Boundary conditions on tank

[Weideman]

Hi,

I am trying to calculate the stresses in a model with a geometry similar to a typical basket, but with the one end longer, hence the center of gravity does not go through the middle of the model.

I modelled half of the model in Patran, with half of the 2 hanger rods (2 triangular hangers attached to tank) as well with the symm plane constrained as well.

I used a pressure load to simulate the liquid in the tank and fixed the top point in all 3 translation directions (representing the crane hook)


The problem is that the deformations I am getting is massive as the model shifts to the left (due to the unbalanced mass and the CG not being in the middle?)

Is this due to the pressure load that is normal to the surface regardless of the tank orientation? How would I constrain the model If I wanted to obtain the stresses in the lugs, hangers and bottom of tank?

If I "hold" the tank at the 2 lugs without the hangers I get better results..

Thanks in advance

 

[goeasyon]

How massive deformation it is? You could:
1. double check the material property definition, such as Young's module.
2. calculate the theoretical results with some assumptions to compare with the numerical results to exclude the software defect.
3. compare with other software with the same configuration

https://welsim.com

 

[gravityandinertia]

In addition to what goeasyon said, if you are running this as a static, linear elastic analysis and the loads are high enough and the walls are thin enough, this problem may need a non-linear solver. If the walls behave as membranes due to their thin nature, then solving in one step underestimates their stiffness significantly because membranes have very low bending stiffness and don't begin to resist much until after they start deforming and tension develops in them. This effect is captured in static analysis.

 

[Mustaine3]

I would also say that you get a motion just because of the imbalance in the system. But this would require a geometric nonlinear analysis, since the solver has to calculate the rigid body movement and the deformation separately.

With a good postprocessor (I'm using Abaqus and Abaqus/CAE) you could remove the rigid body movement afterwards, to see only the displacements that comes from the actual deformation.