HOUR GLASS 3

[prasannaah]

Dear learned ones,

why do the tetrahedral solid and triangular shell elements doesnt have hourglass modes?
Thank you.

Prasannaah

 

[EngAddict]

Because of the inherent nature of a triangle, one end cant rotate about the other. But as stated the downside is shear locking causing your model to be overly stiff.

 

[rb1957]

"hourglass" elements are created in error ... if the proper node sequence for a quad is 1-2-3-4, an hourglass would result from coding the element as 1-2-4-3. tirangles don't have this issue, as noted above; 1-2-3 is the same as 1-3-2.

 

[prasannaah]

Thank you,

But please tell me why only LS DYNA has hourglass mode, in detail.
Thanks again.

 

[prasannaah]

I heard there is some kind of change in formulations involved. Can any one guide me with respect to such stuffs?

 

[rb1957]

? hourglass elements are errors, mistakes, and can't be used in analysis (if you hope to get good, reliable, answers).

 

[prasannaah]

ONE MORE IMPORTANT DOUBT I NEED TO CLEAR IS:

what happens when hourglass is :applied during simulations?
: not applied during simulations?

 

[jhardy1]

ONE MORE IMPORTANT DOUBT I NEED TO CLEAR IS:

what happens when hourglass is :applied during simulations?
: not applied during simulations?

Hourglass elements are a modelling error.

If you use hourglass elements in your analysis, your results WILL be erroneous (at least in the regions containing the hourglass elements).

If you don't have any hourglass elements in your analysis, your results MIGHT be OK. (To put it another way, a complete lack of hourglass elements is a necessary but not sufficient condition to ensure valid finite element modelling.)

If you don't understand why this is, your results are almost certainly meaningless!

http://julianh72.blogspot.com

 

[rb1957]

ditto ... you should check your model for twisted elements, most "front-ends" do this, your FE output will almost certainly flag these elements.

 

[EngAddict]

I think by "hourglass applied during simulations" you mean to say hourglass 'control' applied during simulations. This could be things like selective reduced integration or more specifically the methods stated in the link you posted. They may take longer to run but give better results. Whereas without hourglass control enabled you run the risk of getting some 'funky' displacements depending on your mesh refinement and your application / critical areas of displacement.

 

[IDS]

The linked notes are quite clearly not talking about hourglass elements created by user error. The link below seems to cover the topic well (from a very quick look):

Reduced integration does a lot of good things for an element: it reduces
the computational cost, it reduces shear and volume locking, and it generally
softens the element so that the predicted stress is more accurate. However,
reduced integration also makes elements too soft in the sense that modes
other than rigid body modes aren’t resisted by the element. These modes,
which were originally noticed in finite difference calculations in two dimensions
in the 1960s, are historically called hourglass or keystone modes because
of their shape. For other elements, the modes don’t have these shapes, and
the modes are commonly referred to as zero energy modes.

http://www.feapublications.com/pages/pdfnews/article.3.9.03.pdf

Doug Jenkins
Interactive Design Services

http://newtonexcelbach.wordpress.com/

 

[TERIO]

EngAddicit is pointing you in the right direction. For what it's worth hourglassing (at least in the sense the OP is talking about) is not an error of having the wrong node sequence. Hourglass (zero energy) modes can happen with under integrated (reduced integration) elements. Basically because of the formulation the elements can deform into some shapes without causing strain. I believe it tends to be more of an issue in explicit dynamics (like LS-DYNA) because the small errors add up over many time steps. The result looks something like the attached sketch, i.e., pairs of elements look like hourglasses.

 

[rb1957]

ok, i'll accept i didn't read the attachment (mea culpa).

but "hourglassing" due to loading (i guess it's the torsional mode of the quad (allowing one end to rotate relative to the other) isn't a good thing. i guess too these are membrane quads (not plate elements).

 

[prasannaah]

Hey guys i found a pdf related to tis issue, and i thot i cud share with u all. plz find the attached file.

 

[kellnerp]

Cook explains hourglass modes.

faq1183-1711

TOP
CSWP, BSSE

http://www.engtran.com

http://www.niswug.org

http://www.linkedin.com/in/engineeringtransport

"Node news is good news."

 

[jhardy1]

Apologies for also leaping to the wrong conclusion about what "hourglass" meant in this context. EngAddict and IDS are pointing you in the right direction.

http://julianh72.blogspot.com