PDA

View Full Version : Updated FEA + Solid Modelling summary



Julie Matthews
02-16-1999, 01:19 AM
Dear Biomechers

I would like to thank everyone for their fast responses to my question
about solid modelling and FEA programs.

The summary.

I had 12 responses
2 said the they liked Solidworks with COSMOS
3 said they would recommend ProE with ANSYS
1 sales man recommended ProE! But later states that ProMechanica (which is
the FEA add on to ProE) is best for linear problems. However another
opinion is that this ProMechanica is not a good code. (I don't want to
start a war, everyone is entitled to their opinion)
1 person felt that neither Solidworks or ProEngineer was good for
non-linear problems, but suggests ANSYS (with the necessary non-linear
options), Abacus, or Adina
2 suggests ANSYS
3 people also pointed out other options, including UniGraphics, CATIA,
IDES, CADDS5, ANSYS, NASTRAN, ABAQUS, MARC, SDRC
1 person points out that we can use Materialise to make a 3D model from CT
scans for further FEA analysis

Gu?jon G Karason and Raj Basu reminds us that to get good and valid results
from any program, requires you to know what you are doing.

Marco Viceconti recommends that the training cost and time spent learning a
FEA program should be considered as well as the licence cost.

Naris Charoenporn suggest that www.wohlersassociates.com is a good site to
have alook at.

Thanks again

Julie Matthews

Below is my original post and the replies that I received.
------------------------------------------------------------------------
----------
We are currently considering buying some programs that will enable us to do
some prototype modelling and some finite element analysis of titanium screw
shaped orthopaedic and dental implants. We have looked mainly at
ProEngineer and Solidworks with COSMOS.

Questions
Do you have any suggestions of programs for solid modelling and/or FEA?
Are you using any programs that you can recommend for orthopaedic and
dental applications?

------------------------------------------------------------------------
----------
Marco Viceconti
I guess your question is not proper.
There are no orthopaedic or dental applications; there are more or less
simplified models. Medical problems by their nature are always of extreme
complexity and we always simplify them (but I guess this can be said for
any physical process).

If your target is research then you need the most powerful tools available
(or affordable with your budget) because you never know what level of
complexity you may need tomorrow. I have a very high opinion of Ansys
(which I use) Abaqus (which is probably better for material non
linearities) and I guess Adina can be a good as weel in a research setting.
All these codes have a long history in solving non linear problems (which
is a tricky business) and you can trust them (as much as you can trust any
FEM program, just a little).

In an industrial environment in most cases a serious non linear analysis is
too expensive to carry out no matter which code you have. For linear
problems (and some occasional non linearities) also codes such as COSMOS or
Lusas or NISA, which are less expensive can be fine, I guess. In any case
training costs and time associated to any FEM package are significant and
should be taken into account.

For the CAD side the subject become too big to be treated here. Only
consider that in ANY case you will have problems to have the CAD geometry
read in the FEM envronment. IGES is a big pain and direct translator
sometime are poorly written. Test before you buy, not with a cube and a
sphere but with dirty geometries with sliver patches, trimmed surfaces and
conbination of boolean and free form entities.

------------------------------------------------------------------------
----------
Mark Bliek
Materialise developed several applications for both orthopaedic and
dental applications with CT or MRI data as input.
These programs allow to export data in formats for CAD software and
prototype modeling. Furthermore after the design has been completed in
CAD, the data can be merged back with the image dataset to verify form
and fit on a patient.

A trial version of the software (MIMICS) can be downloaded from
htpp://www.materialise.com

------------------------------------------------------------------------
----------
Raj Basu,
If you plan on modeling and analyzing nonlinear biomedical implants, it
is recommended to buy a strong nonlinear FEA package. MARC, ABAQUS and
ANSYS have strong FEA capabilities. Out of these three, ANSYS is the
most user-friendly and the pre-processor itself supports a lot of solid
modeling tools (including BOOLEAN operations) and parametric modeling
(APDL = ANSYS Parametric Design Language).

You mentioned you were looking at COSMOS. I have used COSMOS myself,
and would not recommend it for non-linear biomedical applications (it
costs less though, ~ 8K).

As for the CAD packages, although UNIGRAPHICS, SDRC and CATIA are very
powerful CAD packages they offer very little or no nonlinear FEA
capabilities (the FEA capabilities are added as in-built "limited FEA"
modules from FEA vendors like ANSYS, NASTRAN, COSMOS, etc.)

The same is with PRO/E. Although PTC sells PRO/MECHANICA (FEA) as well,
but it is not a good FEA code. However, the cost of PRO/E (Foundation)
is only 6K compared to ~ 20K for the other three. The other three are
good for automotive applications where there is a lot of surfacing,
assembling and cosmetic design involved.

I would recommend to buy PRO/E as the CAD solid modeler and then for FEA
to go with either ANSYS or ABAQUS or MARC depending on the complexity of
your non-linear models and the material properties and also on the FEA
experience of your team-members.

------------------------------------------------------------------------
----------
Richard Bastien
Did you take a look at Unigraphics ? (http://www.ugsolutions.com)
They have one of the most important high end CAD systme with many tools
(like FEA, mechanism, simulation...) integrated.
I know of some prostetics companies working with it.

You might think there not a lot of difference with ProE buit, they also
have
a really efficient after sale service and support which is reallly helpfull
and usefull when you face problems and when you want to develop something
that is really difficult.
------------------------------------------------------------------------
-----------
Trevor Hearn PhD
Mr Dan Barker in our group is working with various FEA packages as
part of his PhD concerning metacarpophalangeal joint biomechanics. He
recommends ProEngineer as a one-stop linear elastic package.

