Modelling composite cylinder with smeared properties in abaqus

[Megabeast]

Hi All,

I am trying to calculate natural frequency of a rotor. I have already run one analysis in pro mechanica which gives me first natural frequency to be equal to 2 Hz.
I am now running the same rotor model in abaqus for comparison and I am getting 1.5 Hz. I suspect it is something to do with the cylindrical spool. May be its stiffness is not rightly modelled.
It is a hollow cylinder with a certain thickness and is made of composite material, continuum shell elements with smeared properties have been used.
I read in the manual that in the through thickness direction, there should be only one element. Hence, i tweeked the mesh setting to have just one element in the thickness direction.
The smeared properties used here are same those used in pro-mechanica. Since there is only one ply due to smeared properties i have taken element relative thickness as one.
and stacking direction is what i am confused about, if you look at Figure 1, stacking direction i.e thickness direction could be either 2 and 3, depending on how we look at the cylinder.
I also tried using cylindrical coordinate system instead of cartesian.
Should i use solid homogeneous section instead of composite layup to get the cylinder stiffness right? How about the element stacking direction?

Thanks a lot in advance!

Regards,
Divya

 

[Mustaine3]

The wall thickness looks pretty thin. Why aren't you using regular shell elements?

 

[Megabeast]

Hi Mustaine3,

Thanks for replying.
The spool (cylinder) is split in to two halves, the top half being 9 mm and bottom half being 29 mm.
29mm is a bit thicker, so i used continuum shell elements with composite layup initially. Later i tried with solid elements, assigning solid homogenous section with smeared properties. Now it seems to give a much sensible result than before (i.e) first mode equal to 2Hz. Just that i cant see the twisting mode, its all bending modes. This might be due to smeared properties and modeling it as a composite layup might be the best option.
I can try with shell elements and check now. Is it easier to extract midsurface from solid model in CAE?


Thank you!

Regards,
Divya

 

[Megabeast]

Hi Mustaine3,

I now tried using shell elements and get natural frequency equal to 2 Hz (First mode. The two software (Abaqus and Pro mechanica) gives the nearly same natural frequency.
But the mode shapes look very different. I do not see torsional mode till higher modes in Abaqus. But in Pro mechanica third mode is the torsion mode.
Also, there is more bending (i.e) the modes are more flexible in Abaqus. I am not sure which one to believe and how to validate the results I have obtained in Abaqus.
I have used tied constraints in such a way that there is a master surface and two slave surfaces adjacent to it.
If you have any idea please let me know.

Thank you!

Regards,
Divya

 

[Mustaine3]

Maybe you have Hourglassing. Refine the mesh and use fully integrated elements or go directly to parabolic elements.

 

[Megabeast]

Thank you... I will run it with hourglass control and check.