Selecting finite element software 3

[ddddarkbull]

I am looking for a finite element software for dynamic, static as well as fatigue analysis (linear, non-linear, random, sin, combine random and sin, shock). Our products are typically electronic boxes (MCU) for aircraft application. Ideally, it should be easy to use for analyzing vibration response of printed circuit board as well as components, solder joint etc. The software should be compatible with Solidworks. The important aspects to consider includes user friendliness, built in library, cost. Ideally, the company should offer superior technical support which can help design engineer type of people to perform analysis work. I have used ANSYS in the past for this type of work. Is there any other better choice?

 

[rb1957]

"better" is such a variable.

is there something specific you don't like with ANSYS ?

does Solidworks prepare FE models and show results ? (or is it purely a drafting s/ware)

if your looking for a pre- and post-processor, FeMap is nice, PATRAN is Nasty (IMO)

 

[dwallace1971]

I cringe every time a see a phrase like "help design engineer type of people perform analyis work".

"On the human scale, the laws of Newtonian Physics are non-negotiable"

 

[IceBreakerSours]

ddddarkbull: If I were making this decision, I would think about the following (at least) because once you make this decision, the company will, in all probability, be stuck with this FEA package forever. Besides, but I am not sure since it depends on where you are, some of these decisions may have legal implications as well.

a) How critical/important is FE analysis to what my goals are? Do I need FEA from hereon as long as I can foresee or can I simply hire a professional consultant and get the job done? If not, do I have enough resources (financial, competent human, and time) to invest in FEA?

b) Is my "FE analyst" good at modeling? In other words, does the person "know" what the "black box" really does? Does the person know what needs to be done or are they supposed to learn it? What kinds of
"fancy stuff" will be involved (nonlinearities, plasticity, fracture, material modeling, etc.)? The more the "fancy stuff", the trickier the modeling gets. Should they be trained or should I expect them to learn by going through the documentation? How long will that take? Can I afford to spend that much time (and money)? It is well-known that FEA has a steep learning curve to it. Can my FE analyst solve some benchmark problems in the areas of interest and understand what the limitations are?

c) What types of analyses are of most interest to me? And which software package is relatively more well-known in those areas? For example, in crash testing, one package may be more well-known than the rest. This can be very helpful in the long run because my FE analyst will have an eco-system, and not just the technical support guys, to rely upon.

d) Do I need associated software (pre and post-processor)? Or should I go with a bundled package (ANSYS, ABAQUS, UG, LS-DYNA, and many more)? This may seem like a no-brainer but sometimes pre-processing (preparing the geometry, meshing, applying BCs etc.) can be harder than the analysis itself. The opposite is true in many situations as well. Where do our problems fit in in this continuum?

e) There are cheap and free alternatives too but help will be hard to find! For example, one of the founders of FEA at Berkeley has a freely downloadable software package and another one at Harvard sells his package for free if you purchase his book. There are others as well.

Most of the famous packages will be able to import files from SolidWorks. That said, ABAQUS was recently bought by Dassault Systemes which also sells SolidWorks and CATIA. In the long run, I'd expect these packages to work together very well. For example, in its next release, ABAQUS allows one to make a geometry out of the deformed output. That said, do I care about long-term seamless integration of these packages?

~!ce.

http://www.eng-tips.com/faqs.cfm?fid=376

 

[masomenos]

SAP2000 could likely meet your analytical requirements for $8k (US dollars) or less, including NL time history analysis and PSD random dynamic analysis. Not sure if they have a direct Solidworks interface, but SAP can import Nastran .dat and IGES formats. I've never modeled solder joints, but I imagine they could be done with solid elements.

Regarding "free" FEA software from one of the founders of FEA in Berkeley mentioned by IceBreaker in his post, I assume he is referring to Dr. Edward Wilson, who pioneered SAP development back in the 70's. If so, that free software is a 25+ yr old public domain software version with outdated solvers and outdated element formulations. If you are referring to another free FEA software our of Berkeley from a founder of early FEA development, please specify which one.

Does any FEA program really work "seamlessly" with Solidworks for this kind of application?

 

[IceBreakerSours]

masomenos:

You may be right about FEAP (Dr. Ed Wilson's FEA program) being outdated but, as far as I can tell, there are a few FE developers who like it because it is open source. That being said, FE development is a different ballgame.

http://www.eng-tips.com/faqs.cfm?fid=376

 

[Twoballcane]

“company should offer superior technical support which can help design engineer type of people to perform analysis work”

Be very careful on this decision. FEA is a tool; one will need deep knowledge of engineering theory to solve the problems you’re asking. It is the analyst, not FEA , that will guide you to a solution. The analysis should start with hand calculations and then move on to FEA and then try to correlate with one to the other. Most technical support will answer how to run the software, but I don’t think they will go into depth about engineering theory.

As for choice of software, in Defense you are not taken seriously unless the FEA is Ansys or Nastran.


Tobalcane
"If you avoid failure, you also avoid success."
“Luck is where preparation meets opportunity”

Perception is reality: You may build your skill set, but what you think of it doesn't really count. The only thing that really matters is what your boss and or customers think of it. Likewise, the real you is how others perceive you, not

 

[masomenos]

"As for choice of software, in Defense you are not taken seriously unless the FEA is Ansys or Nastran"

That sounds like a bit much of a sweeping claim. If you google "U.S. Dept. of Defense progressive collapse UFC 4-023-03" you will see in their design manual that the DoD uses SAP2000 for their progressive collapse analysis. Similar google searches show that Abaqus is also used by the DoD. I'd bet that there are a multitude of defense contractors also using Abaqus in various applications too in which the results they provide are taken quite "seriously" even though it's not Ansys or Nastran.

I agree with the importance of knowledge of engineering theory rather than counting on FEA support engineers to be able to help you with specific designs. FEA support engineers will tell you how to perform certain tasks with the program.

Icebreaker, thanks for the specific reference to FEAP. I don't think that was Dr. Wilson's software, but originally developed by Dr. R.L. Taylor who was also a pioneer in FEA development. Dr. Taylor was the first to introduce incompatible modes to avoid 'shear locking' with in plane bending of shell finite elements. The public domain SAP program which Dr. Wilson was involved with and available for free has not been updated in decades to my knowledge. FEAP, on the other hand appears to have some support among the open source community.