Simulating gravity by forces at nodes

Das Forum zu Z88OS (Open Source) / The newsgroup for Z88OS (open source)

Moderatoren: ccad, mz15, auroraIco, Lehrstuhl

Antworten
TomGroves
Newbie
Beiträge: 6
Registriert: Do 2. Jul 2015, 11:57

Simulating gravity by forces at nodes

Beitrag von TomGroves »

I'm trying to simulate gravity on a 2D object by distributing the force across the nodes (as described in z88mane.pdf, page 97, fig.2). The object is supported by fixed nodes with zero displacement at the bottom.
I'm seeing an odd result:
ForceConcentrate.png
(11.59 KiB) 1608-mal heruntergeladen
Why is the stress concentrated at the corners of the supports? Surely it should be distributed across the base?

Thanks,
Tom
Dateianhänge
FEM.zip
(559.99 KiB) 233-mal heruntergeladen
DanBil
Alumni
Beiträge: 39
Registriert: Do 5. Jun 2014, 06:25

Re: Simulating gravity by forces at nodes

Beitrag von DanBil »

Hi Tom,

the stress is very high at the supports, because there are local singularities (high stress gradient within an element). This is a typical phenomenon in finite element analysis. The reason for that is, that a linear finite element analysis can not consider the local plasticization, which would lead to downsizing and shifting of the stresses. Because of that a stress evaluation at the position of boundary conditions should be made in an integral way after a refinement of the mesh.

Kind regards,
Daniel Billenstein
TomGroves
Newbie
Beiträge: 6
Registriert: Do 2. Jul 2015, 11:57

Re: Simulating gravity by forces at nodes

Beitrag von TomGroves »

Really? Surely there's no local plasticization when a (relatively light) steel object sits on a hard surface?

Is there a guide somewhere on how to do the refinement and re-evaluation at the singularities? I'm new to FEA, and I was not expecting this complication :-/

Thanks,
Tom
DanBil
Alumni
Beiträge: 39
Registriert: Do 5. Jun 2014, 06:25

Re: Simulating gravity by forces at nodes

Beitrag von DanBil »

Hi Tom,

of course there is local plasticization in reality, but a linear finite element analysis can not consider this effect. For that you need to use a solver for nonlinear material behaviour.
Z88Aurora has one, but only for hexahedron and tetrahedron. So you have to make a new finite element structure with hexadrons or tetrahedrons to consider this effect within Z88Aurora.

Kind regards,
Daniel Billenstein
Antworten