Follow along with the video below to see how to install our site as a web app on your home screen.
Note: This feature may not be available in some browsers.
Largest Manufacturing Technology Community on the Web
Stay Connected:
new to FEA
- Thread starter 1993terry
- Start date
- Replies 21
- Views 4,789
I would suggest you get a good textbook on materials science and one on stress/strain/mechanics and look for as much background info as possible on this topic. As a short answer to your question, you need to know what the properties of the material are. ALL properties. YOU have to decide which of those properties are important to your design effort and the function that the design is to have.
As an obvious question, what effects on your design functionality would be incurred with the choice of glass, or aluminum, or polypropylene as the basic material? Vastly different properties, but perhaps any one of them would work FOR the FUNCTION.
As an obvious question, what effects on your design functionality would be incurred with the choice of glass, or aluminum, or polypropylene as the basic material? Vastly different properties, but perhaps any one of them would work FOR the FUNCTION.
Some good starting points would be "Mechanics of Materials" by Hibbeler and "Concepts and Applications of Finite Element Analysis" by Cook. Also, if you're learning finite element analysis, Solidworks Simulation may be the worst choice. It doesn't give you good control over meshing and its abilities to look at the underlying data, rather than the pretty picture, are limited. Ansys (classic, not Workbench) is my top choice for learning FEA. It will let you actually learn how FEA works and its pitfalls rather than covering it up. Once you know the basics, switching to Workbench or Solidworks Simulation is very easy.
Workbench and SW are much easier to use, but along with that comes being much easier to use wrong.
Workbench and SW are much easier to use, but along with that comes being much easier to use wrong.
triumph406
Diamond
- Joined
- Sep 14, 2008
- Location
- ca
the FEA package will have a library of common materials, occasionally you may have to add a material that's not in the library.
If your asking a question whether choosing the wrong will impact the results, yes it's fairly obvious it will. If the part is 4140, and you choose 7075, or vice versa then the results will be incorrect.
If you aren't capable of choosing the right material, you shouldn't be doing an FEA analysis.
Are you a student asked to design and analyze a part, or a recent graduate engineer starting at a company being asked to do an FEA analysis because they assume you know how?
Eng-Tips Engineering Forums is the proper place to ask this.
I_AM_MACHINE
Aluminum
- Joined
- Jul 14, 2015
- Location
- Nebraska, USA
I'm taking a course in finite element analysis in graduate school this semester. It is a fascinating subject. My experience is only based on the Abaqus software package, though I would believe that all the software packages would be similar with respect to your question. The typically important material aspects for deformations, stresses, and strains are typically Young's modulus and Poisson's ratio for the material. Steel typically used in the construction industry (I'm a civil engineer), for example, has a Young's modulus of 30,000 ksi and a Poisson's ratio of .3. If you are trying to model heat transfer or other types of analysis, however, you may need other material characteristics as well. I hope this helps!
Kevin
TeachMePlease
Diamond
- Joined
- Feb 11, 2014
- Location
- FL
Well, just in case you were unaware, punctuation, capitalization, and details are all free here, so don't be afraid to use/give them.
Mike Henry
Titanium
- Joined
- Aug 18, 2001
- Location
- Batavia, IL USA
That's inconvenient to do from a cell phone, if that's what the OP is using.
TeachMePlease
Diamond
- Joined
- Feb 11, 2014
- Location
- FL
Sure, if you're using T9 on your flip phone...
For most linear elastic structural modelling, E and nu will get you there. Things can get much more complicated than that though. 29000 ksi though for steel, not 30.
I've never been a big fan of using the FEA software to determine safety factors though. I prefer to handle the allowables in a different program. Otherwise you end up with a million essentially identical materials in your library. This is especially true when you're dealing with A-basis allowables and Fy and Fu vary with thickness and what shape it is.
As others have said, a course in FEA (or a good set of books) are highly recommended. The books by Wasim Yunis are a good start. They are aimed at Autodesk Inventor, but a lot of the theory is transferrable.
I do a lot of structural and mechanical design (not engineering) and do a certain amount of my own analysis before I bother paying the engineer to look at it. Both FEA and hand (with spreadsheets) calcs. But engineering analysis IME is a lot more about the engineer having the experience to ask the right questions and determine correctly how the item is going to be loaded than it is about the math.
