How to import complex geometry like an airfoil in .i3d file ?

[gauravity30]

I have some airfoils over which i want to run the simulations. How do I import the airfoil geometry for simulation ? One of the way I found was to replace the equation of cylinder with the equation of the geometry in the BC_Cylinder.f90 file. Is there any other way to import the geometry like using the coordinates of the airfoil.

[slaizet]

Hi,
We are going to prepare a video tutorial on how to do import a stl file in Xcompact3d (it is possible via the Sandbox). In the meantime, using equations for your airfoil is the best option :-)
Best,
Sylvain

[gauravity30]

Thank you ! It will be very helpful.

[gauravity30]

Hello @slaizet !
I tried using the sandbox mode through the python code xcompact3d-toolbox. I was facing trouble installing the library. The dependencies of the library were not supporting each other. So, I could not use the option to import STL using that.

So, with some reading i developed my own python code to produce Epsilon function for any STL geometry. Currently, I am trying to input the geometry in cylinder case, by reading the indices of the points inside geometry and setting epsi(x,y,z) = remp for those coordinates.
(Check out the python code I used, here)

if (nxi.eq.1.and.nxf.eq.nx.and.&
         nyi.eq.xstart(2).and.nyf.eq.xend(2).and.&
        nzi.eq.xstart(3).and.nzf.eq.xend(3)) then
         open(2, file = 'src/epsilon.dat', status = 'unknown')
         read(2,*) len
         allocate(ex(len),ey(len),ez(len))
         read(2,*) (ex(i), ey(i), ez(i), i=1, len)
         close(2)
         do i=1,len
            epsi(ex(i),ey(i),ez(i)) = remp
         enddo
      endif

This is the modification I made in BC-Cylinder.f90. The simulation starts but all the values are NaN. Can you please me how to input the geometry correctly from a epsilon file ? Do we need to write an extra subroutines for this ?

[slaizet]

Dear Gaurav,

You should not modify BC-Cylinder.f90. The file of interest for you is genepsi3d.f90. This is where we prepare all the files and arrays for the immersed boundary method approach, in particular the subroutine geomcomplex_io.

Best,

Sylvain

[gauravity30]

Dear Sylvain,
Thank you for the reply I actually looked in the genepsi3d.f90 and subroutine geomcomplex_io. I was a little confused. If I opt for reading the epsilon file using the geomcomplex_io subroutine, then I have to read the files for the other parameters nobjx,nobjy,nobjz,xi,xf,yi,yf,zi,zf, nxipif,nxfpif,nyipif,nyfpif,nzipif,nzfpif,nobjmax,npif. Can you tell what these parameters are ? I couldn't understand what these arrays actually mean.

Interestingly, the code works with my modification if I set the IBM method as forced boundary method (1) instead of method mentioned in "A DNS study of jet control with microjets using an alternating direct on forcing strategy". I am testing whether I am getting good results with the forced boundary method. Also, I need to check if it works with the other method.

Regards,
Gaurav

[gauravity30]

Hello @slaizet,
I was going through the genepsi3d.f90 file. I am little confused in the part where nobjy and nobjz are computed. What is the call transpose_x_to_y(ep1,ep2) and transpose_y_to_z(ep2,ep3). I tried checking what it was doing so I tried printing the size of the arrays ep1, ep2, ep3. The shape of the array is constant and not changing, so what the calls are doing ? I also noticed these calls in verif_epsi subroutine while generating nyipif, nyfpif, ... I apologize for asking so many questions. I am trying to write the code in python, so I am trying to understand the inner working. Thanks a lot !

Thanks,
Gaurav

[slaizet]

Hi,
As you should know, all 3D arrays ending with 1, 2 or 3 are 3D arrays in X-, Y- or Z- pencils. They do not have the same sizes. The transpose subroutines are tools to switch an array from one pencil decomposition to another. Please have a read at the relevant publications for more information.
Best,
Sylvain

[gauravity30]

Hii,
Thanks for the explanation. So, the transpose subroutines are relevant to the domain decomposition process. Since, I will be generating the nobjx, nobjy, .. etc. in python without any domain decomposition. I think I wouldn't require to implement the transpose subroutines. Does, the geomcomplex_io subroutine handle this domain decomposition after it reads the data from the .dat files ?

Thanks,
Gaurav

[slaizet]

Hi,
I am not quite sure what you are trying to do here (doing in Python what is currently being done in Fortran?). If you are trying to generate in Python all the arrays needed for the reconstruction for the IBM then yes you do not need to worry about the transpose subroutines.
Best,
Sylvain

[gauravity30]

Hi,
Thanks a lot for the clarification. Yes, I am trying to generate the arrays in Python for reconstruction of IBM.

Thanks,
Gaurav