Note: This discussion is about an older version of the COMSOL Multiphysics® software. The information provided may be out of date.

Discussion Closed This discussion was created more than 6 months ago and has been closed. To start a new discussion with a link back to this one, click here.

Simulation results dont match analytical ones

Please login with a confirmed email address before reporting spam

hi,

I m trying to simulate a surface meshed by shell elements with constraint at one end and a vertical force in the opposite end.

I m running the simulation and everything seems fine except that the result of the surface decline defer from the analytical result by 15 percent. Despite the fact that the forces and material moduli account for small elasticity.

When i m trying to do such a comparison for a thin and long beam the anayltical calculation of the decline and the simulation result are perfectly the same.

What could be the problem?

thx a lot

8 Replies Last Post 2011年12月14日 GMT-5 01:30
Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月12日 GMT-5 03:30
Hi

have you checked that your results are mesh independent ? and that you respect the shell hypothesis (at least a thin element, check the doc for the rest)

Mostly I'm happy with 5-10% coherence for FEM, particularly when using "simplified" models

--
Good luck
Ivar
Hi have you checked that your results are mesh independent ? and that you respect the shell hypothesis (at least a thin element, check the doc for the rest) Mostly I'm happy with 5-10% coherence for FEM, particularly when using "simplified" models -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月12日 GMT-5 04:33
Thx a lot for the fast reply

My model is basicly a surface with length a width equal to 1 cm. The Shell's thickness has been defined as 0.3 mm (and i assume that this thickness is thin enough).

I m only trying to deal with Small Elasticity so the Z deformation resulted in approximatly 3*10^-6 [m] at the edge where the vertical Z force has been defined (the force per length is 3.6*10^-6 N). The analytical decline calculation produced 3.555*10^-6 [m] ( Z deformation ).

As far as i see, the results are not really mesh independent, because refining the mesh is changing the results by almost 10 percent.

Another strange thing is that although the force (per length) has been defined for the entire edge, i get in the results parabolic Z deformation when looking down the X direction of the surface, instead of a uniform Z deformation per X (The edge with the force has a normal in the X direction).

I hope this added information would help.
thx a lot
Thx a lot for the fast reply My model is basicly a surface with length a width equal to 1 cm. The Shell's thickness has been defined as 0.3 mm (and i assume that this thickness is thin enough). I m only trying to deal with Small Elasticity so the Z deformation resulted in approximatly 3*10^-6 [m] at the edge where the vertical Z force has been defined (the force per length is 3.6*10^-6 N). The analytical decline calculation produced 3.555*10^-6 [m] ( Z deformation ). As far as i see, the results are not really mesh independent, because refining the mesh is changing the results by almost 10 percent. Another strange thing is that although the force (per length) has been defined for the entire edge, i get in the results parabolic Z deformation when looking down the X direction of the surface, instead of a uniform Z deformation per X (The edge with the force has a normal in the X direction). I hope this added information would help. thx a lot

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月12日 GMT-5 05:44
Hi

why should you get a uniform linear deformation ?, that is what I expect from a force load, you will get an arc of a circle if you load with a pure moment. But I'm not 100% sure I got you coordinates & load right.

In anycase I find your simulation and analytical results rather close 45/3600 is about 1.25% given all assumptions on the analytical and FEM I'm not surprised

--
Good luck
Ivar
Hi why should you get a uniform linear deformation ?, that is what I expect from a force load, you will get an arc of a circle if you load with a pure moment. But I'm not 100% sure I got you coordinates & load right. In anycase I find your simulation and analytical results rather close 45/3600 is about 1.25% given all assumptions on the analytical and FEM I'm not surprised -- Good luck Ivar

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月13日 GMT-5 00:00
Hi,

Unfortunately the comparison should be between the analytical Z deformation, which resulted in 3.55*10^-6 [m] to the result of the simulation which resulted with 3*10^-6 [m]. So the error is about 15 percent.

The model is a surface with width and length both equal to 1 cm. The surface is meshed with Shell elements with thickness of 0.3 mm. There is a constraint at one edge and force load at the opposite edge equals to 3.6*10^-6 [N/m].
The 2 other edges are free.

What could be the problem?

Thank you
Hi, Unfortunately the comparison should be between the analytical Z deformation, which resulted in 3.55*10^-6 [m] to the result of the simulation which resulted with 3*10^-6 [m]. So the error is about 15 percent. The model is a surface with width and length both equal to 1 cm. The surface is meshed with Shell elements with thickness of 0.3 mm. There is a constraint at one edge and force load at the opposite edge equals to 3.6*10^-6 [N/m]. The 2 other edges are free. What could be the problem? Thank you

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月13日 GMT-5 00:02
Of course the force load is directed in the Z direction while the surface is at the XY plane.
Of course the force load is directed in the Z direction while the surface is at the XY plane.

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月13日 GMT-5 02:55
Hi
you have forgotten to tell us the material used

I get 2.5e-9 mm for "structural steel" in 3D shell 1cm^2 and 0.3 mm thick, with 5854 elements

and 2.37e-9 mm for a 2D "solid" model with 802 elements

all in linear theory for me rather coherent

--
Good luck
Ivar
Hi you have forgotten to tell us the material used I get 2.5e-9 mm for "structural steel" in 3D shell 1cm^2 and 0.3 mm thick, with 5854 elements and 2.37e-9 mm for a 2D "solid" model with 802 elements all in linear theory for me rather coherent -- Good luck Ivar


Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月13日 GMT-5 14:06
Of course,

Module Young, E, is 150 kPa and poisson ratio is 0.495.
My simulation result was 3.0382*10^-6 [m] for the maximum Z deformation, while the analytical result is 3.555*10^-6 [m].

Thank you
Of course, Module Young, E, is 150 kPa and poisson ratio is 0.495. My simulation result was 3.0382*10^-6 [m] for the maximum Z deformation, while the analytical result is 3.555*10^-6 [m]. Thank you

Ivar KJELBERG COMSOL Multiphysics(r) fan, retired, former "Senior Expert" at CSEM SA (CH)

Please login with a confirmed email address before reporting spam

Posted: 1 decade ago 2011年12月14日 GMT-5 01:30
Hi

I do not get the same values, but I suspect that my model is not correct, and that I must have missed something from your description.
Another point not to forget, are you using a max or an average displacement ? as depending on how you define your model these two might differ. For precision measurements, I always take an average over a boundary as my driving value, it's too easy to get a "hot spot" and conclude from a peak

--
Good luck
Ivar
Hi I do not get the same values, but I suspect that my model is not correct, and that I must have missed something from your description. Another point not to forget, are you using a max or an average displacement ? as depending on how you define your model these two might differ. For precision measurements, I always take an average over a boundary as my driving value, it's too easy to get a "hot spot" and conclude from a peak -- Good luck Ivar

Note that while COMSOL employees may participate in the discussion forum, COMSOL® software users who are on-subscription should submit their questions via the Support Center for a more comprehensive response from the Technical Support team.