FEA - Theory vs Practice 4

[DanielHosken]

What is the best way to learn about Finite Elements for CAE?
I'm mechanical engineering student. Lately (two weeks ago) i have read about FEA and i'm using softwares AutoDesk CFD and Ansys.
I'm reading this books:

Introduction to finite elements in engineering;
A first course in the finite element method;
The_finite element method a practical_course;
Um primeiro curso em elementos finitos (portuguese)
Elementos_finitos - a base da tecnologia (portuguese)

But there is theory a lot!
Its not a problem, but it seems so far from what i use in sofware.

For example:
iI followed all tutorials from Autodesk CFD... and i never needed to create a mesh. All is automatic! I just needed to select the materials and my boundary conditions (velocity, pressure, heat flow... etc). All very simple!
(

http://www.autodesk.com/simulation-cfd-tutorial-2015-enu

--- link with tutorials).

But i feel that theory is very important, even the software doing all alone.
So, i have two questions:

1 - Is there some good material that can help me? I think in some more practice... "Learn doing". I need connect theory and practice.
2 - What theoretical subjects i must learning? I have special interest about CAD applied in fluids and aerodinamic. But i dont know for where begin. I know about heat transfer, fluid mechanical.. but the classical concept. FEA looks another world thing...

Sorry if this question is so basic, but i really need some orientation to improve my study. In my university i dont study about it.

 

[rb1957]

you need to very careful of automesh ... sure it's easy, but you need to understand meshing and the problems it can make wit the results. sure you can automesh, but you should be able to mesh manually. Try some simple geometries, that you know the answer, like Kt = 3 at an open hole. Always be aware that you're using (probably) a linear FEA, which will over-predict stress in a notch ('cause it doesn't know about yielding). another simple model is a beam, modelled with caps and web.

another day in paradise, or is paradise one day closer ?

 

[stressebookllc]

Theory is important, but what is your goal in terms of a job?

If you want to work as a technical support engineer or an applications engineer at a company that makes FEA software, then absolutely you need to know how the elements are formulated, their important characteristics, solutions types, how the solver code works and also the application side of it.

But if your goal is to work in the aerospace industry for example, then depending on how much time you have, focusing on the application side may be better, with a decent understanding of how the elements work. That is just my opinion.

But if you are really interested in getting into the mathematics of FEA software and FEA, then you may need to get into a graduate school or a masters level FEA course, which goes into the FEA theory quite a bit, people even go so far as to do PhD's with special user subroutines that modify the way the code works, create custom element types etc. Then you can join the FEA software producing companies and get into the practical side of the application of that theory in an FEA software.

I personally have taken graduate school level FEA courses but I rarely find myself going into the depths of the code or element interpolation functions or isoparametric space etc.

So far I mainly had to focus on understanding the element behavior, solution types, when to use what element or solution type, using the FEA software for modeling efficiently, and also learning the other important mechanics and physics of aerospace structures as well as the procedures of the trade.

My course may be of use to you, may be not, but feel free to check it out:

http://www.stressebook.com/finite-element-analysis-course/

Aerospace Stress Analysis and FEA Courses

http://www.stressebook.com

Stressing Stresslessly!

 

[GregLocock]

In my opinion the best FEA engineers have more than a little experience with the real life equivalent of their models - stress measurement with strain gauges and so on, experimental modal analysis, fatigue testing, etc.



Cheers

Greg Locock


New here? Try reading these, they might help FAQ731-376

http://eng-tips.com/market.cfm?

 

[Shz713]

I agree with others.
try to look up FEA related specifically to your major, there are plenty in the market (just google or search your library). This is a good starting point too Link

Shoot for the Moon, even if U miss, U still land among Stars!

 

[DanielHosken]

Thank you.
Everyone help me a lot.

A special thank you to stressebookllc. Great answer!

