STEP import and wrong units

The graphical user interface of Elmer
Post Reply
Newmax
Posts: 2
Joined: 10 Dec 2022, 21:38
Antispam: Yes

STEP import and wrong units

Post by Newmax »

Hi all

i'm hitting the head in the wall but i'm not finding a solution to solve this problem.
I load a STEP file in elmergui , its a steel bar 10x10x300 mm .
Coordinates in the STEP file are in meters :

----
#8=(LENGTH_UNIT()NAMED_UNIT(*)SI_UNIT($,.METRE.));
----

Checking point coordinates i can see they are written in metres in the STEP file.
I import in elmergui and Model Summary shows me :


BOUNDING BOX
X-coordinate: [ -5 , 5 ]
Y-coordinate: [ -5 , 5 ]
Z-coordinate: [ -150 , 150 ]

So at this point elmergui ( or netgen ? ) has converted yet the file in millimetres.

So i add in Setup coordinate scaling a 1e-3 value to scale down again to metres.

I assign a steel material to the part, Young's modulus : 200e9
I add a 200 Newton load at one end and 0,0,0 fix position to the other end of the bar.

Launch the solver, all ok.
I open results in paraview, i turn on axes grid and the bar is long 0.3 meters ( so it seems the coordinate scaling is working ).
But the results are wrong , max deflection on the tip of the bar is 7e-7 meters .

If i change Young's modulus to 200e5 and i launch the solver then in Paraview I see 7e-3 meters deflection ( that's quite ok ! good )

Max Von mises stress is 4e04 in both cases ( i don't know the unit at this point ) , it should be around 400 N/mm2 .

Where is the error ? in elmergui or in Paraview ?
Calcs are good and visualization is wrong ?
How i can solve ?


test_bar.zip
(330.23 KiB) Downloaded 58 times
kevinarden
Posts: 2237
Joined: 25 Jan 2019, 01:28
Antispam: Yes

Re: STEP import and wrong units

Post by kevinarden »

The correct answer for the deflection is 1.08E-6. Linear tetrahedron elements are not good for linear elastic problems. Using higher order elements, Elmer produces exactly the correct deflection of 1.08E-6.

When meters is specified as the units in step file, then the tools like gmsh etc., automatically convert to mm.
If you remove the unit description line from the step file, it will import the values as is without doing any conversions.
Newmax
Posts: 2
Joined: 10 Dec 2022, 21:38
Antispam: Yes

Re: STEP import and wrong units

Post by Newmax »

Thank you for the hint.
I tried with a mesh with second order elements and I obtain too 1.1e-6 deflection but I'm not sure it's correct.
A think the correct result is 10.8 mm so 1.08e-2 metres.
Checked again calcs using theory and Freecad with calculiX and the correct answer is 1.08e-2 and the Von Mises stress is 350 GPa.
So it seems deflection is off by 10^2 and stress is off by 10^4 in Elmer.

EDIT :

viewtopic.php?t=794

It seems i did the same mistake.
I need to multiply load by a 10000 factor cause the area of surface i applied the force to has an area of 0.01x0.01 metres !
Post Reply