Dear all
For the Seismic Analysis of Liquid-Storage Tanks, refer eurocode 3 and other documents, Hydrodynamic pressure have been modeled like convective mass with spring having stiffness.
As picture bellow
Anyone tell me, how can i model convective mass on RSA with spring (kc)
Thanks in advance
never done it but Probably you can can try very rigid bars with no mass and use bar elastic release to simulate the kc spring
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
Dear Rafa
I've model as that, but, have a problem as bellow
could u check it for me
Dear all,
i've modeling as this method
however analytical results have been problem, as bellow
i've att my model, anyone can be help me again
Hard to say without further information.
Which load case?
"supposed like this" is vertical or horizontal displacement ?
If it´s modal analysis ,modes 4 and 5 show similar displacement to your picture
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
If you mean the deformation in the horizontal direction you may refer to vibration modes 4 and 5, mentioned by Rafael.
Below are some screen captures showing top view of them for bars made of LINK-1 section.
Looking at these modes 2 observations can be made:
The screen capture in your first post and the document attached to it suggest that the convective mass should be connected to the walls of the tank by the set of simple longitudinal springs, acting in horizontal directions. The simplest approach for it in Robot seems using weightless truss bars with length, section and material corresponding to required stiffness. In your model the length is 5 meters (radius of tank), stiffness defined in releases is 1000 kN/m, "Nomass" material defined by you has Young's modulus E=2e5 MPa. In such case the section area corresponding to these parameters is 1000*5000/2e5 = 25mm2
It means that the truss elements to model the springs can be made of 5x5mm square sections. I have defined such section with label "spring", assigned it to bars 3to6 or your model, declared them as truss bars (Geometry>Additional Attributes>Advanced Bar Properties), deleted now unnecessary releases and defined additional support to prevent the node with convective mass from vertical displacement and rotations.
I have attached such model to this post. You can see the change of vibration frequencies of modes 4 and 5 for it comparing to the original model (3.57 Hz instead of 4.47 Hz)
For this model also the effect described by me in point 1. can be observed - the vibrations shapes related to convective mass are not along global directions but 2 mutually orthogonal oblique directions.
Best regards,
Hi Pawel,
That´s also a good way!
I just prefer rigid bars and elastic releases because there´s no need to create new supports and calculate new section properties. In his original model just a bar like in the picture below , would give the same results as your model alternative!
On my almost total ignorance in dynamics , I just keep thinking, why not link this mass to all tank nodes ?
At least 4x90 degrees ,but to the nodes in the whole height lenght
Wouldn´t it be more realistic?
Rafael Medeiros
Did you find this post helpful? Feel free to Like this post.
Did your question get successfully answered? Then click on the ACCEPT SOLUTION button.
Hi Rafael
"there are many ways to skin a cat"...
I prefer avoiding "rigid bars" because it is not quite clear and precise when explaining to others. You have defined a "rigid bar" with Ax=1000cm2 and Ix=Iy=Iz=1cm4 - it resulted in frequensies as in my model. But using "rigid bar" with the same Ax=1000cm2 but for instance circular or square shape would result in different results - because it would be "rigid" not only in longitudinal direction..
Not all nodes are linked because it is easier to model it this way - the number of springs is limited and their stiffness can be simply derived from stiffness related to convective mass.
In case of linking all nodes it would be more difficult to calculate the tributary stiffness corresponding to single link - especially in case of non-uniformly distributed nodes.
Regards,
