Learning Forum

[Nizar]

Hello everyone,

This is my first post here, but I'm really stumped with this problem. I'll try to explain it the best I could, and you can ask me any question for further explination. So I am trying to model a soil-tunnel-structure model (I'll ask about that later in another discussion). In order to do that, I have to model the soil and its layers. It should be simple right ? I have defined my materials (only elastic properties and damping coefficients), applied a fixed boundary condition (I want to Apply eventually an absorbing boundary condition so I would appreciate if you could point me to some useful article about that), applied a nodel force (tabular force, and divide load by nodes option is turned off), defined a couple of nodes to view their acceleration results (in the direction of the force). 

 

So as you can see, here are the nodes. I am using quadratic order elements. The problem I am having is that if I check the B and C nodes' results, they should have the same acceleration value in the Z direction as they are at the same distance from the force, but that is not the case

 

Node B

Node C

So as you can see there is a big difference in the values. However if I try changing the mesh (it hugely dépends on the sizing), or the order back to linear, I'm not having such problems, however I feel like linear order is not accurate enough. I've done some research, and the only thing I can think of is the way Ansys distributes the forces to the adjacent nodes, or that it got something to do with me placing the nodal force on a mid node. I would like to clear this problem, in order to continue my work without worrying.

Thank you in advance,

Regards,

Nizar

P.S: I didn't attach the file, because i didn't know which one I should attach, and the associated folder is huge.

[peteroznewman]

Don't use Nodal Forces on quadratic elements without taking into account the shape function of the element.

A uniform pressure on a set of quadratic elements does not result in equal forces on each node.

Linear elements have a more predictable nodal force generated from a uniform pressure.

Use the geometry editor to divide the face and then apply a pressure or force to the face. Mechanical will properly distribute that force to the nodes, taking into account the shape function of the elements, be they linear or quadratic.

[Nizar]

Hello Peter,

Thank you very much for your answer, I did create such model and it did work, however I thought it was a lot of hassle to divide a solid into 4 parts just to apply a point load since I can apply it directly on a node. In any case, I will stick with  a geometric division.

Regards,

Nizar