------------------------------------------------------------------------
-----------
Hamid Katoozian, Ph.D
If you just want a solid modeling package, you can use ACAD.
For solid modeling and preprocessing prior to finite element analysis
PATRAN could be a good choice.
Selecting a good FEA code depends on the type of analysis that you are
going to do. For most orthopaedic applications particularly in case
of nonlinearity such as frictional interface or material nonlinearity,
I've used ABAQUS FEA code and I got very good results with a reasonable
convergence time.

------------------------------------------------------------------------
-----------
Myles Sussman/ George Pappas
They could buy Pro/E and use Mechanica. It's more expensive but they
don't have to be FEA experts.

See this book, FEA made easy with Pro/MECHANICA, for learning how to do it.

http://www1.clbooks.com/asp/bookinfo/bookinfo.asp?theisbn=1566901448

OR

http://search.borders.com/fcgi-bin/db2www/search/search.d2w/Details?&med
iaType=Book&prodID=15470554

Second email from Myles
I help to write parts of Pro/MECHANICA, so it's true I am a bit biased.
:-)

If you need to solve problems with very complex variations of porous
material properties, then Pro/MECHANICA is probably not the right tool.

If you can get by modeling with standard isotropic, orthotropic, or
transversely isotropic material properties, and you want to solve linear
static analysis *or* frictionless contact analysis (or what I just
helped write, geometrically nonlinear analysis of isotropic materials)
then MECHANICA does a good job. In order to get accurate results out of
most FEA codes you need to be a guru user, we shoot for providing
solutions for the non-guru.
------------------------------------------------------------------------
-----------
amit roychowdhury
As far as my experience is concern you can go ahead with Proengineer and
ANSYS 5.5 .
I have modelled 3D dental implant and total 3d knee joint using
Ansys5.3. But modelling part of ansys is not very user friendly,
but you can generate any type of surfaces and volumes.

------------------------------------------------------------------------
-----------
Gu?jon G Karason
I can not give any magic solutions regarding CAD and FEA software.
However, my opinion is that it is generally not the program the sets limit
to your work, but much rather your knowledge on the program used. I will
nevertheless try to give you some input on the subject.

Even though I have not done anything using ProE, I have heard some very
good comments on that program and its usability. It would be very
surprising if you wanted to do something beyond what the program is capable
of.

I had some experience of Cosmos a couple of years ago, and even though the
program is most likely more than enough for your application I would chose
Ansys for FEA.

But like I said before, be sure that you have someone to use the programs
before you buy anything.

------------------------------------------------------------------------
-----------
Sam Evans PhD
I have used Solidworks for a couple of years for designing
orthopaedic implants and it is excellent. It all works how it is
supposed to, it is clear and easy to use and even the manuals are
easy to read and tell you what you want to know. It is also
remarkably fast and powerful. I have found, though, that complex
screw shapes can be very slow to rebuild, and drawings of them in
particular can be slow and difficult to work on. Certainly you need
a fast PC with plenty of memory if you want to do this type of work.
Regarding FE software, I have used LUSAS in the past but this is
fairly limited in its solid modelling and irregular volume meshing,
although it is excellent for contact problems. I think now that
COSMOS/Works is probably the best bet for complicated 3D models and I
am trying to get a copy at the moment. Another option is MARC, which
is very good for bone remodelling or if you want to include your own
material models, because you can easily define your own subroutines.
Good luck!

------------------------------------------------------------------------
-----------
Naris Charoenporn
In addition to the software you mentioned, there are other CAD and FEA
software for creating solid modelling and analysing it with Finite
Element method. For example UniGraphics, CATIA, IDES, CADDS5 and others
those are CAD software enabling to create solid models and to link with
some FEA software such as ANSYS, NASTRAN, ABAQUS etc.

It is not easy to creat complicated geometry of human body parts such as
bone via the CAD or FEA software mentioned above. I have known that some
medical software enabling to interface with data for CT scanner and to
select and converse that data to be 3D solid model of bone and other human
parts. These 3D solid model will be saved in format of STL or IGES file
and then transfer to FE software for meshing and analysing them.

A good website concerning with CAD/CAE in medical fields is
www.wohlersassociates.com

------------------------------------------------------------------------
-----------
John Rasmussen
These options have the benefit that they easily model complex geometries.
Unfortunately, I am afraid that they will not do the job you are planning.

The reason is that an implant problem requires
- very complex anitotropic bone material data that vary from point to point
in
the structure
- contact problems
- dubious strain/stress ralations in the porous bone material

Neither of these facilities, I believe, are covered by the systems you
mention.
I fear that to treat the problem properly you have to go for advanced
systems
like ANSYS (with the necessary non-linear options), Abacus, or Adina. These
are
among the most advanced systems in terms of analysis facilities, but they
are
much more difficult to use and effectively require expert users. The
geometric
modeling facilites of these systems are typically less convenient than for
Pro/E and SolidWorks.

In spite of what the FEA vendors will tell you, there is regrettably no
shortcut to accurate finite element analysis of bio-related problems.

---------------------------------------------------------------
To unsubscribe send SIGNOFF BIOMCH-L to LISTSERV@nic.surfnet.nl
For information and archives: http://isb.ri.ccf.org/biomch-l
---------------------------------------------------------------