Problem with reactions

Johan_S

I have sent the below mail to support@mecway, but it does not seem to get delivered.

My problem concerns a tank loaded hydrostaticly in various combinations of g acting in the X,Y and Z directions. Any individual direction gives me the expected reactions, but as soon as I combine them, the reactions are grossly overstated.
I need to use the calculated stresses and my concern is that these are overstated as well. (The shape of the tank is such that it can not be checked with hand calcs)

My mail reads:

I just want to make sure that you have received this mail, because i need an answer urgently, as I have a client breathing down my neck.

I have since run a LC with the hydrostatic pressure applied in the X direction only, and the reactions are as expected.

**************************

Hi

My apologies for the big file, but this is what I am stuck with.

I hope I am not doing something stupid, but will appreciate any help.

In the attached file, my problems are:

The z-reaction in LC1 and LC4 are allmost exactly the hand calculated value.
The reactions in LC2 seems to be out by root3.
And LC3 ????
The Flexible joint/bonded contact on the +X side seems to give reasonable results. If defined the same way on the -X side the results are even funnier than what it is now.
This model runs for about an hour per LC. Can I make it more efficient?

I appreciate your help.

The file is in Dropbox : https://www.dropbox.com/s/qioelbxwdljp2ie/31_Cub_M_1.zip?dl

Johan_S

I attach a simple model to confirm my concerns.

In this case the force ratio is 1.5

disla

.

disla

I have found that my previous posts are only valid in particular cases which would not solve your more general problem. As far as I can see introducing two gravity loads do not manage them as independently. Hydrostatic pressure is distributed as if the points on the free surface also change the gravity direction.
I have tried to apply each hydrostatic pressure , one into the internal surface and the second to the external trying to separate them but seems to couple anyway.

( Ccx only allows one gravity load)

disla

There could be a workaround directly applying the hydrostatic pressure loads using the Heaviside function to exclude the empty side of the container.
I have separated each load on internal and external surface to be able to check if they are correctly applied.

Johan_S

Disla
Thanks for your reply and your efforts.
I have now only defined one hydrostatic pressure per compartment (as per your first post) and the different g's then sorts it out.
Thank you

PS The big file is removed from Dropbox.

Victor

@disla, that's interesting regarding multiple gravities. Perhaps it doesn't make physical sense to sum accelerations? Do you think that might cause some sort of impossible/incorrect/ambiguous situation?

The multiple gravity loads are internally summed and only the sum is used by hydrostatic pressure, so allowing multiple gravities is just a convenience for cases like this.

What's the advantage of using pressure with heaviside instead of hydrostatic pressure? They should be equivalent except hydrostatic pressure captures the transition between zero and non-zero pressure more accurately on a coarser mesh.

disla

@Johan_S First of all I have to say in no way I’m criticizing or questioning your model. Your model is impressive !!
My knowledge about this particular problem comes from traditional calculation of tanks under seismic actions.
Any particular code has their requirements and my comments are based on API650 where it requests to consider the sloshing height on the walls.

@Victor.

Impossible/incorrect situation: For me, multiple accelerations have perfect physical sense the same as multiple forces. The superposition principle can be applied as far as the material behavior is linear.
Multiple Gravities could be very useful to prepare different configurations of a container in horizontal (transport) and vertical (assembly) position for example.
Ambiguous situation : It may confuse the user if they think that two gravities (GX,0,0) and (0,0, Gz) instead of one (Gx,0,Gz) generate two independent hydrostatic pressure distributions.

The workaround of pressure with Heaviside is more related with this particular case as I was considering that itsn’t solved the hydrodynamic model. One should make a supposition on how the fluid and pressure will distribute inside the container once the acceleration arises. That makes the difference.
For a partially loaded Tank (50%), I don’t think the surface will lie perfectly into the diagonal under a (g,0,g) seismic acceleration. That would be the hydrostatic solution not the hydrodynamic.
Pressure with Heaviside would allow to be more conservative while passing the Sum of Reactions Check.
I have prepared a file showing the difference between “Two Gravities” versus “Two Heaviside pressures” for each direction.
Any comment will be appreciated. As I comment I never solved this problem with FEA.

Victor

@disla. I see. Two hydrostatic pressures with different gravities is more conservative than one hydrostatic pressure using the summed gravity. I guess this might be a problem best solved by a (not yet existing) feature to sum solution values across multiple configurations so you don't have to write those formulas.