 

[aerostress82]

Every advice above looks good. I wanted to give you another approach to this problem.
As you said, you are mostly into CFD, you will probably be using FEA at a basic level just to help you with your CFD projects. And with time and experience, you can easily master FEA as well. (2-3 years)

But for a beginner level, I would advise you to do some tutorials from Nastran/Patran (Nastran SOL 101 - also known as "linear static analysis") and understand what is going on behind all those calculations. (Sol 101 for Nastran. You can use other software too, but Nastran is the most common among really big aerospace companies as it has the most history among other software and was originally developed by NASA in the 1950s. Most probably your CFD loads in your future job will come from a software that is Nastran compatible. That's what I've seen so far. Some companies use ANSYS too if the structure has lots of CFD effects in it - for example: Aircraft Engines / Pratt & Whitney)

Rather than going into all those heavy text books, you could do this instead:
[ol 1]
[li]Just have 3 beams connected to each other in a horizontal line,[/li]
[li]Applying tension for from the right end of the beams,[/li]
[li]Apply 123456 constraints from the left end of the beams, (x translation, y translation, z translation, x rotation, y rotation, z rotation)[/li]
[li]Run the analysis in Nastran or any other software that you are using,[/li]
[li]Check the results,[/li]
[li]From the text books, try doing the same calculation by using Matlab or Excel. (3 beams will have 4 nodes in total, so your equation for manual hand calculation will have 4 * 6 = 24 lines in the matrix calculation - so you may use Excel too if you know how to perform matrix calculations. If not Matlab is an already automated tool for this kind of task)[/li]
[/ol]

This will help you understand how FEA works with nodes and elements and forces. So, no matter how complex the FEA is in the future, you will have some understanding to what is happening behind the scenes.
If you don't understand what I described above, just have it printed somewhere - you will understand it when you start doing workshops for SOL 101, SOL 103, SOL 105, SOL 106 and others.

This is how I started FEA in my graduate study while working on Airbus projects in an engineering office at the same time. So, it really helps you understand what you are doing in the long run..

And since you are a new graduate, you may want to put in 12-14 hours of work on a daily basis. The first 2-3 years after you graduate is crucial. I remember studying on my Master's courses and working on projects at a total of 12-16 hours a day sometimes. But always try not to burn yourself with too much studying. A good equivalence of fun and work will keep you happy during these first years. After you are specialized in an area, you will be more comfortable and won't have to study as much anymore. I don't work and study more than 50 hours anymore. After the first 3 years, it really calms down and you know what you are doing most of the time.

Spaceship!!
Aerospace Engineer, M.Sc. / Aircraft Stress Engineer

 

[aerostress82]

I think I must add a Step 7 to my above post:

7. Calculate Stress, Strain, Node Displacement, Element forces "for each element" after you solve your 24*24 matrix equation. As your calculation will be [K]*[Q]=[F], your whole matrix calculation will give you [Q] matrix values as a result (K: stiffness matrix, Q: displacement matrix, F: force/load matrix). FEA is a setup of your [K] matrix (for elements) and [Q] & [F] matrices (for nodes). So, you will understand how the whole calculation takes place only by checking a simple model of 3 beams (CBAR).





Spaceship!!
Aerospace Engineer, M.Sc. / Aircraft Stress Engineer

 

[jean15paul]

Let me start off by saying that I have a lot of experience with linear static FEA, but none with CFD, so my comments will reflect my experience. There is some good advice in this post, but not a lot of specifics. So I will provide a few specific resources.

First I'd say that most CAE engineers don't need all the academic theory and math behind FEA, unless you're a researcher or work for a company developing the code. I've taken graduate level FEA courses, and the theory I learned has not been very useful. Below are 2 resources that I do find very useful for FEA application.

1) The book "Building Better Products with Finite Element Analysis" by Adams & Askenazi.

http://www.amazon.com/Building-Better-Products-Element-Analysis/dp/156690160X/

This book is special because it's written by FEA experts, but it's about application, not theory. If you don't have a good FEA mentor at work to train you, then I'd highly recommend reading this entire book. Even if you do get training at work, this book is still an excellent reference.

2) The free whitepaper "Guideline for Evaluation of Finite Elements and Results" by the Ship Structures Committee.

http://www.shipstructure.org/pdf/387.pdf

This whitepaper is not a comprehensive guide to everything in finite elements, but it does give you some very useful guidelines for evaluating the quality of your FEA model. This is a good summary of the things you have to consider when building a GOOD QUALITY finite element model. Unfortunately some of it is specific to models made up of 2D elements, but in my experience (linear static structures) that's the most common kind of model.

Best of luck with your career.