Newbie Question 2

[haynewp]

I have hardly ever done FE and have a basic question about finite element software, just out of curiosity. We have RISA and RAM Advanse which have finite element capability. But how would these compare to a program like Nastran or Abaqus? I have always heard about these programs...Besides the number of elements available, what makes these programs so much better than RISA or RAM?

 

[GregLocock]

Horses for courses. If you are analysing a structure in a way that is compatible with codes written around hand analysis, where the loads are well defined (if not necessarily realistic) and the details of the structure and connection types are limited and well understood, then the additional flexibility of a general purpose FEA program is of no value.

As an example - if you have a bolted joint with 3 bolts, in clearnace holes, in the beam, and six in the column, you can probably specify that for a building FEA in about 3 seconds flat (guessing).

Whereas if we were to analyse it the way we do car structures we'd use an element size of about 1mm and would need a supercomputer, given that you've got clearance holes and contact. There is a joint pretty much like that in a car, some poor bugger's life work is meshing it.




Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[GregLocock]

On the other hand - if we analysed a building the same way we analyse a car, then we could look at the stresses during the earthquake, and model what happens as each joint fails, and what happens when it all hits the ground.

Fun, but useless, for buildings. Not much fun, but very useful, when we crash cars into barriers.




Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.

 

[GBor]

I think Greg hit it, but here's an example:

Some friends of mine (civil engineers...I mix with a bad crowd recently had an issue where they had designed a non-traditional joint in a building and someone questioned the accuracy of their RISA analysis. This building was extremely complicated, but their beam model did a good (and very fast) job of calculating reactions in the beams...shear and moment diagrams, etc. It was, as Greg mentions, the joints that were the area of concern.

In the joints, they had connected several beams by welding "boxes" with multiple faces that lined up to where the beams would come in. I extracted their loads from RISA and modeled the detailed joint using very refined plate elements to get a detailed load path. It showed some issues and how to quickly and easily resolve them.

I love FEA...bit of a fanatic...play with software in my off hours after the children go to bed. NEVER FORGET...it is a tool. Don't bring a rack to dig a hole...don't bring a knife to a gun fight. I support three different packages that, in my opinion, are useful for three different functions. One is very inexpensive with great tools for importing and cleaning up CAD models. One I've used for over a decade...I'm very fast with it and it handles most of the problem types that I may need. The third has very powerful contact algorithms and an explicit solver and runs a good bit more efficiently than most softwares (don't need a supercomputer to run significant contact).

Make sure you know what types of problems your tool can address, and learn the right way to use it to its fullest capabilities. Don't try to use it for something that it isn't designed to do, and make sure you are using it correctly by periodically running a problem that you CAN hand calculate and comparing to your results.

 

[abusementpark]

Some other features of more advanced FE software include higher order elements. ANSYS would put structural FE program to shame in terms of element library.

Also, the non-linear, time-history, analysis features are much more sophisticated. Some programs allow you to define you stress-strain curve for a given material.

ANSYS, ABAQUS, etc... all have capabilities that help you build highly refined models in more automated fashions.

The structural programs like RAM and RISA are more geared toward design (and they do a great job!) with the implementation of code checks and section optimization features. They also have better GUI's imho.

 

[hickdawg]

Interesting thread. An analogous situation to this discussion is one I frequently remind my crew about - there is no justification for spending $100,000 in analysis to save a thousand puonds of steel in a static structure.....assuming that the structure is on land, that field fab is acceptable, that there are no extenuating circumstances that preclude the use of a structure that is 1000 lbs heavier, ad infinitum, ad nauseum. The application and finacial realities of the project decide; not only which FEA approach is best but which method of analysis in general is appropriate. Sometimes I grab a steel handbook - sometimes I spin up an FEA model. I find that some of the best "art" in engineering is selecting the most appropriate flawed solution for the problem at hand.

As a response to the initial question about what makes one better than the other - try triming your fingernails with a bolt cutter or splitting firewood with a scalpel - its about choosing the right tool for the job.

 

[feajob]

Hi GBor,

You wrote that

"The third has very powerful contact algorithms and an explicit solver and runs a good bit more efficiently than most softwares (don't need a supercomputer to run significant contact)."

Could you please mention the name of the code you are using for contact analysis? And, if it is possible please mention what kind of structure you are dealing with?

From my personal experience, doing the contact analysis efficiently is usually a good challenge.

A.A.Y.

 

[GBor]

The software with the advanced contact algorithms is called Advanced Multi-Physics Simulation (AMPS). The most significant contact analysis I've done with this tool can be seen here:

http://www.ampstech.com/ampstech/Asp/Solvers.asp

It's the one on the left.

 

[GregLocock]

Crush loads on signal cables eh ? I worked on marine seismic streamers for a couple of years.

Cheers

Greg Locock

SIGlease see FAQ731-376 for tips on how to make the best use of Eng-Tips.