I know that sounds stupid, but engineering is very much garbage in garbage out. Wrong assumptions with perfect math will screw you every time. The scary thing about FEA (not that FEA is bad) is just that it is so easy to set the analysis up wrong and/or analyze the results incorrectly. Because the results look pretty and it's easy to spot out a report that looks official we want to trust it. Hand calcs can be the same way but I think we tend to question them more.
That's maybe a bit OT but I say that to say pay particular attention to the boring aspect of the boundary conditions of your analysis. The constraints, loads, material, etc. Then check your results with whatever hand calcs you can to make sure you didn't screw up.
Two easy mess ups:
#1 Not setting up your analysis properly and ending up with a vector load when you need a moment load.
#2 Failing to account for buckling. Most basic FEA does not do buckling analysis out of the box. (Inventor doesn't without Nastran). Even if they do it's easy to get unrealistic results because in the computer you can have perfectly centered forces. It's alarming how much sooner a column buckles with only a few pounds of kick force instead of a perfectly vertical load.
FEA is a super cool learning tool though. You get to see how things fail in a very visual way and that is very helpful for folks like myself who are not engineers to get a handle on all that hocus pocus.
Good luck!
Leviathan
Sent from my SM-G930W8 using Tapatalk
I do a lot of structural and mechanical design (not engineering) and do a certain amount of my own analysis before I bother paying the engineer to look at it. Both FEA and hand (with spreadsheets) calcs. But engineering analysis IME is a lot more about the engineer having the experience to ask the right questions and determine correctly how the item is going to be loaded than it is about the math.
I know that sounds stupid, but engineering is very much garbage in garbage out. Wrong assumptions with perfect math will screw you every time. The scary thing about FEA (not that FEA is bad) is just that it is so easy to set the analysis up wrong and/or analyze the results incorrectly. Because the results look pretty and it's easy to spot out a report that looks official we want to trust it. Hand calcs can be the same way but I think we tend to question them more.
That's maybe a bit OT but I say that to say pay particular attention to the boring aspect of the boundary conditions of your analysis. The constraints, loads, material, etc. Then check your results with whatever hand calcs you can to make sure you didn't screw up.
Two easy mess ups:
#1 Not setting up your analysis properly and ending up with a vector load when you need a moment load.
#2 Failing to account for buckling. Most basic FEA does not do buckling analysis out of the box. (Inventor doesn't without Nastran). Even if they do it's easy to get unrealistic results because in the computer you can have perfectly centered forces. It's alarming how much sooner a column buckles with only a few pounds of kick force instead of a perfectly vertical load.
FEA is a super cool learning tool though. You get to see how things fail in a very visual way and that is very helpful for folks like myself who are not engineers to get a handle on all that hocus pocus.
Good luck!
Leviathan
Sent from my SM-G930W8 using Tapatalk
Mike Henry
Titanium
- Joined
- Aug 18, 2001
- Location
- Batavia, IL USA
No argument from me, aside from the typo.
TeachMePlease
Diamond
- Joined
- Feb 11, 2014
- Location
- FL
Doug's just channeling Ox... He regularly says "doo"... Doooo not ask me why.
Mike Henry
Titanium
- Joined
- Aug 18, 2001
- Location
- Batavia, IL USA
FEA is shit in, shit out. I've been mislead on many occasions to believe the colorful plots at first glance. Basically a lot of thought should go into the meshing. Areas on the part which you know will see the most "interesting" stresses should have a finer mesh than the rest. Due to computational limitations (you don't want the computer to churn all day unnecessarily) it's wise to look at symmetries and only simulate on a half part with half load for instance. I always eyeball things such as deflection against simplified hand calcs unless the part is ridiculously complex. Usually, features can be simplified to simple rectangular boxes so that you can figure out roughly what the FEA should spit out.
I pay someone else with the sheepskin to doo FEA, after I design the part and optimize it best I can, using the images in my head. Certainly you can picture how the FEA is gonna look.
Rarely do I need to revise the part much at all.
Milland
Diamond
- Joined
- Jul 6, 2006
- Location
- Hillsboro, New Hampshire
Good post - just watch some crane failure videos for graphic proof of how easy a buckling failure can occur.
YouTube (~30 seconds in. And I swear, the ship's name is only a coincidence...
).[Yes, I know it's an "O", not a "D"]
