Carving Out a Cylinder using OBST and MULT

[kasrash]

Hi everyone,

I am trying to create a cylindrical obstruction using OBST and MULT following FDS User's Guide. However, the cylinder is not created correctly or as expected. I would appreciate the help to know what I am doing wrong. Here is how I am defining the cylinder:

&MULT ID='cylinder', DX=0.00125,DY=0.00125,DZ=0.00125,I_UPPER=31,J_UPPER=31,K_UPPER=31 / &OBST XB=-0.01905,-0.0178,-0.003175,-0.001925,0.00125,0.0025, MULT_ID='cylinder', SURF_ID='vegetation', SHAPE='CYLINDER', RADIUS=0.003175, HEIGHT=0.0381, ORIENTATION=1,0,0, XYZ=-0.01905,0,0.004425/

The cylinder should be parallel to X axis, its bottom should be placed 0.00125 above ground (Z=0.00125).

Attached is what I am getting with the provided definition, which does not look right (see the two flange-like shape and overall cuboid shape)

Here is the input files as well. The main input is "InputFile_MeshAdjust32.txt", the mesh detail is available in "MeshAdjust128.txt", and the cylinder material is in "vegetation_model.txt". Sorry for the messy input. The model is still in development/calibration phase.
InputFile_MeshAdjust32.txt
MeshAdjust128.txt
vegetation_model.txt

Thank you for your help in advance!

[rmcdermo]

Please post the whole input file. I cannot tell what your grid resolution is.

[kasrash]

Please see attached for the inputs. The main input is "InputFile_MeshAdjust32.txt", the mesh detail is available in "MeshAdjust128.txt", and the cylinder material is in "vegetation_model.txt". Sorry for the messy input. The model is still in development/calibration phase.

InputFile_MeshAdjust32.txt
MeshAdjust128.txt
vegetation_model.txt
is 0.00125 m (1.25 mm) in all direction. The mesh detail is in

[rmcdermo]

I admit that this is somewhat difficult to grasp at first---it took me a bit of time to remind myself how it works. Which argues for an update of the user guide, and I will work on that.

In the meantime, the way to think about this process is to imagine you are a sculptor. The MULT and OBST lines---without any SHAPE arguments---should generate a block of OBST that you will sculpt into the SHAPE you want. Below I have provided an example, not the most efficient, but it is meant for clarity first. Imagine you want to put an OBST in every cell of the domain, you could do something like this:

&HEAD CHID='test' /

&TIME T_END=0 /

&MESH IJK=76,13,13, XB=-0.0381,0.0381,-0.00635,0.00635,0.0,0.0127 /

&MULT ID='cylinder', DX=0.001,DY=0.001,DZ=0.001, I_UPPER=76,J_UPPER=13,K_UPPER=13 /
&OBST XB=-0.0381,-0.0371,-0.00635,-0.00535,0.0,0.001,
      MULT_ID='cylinder',
      SURF_ID='vegetation',
      ! SHAPE='CYLINDER',
      ! RADIUS=0.003175,
      ! HEIGHT=0.0381,
      ! ORIENTATION=1,0,0,
      ! XYZ=-0.01905,0,0.004425
      /

&SURF ID='vegetation', COLOR='BROWN'/

And you would end up with the whole domain filled, like this:

Next you generate a SHAPE that will carve out (remove) the OBSTs that you do not want. Obviously, the cell resolution , etc., matter here for what will end up getting culled in the subtraction process. In this case, the SHAPE is defined by RADIUS, HEIGHT, ORIENTATION, and XYZ. The XYZ is the center of the cylinder opposite the direction of the ORIENTATION, so the left side in this case. If the ORIENTATION were 0,0,1, the XYZ is the bottom center of the cylinder.

