Velocity distribution in input from pipe

Numerical methods and mathematical models of Elmer
Post Reply
Luve
Posts: 9
Joined: 22 May 2022, 16:07
Antispam: Yes

Velocity distribution in input from pipe

Post by Luve »

Hello

I tried the tutorial 23 "Thermal flow in curved pipe" and made a mistake.
By test I have exchanged input and output of the pipe.
Because of this observation I have questions:

1. is it correct that the boundary condition:

Navier-Stokes
Velocity 1 = 0.0
Velocity 2 = 0.0
Velocity 3 = Variable Coordinate
Real MATC "100.0*(1.0e-4-tx(0)^2-tx(1)^2)"

is only valid for absolute coordinates tx(0) and tx(1)? This could explain
why ELMER aborts with an error message if the input boundary of the pipe is wrong.


2. could you publish the coordinates (numerical values) which result in
Real MATC "100.0*(1.0e-4-tx(0)^2-tx(1)^2)"?
As a user I can in the mesh: "curved_pipe.grd" not read these values. For me this
information is important to model my own examples correctly.

A suggestion: Mark in the tutorial 23 with a graphic input
of the pipe. This helps the newbies.

Thanks

With kind regards
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Velocity distribution in input from pipe

Post by kevinarden »

You can use ElmerGrid convet a grd file to gmsh or vtu file (paraview) then you can interrogate the coordinates. Elmer accepts meshes from many mesh creating software, it does not have to be a grd file. If you convert it you can see that the input is in the xy plane at z = 0. the boundary condition
x=0, y=0 means flow perpendicular to pipe inlet is zero. flow down the pipe in the Z direction is a function of the x, y coordinates resulting in a parabolic flow
Velocity 3 = Variable Coordinate means the MATC equation will use the coordinates x, y, z where tx(o)=x, tx(1) =y, and tx(3)=z

When modeling your own problems you would have control and knowledge of the coordinates.
curvedpipe.png
(244.33 KiB) Not downloaded yet
Luve
Posts: 9
Joined: 22 May 2022, 16:07
Antispam: Yes

Re: Velocity distribution in input from pipe

Post by Luve »

Thanks for the information.

With Paraview I had tried it. When ELMER stops the calculation
no case_t0001.vtu file is created. Therefore Paraview cannot be used.
be used.

A simple suggestion how the graphic in the documentation could be changed
could be. Proverb: "There are many ways to Rome!"

with kind regards
Attachments
Input_Orientation.jpg
(227.32 KiB) Not downloaded yet
Luve
Posts: 9
Joined: 22 May 2022, 16:07
Antispam: Yes

Re: Velocity distribution in input from pipe

Post by Luve »

Addendum:

I have now extracted the absolute coordinates with Paraview and saved them as a *.csv file.
A 3D surface is created by the software Rstudio. (see attachment)

As Z-coordinate I used the function from the tutorial

z <- 100.0*(1.0e-4-x_coord^2-y_coord^2)


Important: x_coord and y_coord are the absolute coordinates. That means if
exactly the same flow profile occurs at another position in your example than
the function

Real MATC "100.0*(1.0e-4-tx(0)^2-tx(1)^2)"

must be recalculated and rewritten!

Luve
Attachments
Velo_Distribut.png
Velo_Distribut.png (51.35 KiB) Viewed 366 times
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: Velocity distribution in input from pipe

Post by kevinarden »

From a command line I executed
ElmerGrid 1 5 curve_pipe
to convert the grd file to a vtu file
Post Reply