Answering a reasonable question: what is Finite Element Analysis?
Reading the principles of Finite Element Analysis (FEA) had always
spiked my hair back horizontally, for negligible retention.
It's a learning style issue. I've tried reading mathematics books on
mathematical methods I have "reinvented" independently and I can't
understand a word the book says.
So I came to the idea that I should play around with a FEA package,
letting the "tool" do its thing, then let the accumulating
observations and interest draw me towards the theory.
That also matches my path to now, as a practical learner.
Be aware though that my practical and playing-around approach can make
me an innovator. I have broken new ground in science and engineering.
Just in case anyone thought the approach is "lesser".
Support for this view came from an interesting 11 minute "YouTube" video of a
mathematician Gilbert Strang expounding his thoughts and perceptions
on the Finite Element method ("Finite element method - Gilbert
Strang").
He describes Finite Element Analysis as being the biggest dispersed
teamwork endeavour in human history, with progress flowing back and
forth between engineers, mathematicians, computer scientists and
others (8min 05s on video). Incrementally feeling their way forwards,
finding their way, staging by useful intermediate goals.
So the human picture is no one human ever went in possessing a
comprehensive forward view.
Therefore, it seems reasonable that I should not expect myself to have
go in guided by a comprehensive forward view...!
My Web search and read on FEA had an encouraging message; the FEA I
wanted to achieve, stresses in structures, is the simplest case of
FEA.
The solution is static (you put a load on and the structure
takes up a stressed state; you come along and deduce what that
unvarying stressed state is).
It's linear-elastic because of the application (for structures, you
generally don't want the structure deforming; therefore stresses
for eg a bridge should be in the elastic range - whereas
deformation introduces non-linear behaviour which can now be modelled
but is an extra sophistication)
[ Personal backdrop to this endeavour studying FEA]
So I felt ready to go...
The tutorials for LISA FEA ( LISA website , see tutorials ) chimed-in right for me. Explanations were simple. Emphasis on nodes and elements, getting them in place to make early progress on projects, seemed promising for the "get started somewhere" approach I visualised.
LISA FEA is not "free software"; but the limited version capped at 1300 nodes is free. I'll see where I go onwards later...
I followed the first LISA FEA tutorial ("Tutorials -> Analysis types
-> Static").
It worked! So I'd got an installed FEA program and could broadly get
it to respond.
My first ever FEA simulation (precious only to me, for sure!)
In 2011 I did research on fatigue-resistant T-fillet welds, for naval
and structural applications.
eg shown in this webpage
Could I now model those welds?
Here is where I had a go....
I tried modelling the stress-concentration field around the tip of a
crack.
In any practical sense that's totally inadvisable.
However, it produces pretty pictures and the actual stress-fields are
known from engineering investigation trying to avoid fracture of
components. So you can kind-of get some idea of how well you are
doing...
Modelling of stresses in a cracked sample
As previously mentioned, I've done a research programme on fatigue-resistant welds. That went to some detail; including metallography of the weld toe showing early-stage fatigue-cracking.
Those findings in 2011 were controvertial.
With developing ability at applying the FEA computational tool, could
I now perform FEA simulations providing useful explanation of those
observed outcomes?
I'm applying planes-of-symmetry to FEA solutions by this stage.
Here is the outcome: simulations of fillet-welds with different weld-toe features .
This programme of work is learning how to use shell-elements to build models of pressure-vessels
Reminding readers that this is early-stage work, and there is a high
chance of "a well of sense in an abyss of stupidity" (!!!) errors.
So these "results" should not be taken too literally.
Copied the tutorial - it worked. No pictures taken.
Putting hemispherical ends on a cylindrical pressure vessel would have produced more than one challenge at once. The single-step next challenge was therefore a spherical pressure vessel like a Horton sphere , as seen at petrochemical refineries.
My spherical pressure vessel simulation.
Quick check of some fundamentals, with this simple cylindrical vessel simulation.
Hemispherical-ended cylindrical pressure vessels are used in real
applications.
The stresses in the cylindrical body should be different from that of
the hemispherical ends; so the output is anticipated with interest
My simulation of a hemispherical-ended cylindrical pressure vessel .
A couple of FEA simulations applying axisymmetric elements .
Line-elements enable simulation of structures with a lot of space in-between structural components which are long and of constant cross-section.
Some initial simple "practice" "beam" line-element simulations .
These modelled lattice-structures representing truss-bridges are simulated using line-elements.
Warren-truss and Vierendeel-truss .
Beyond here is breaking-out beyond what fitted in a calendar month
from first downloading a Finite Element Analysis program and giving it
a try...
Comment on
the backdrop
to that first month.
My onwards efforts...
My mixed line-element and shell-element FEA model
My orthotropic bridge deck FEA model
This is an "underlying science" / "engineering fundamentals" exercise.
A significant application of FEA is visualising the non-uniform stress
distributions associated with features of the physical object.
Here is
stress distribution in a simple centrally-loaded cylinder
;
as an exercise in thinking about how stresses distribute in a body.
Another in this series; double the length, at 4m . Diameter and loading condition stay the same.
These Warren-truss and Vierendeel-truss simulations are identical in every aspect apart from being different types of truss.
This presentation is a new departure, as I am now inviting engineering comparative evaluation of the merits of different structures on the basis of Finite Element Analysis models of them which I have done.
Comparing otherwise-identical Warren and Vierendeel Truss models .
I've
summarised the outcome
of this, my initial Finite Element Analysis endeavour.
At this juncture, it looks that the way ahead is a new different
programme of work, of a different character, starting a new chapter.
I hope you have enjoyed sharing my initial Finite Element Analysis modelling and simulations experiences.
(R. Smith, 17May2016, frequent updates to 02June2016, 06June2016,
14June2016, 16June2016)