If you now uncomment the SHAPE parameters you get this for the selected resolution:

&HEAD CHID='test' /

&TIME T_END=0 /

&MESH IJK=76,13,13, XB=-0.0381,0.0381,-0.00635,0.00635,0.0,0.0127 /

&MULT ID='cylinder', DX=0.001,DY=0.001,DZ=0.001, I_UPPER=76,J_UPPER=13,K_UPPER=13 /
&OBST XB=-0.0381,-0.0371,-0.00635,-0.00535,0.0,0.001,
      MULT_ID='cylinder',
      SURF_ID='vegetation',
      SHAPE='CYLINDER',
      RADIUS=0.003175,
      HEIGHT=0.0381,
      ORIENTATION=1,0,0,
      XYZ=-0.01905,0,0.004425
      /

&SURF ID='vegetation', COLOR='BROWN'/

You can increase the resolution, but you must keep the resolution of the MESH and OBST consistent if you want to see improvements in the shape of the cylinder (sorry, to be general, these need to be independent, but you have control over them).

&HEAD CHID='hires' /

&TIME T_END=0 /

&MESH IJK=76,130,130, XB=-0.0381,0.0381,-0.00635,0.00635,0.0,0.0127 /

&MULT ID='cylinder', DX=0.001,DY=0.0001,DZ=0.0001, I_UPPER=76,J_UPPER=130,K_UPPER=130 /
&OBST XB=-0.0381,-0.0371,-0.00635,-0.00625,0.0,0.0001,
      MULT_ID='cylinder',
      SURF_ID='vegetation',
      SHAPE='CYLINDER',
      RADIUS=0.003175,
      HEIGHT=0.0381,
      ORIENTATION=1,0,0,
      XYZ=-0.01905,0,0.004425/

&SURF ID='vegetation', COLOR='BROWN'/

I leave it to you to work this into your case. Note: you do not need to fill your domain. But your MULT and OBST blocks without the shape parameters should generate a solid volume that more than covers the SHAPE you are trying to carve out. Play with this using the T_END=0 trick to just look at the geometry.

[kasrash]

Thank you so much for the thorough demonstrated explanation. I tried your method on a smaller domain, and then, tried to reduce the size of the base block to make it more efficient. For example, the following sample:

`&HEAD CHID='test' /

&TIME T_END=0 /

&MESH IJK=76,13,13, XB=-0.0475,0.0475,-0.008125,0.008125,0.0,0.01625 /

&MULT ID='cylinder', DX=0.00125,DY=0.00125,DZ=0.00125, I_UPPER=40,J_UPPER=8,K_UPPER=8 /
&OBST XB=-0.02,-0.01875,-0.004,-0.00275,0.0,0.00125,
MULT_ID='cylinder',
SURF_ID='vegetation',
SHAPE='CYLINDER',
RADIUS=0.003175,
HEIGHT=0.0381,
ORIENTATION=1,0,0,
XYZ=-0.01905,0,0.004425
/

&SURF ID='vegetation', COLOR='BROWN'/`

generates:

This is reasonable considering the grid size of 0.00125. Then, I tried to put this in my original domain size:

`&HEAD CHID='test' /

&TIME T_END=0 /

!&MESH IJK=76,13,13, XB=-0.0475,0.0475,-0.008125,0.008125,0.0,0.01625 /
&MESH IJK=160,80,60, XB=-0.1,0.1,-0.05,0.05,0.,0.075, /

&MULT ID='cylinder', DX=0.00125,DY=0.00125,DZ=0.00125, I_UPPER=40,J_UPPER=8,K_UPPER=8 /
&OBST XB=-0.02,-0.01875,-0.004,-0.00275,0.0,0.00125,
MULT_ID='cylinder',
SURF_ID='vegetation',
SHAPE='CYLINDER',
RADIUS=0.003175,
HEIGHT=0.0381,
ORIENTATION=1,0,0,
XYZ=-0.01905,0,0.004425
/

&SURF ID='vegetation', COLOR='BROWN'/`

which resulted in:

Now, I am confused why the shape is different than the smaller domain. Could it be a problem with smokeview visualization only and the actual OBST is correct, or something is wrong? For what is worth, I tried increasing J_UPPER and K_UPPER to domain limits, but no difference in the end result.

[rmcdermo]

Hare for me to tell without seeing the mesh. Toggle "g" to see the mesh (defaults to y orientation, hit "x" and/or "z" to see other planes). Toggle "o" to see mesh outline. Toggle "q" to see OBST as specified and what FDS snaps to the grid. These may help show what is going on.