnergyFile.mdl with the time history of the kinetic energy. If the value does not change it could be indicative of a steady state. What should be the tolerance depends on the problem.
Alternatively you could be comparing flow field snapshots (.chk) subtracting one from the other using FieldConvert utility. Than calculate an integral with e.g. the ParaView. What would be the tolerance, I do not know.
Also it is possible to introduce points at which values are tracked (see the manual), and track their time evolution.
Regarding the linear mesh equivalence. Are you asking about the mesh resulting in the same number of DOFs? So just splitting the HO mesh? The FieldConvert utility visualization does that - you can experiment with NUMMODES (<E COMPOSITE="C[10]" NUMMODES="4" TYPE="MODIFIED" FIELDS="u"/>), visualize the mesh to see the number of elements.
If you ask what is the number of linear elements in the mesh that will allow for a solution of similar quality. It is not so easy. Might be the developers have some illustrative examples. Could be some of the publications listed on the Nektar web page?
Best,
Stan Gepner On 14.09.2016 09:59, mohammadali alidoost wrote:
Dear Nektar++ Developers,
We intend to do a blood flow simulation for a *steady* case by using the “velocity correction scheme”. In order to achieve the *steady* state, we have questions as follows:
1) What quantity(s) should be monitored?
2) Where should this quantity be tracked?
3) What criteria with which tolerance should be considered?
Moreover, if we have a high order mesh produced by Nektar++ consisting of different types of elements with Pth order polynomials, what will be the equivalent number of linear elements of that mesh in 2D and 3D?
We would be so pleased if you could help us about above-mentioned questions.
Kind Regards,
Mohammadali Alidoost
_______________________________________________ Nektar-users mailing listNektar-users@imperial.ac.ukhttps://mailman.ic.ac.uk/mailman/listinfo/nektar-users
_______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk https://mailman.ic.ac.uk/mailman/listinfo/nektar-users
Hi Mohammadali, I agree that Stan’s response was good and covered the key issues. With regard to steady state monitoring another option is to put in some history points or indeed force calculations and then you and monitor when they become steady. Detail of both should be in the user-guide. Classifying equivalent mesh is more complex. As Stan has suggested you can get the local and global degrees of freedom from the post-processing however what you must consider is when an error of h^2 is equivalent to an error of h^4. Depending on the value of h you choose dictates how many linear elements would be required to get the same accuracy. Cheers, Spencer. PS We did do some work on Selective Frequency Damping (see papers by Bastien Jordi) however we have not tested this option very extensively, particularly in 3D. On 21 Sep 2016, at 16:40, mohammadali alidoost <alidoost.ma@gmail.com<mailto:alidoost.ma@gmail.com>> wrote: Dear Nektar++ Developers, We really appreciate Dr. Gepner for his responses to our questions. However, we would be so pleased if the other Nektar++ developers could provide different suggestions about the questions related to the “Steady State” and “Mesh Issues”. Best Regards, Mohammadali Alidoost On Thu, Sep 15, 2016 at 1:51 PM, Stanisław Gepner <sgepner@meil.pw.edu.pl<mailto:sgepner@meil.pw.edu.pl>> wrote: Hi, 1. What I do is track the energy using the Energy filter: <FILTERS> <FILTER TYPE="ModalEnergy"> <PARAM NAME="OutputFile">EnergyFile</PARAM> <PARAM NAME="OutputFrequency">100</PARAM> </FILTER> </FILTERS> This will produce a file EnergyFile.mdl with the time history of the kinetic energy. If the value does not change it could be indicative of a steady state. What should be the tolerance depends on the problem. Alternatively you could be comparing flow field snapshots (.chk) subtracting one from the other using FieldConvert utility. Than calculate an integral with e.g. the ParaView. What would be the tolerance, I do not know. Also it is possible to introduce points at which values are tracked (see the manual), and track their time evolution. Regarding the linear mesh equivalence. Are you asking about the mesh resulting in the same number of DOFs? So just splitting the HO mesh? The FieldConvert utility visualization does that - you can experiment with NUMMODES (<E COMPOSITE="C[10]" NUMMODES="4" TYPE="MODIFIED" FIELDS="u"/>), visualize the mesh to see the number of elements. If you ask what is the number of linear elements in the mesh that will allow for a solution of similar quality. It is not so easy. Might be the developers have some illustrative examples. Could be some of the publications listed on the Nektar web page? Best, Stan Gepner On 14.09.2016 09:59, mohammadali alidoost wrote: Dear Nektar++ Developers, We intend to do a blood flow simulation for a steady case by using the “velocity correction scheme”. In order to achieve the steady state, we have questions as follows: 1) What quantity(s) should be monitored? 2) Where should this quantity be tracked? 3) What criteria with which tolerance should be considered? Moreover, if we have a high order mesh produced by Nektar++ consisting of different types of elements with Pth order polynomials, what will be the equivalent number of linear elements of that mesh in 2D and 3D? We would be so pleased if you could help us about above-mentioned questions. Kind Regards, Mohammadali Alidoost _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users _______________________________________________ Nektar-users mailing list Nektar-users@imperial.ac.uk<mailto:Nektar-users@imperial.ac.uk> https://mailman.ic.ac.uk/mailman/listinfo/nektar-users Spencer Sherwin McLaren Racing/Royal Academy of Engineering Research Chair, Professor of Computational Fluid Mechanics, Department of Aeronautics, Imperial College London South Kensington Campus London SW7 2AZ s.sherwin@imperial.ac.uk<mailto:s.sherwin@imperial.ac.uk> +44 (0) 20 759 45052
participants (3)
-
mohammadali alidoost
-
Sherwin, Spencer J
-
Stanisław Gepner