[ATM] Re: ATM Digest, Vol 12, Issue 24

[artbianconi@blast.net]

In the early 90's I was an Application Engineer for PTC for Pro 
Engineer (Often referred to as Pro E). The VAR I worked with gave 
up Pro E and took on SolidWorks so I soon found myself helping 
clients find solutions with that. That was ten years ago and I 
can assure you that if anything Solidworks has become more 
powerful, not less as your statement erroneously suggests.

For one thing, Solidworks does not have to do linear or non-
linear calculations because it is a mechanical design software 
optimized for modeling complex assemblies. It is the not an FEA 
package nor has it ever postured itself as such. Ditto with Pro 
Engineer. 

Their ability to do any kind of structural analysis has always 
been dependent on the add-on capability from a third party 
developer. Several FEA software companies offered altered 
versions of their products so end users could run the FEA 
programs inside the parent CAD software and thus test their 
designs without the need for going back and forth through  
translation cycles. The physical constraints and spatial 
relationships were assigned in the solid modeler and were still 
present when the model was brought into the FEA software. The  
assembly constraints added in CAD were still there in FEA. The 
engineer only needed to specify materials, apply loads and select 
the constraints needed to emulate stresses. I distinctly remember 
doing a design of an overhead cam engine's intake system along 
with the spring retainers, spring, keepers, valve,  etc. And then 
watching as the spring compressed during the intake cycle. 
Because the valve spring was not being pushed into the realm of 
deformity, it was done with the linear software. Of that much I 
am certain of for no other reason was that my brain short 
circuits like Robbie the Robot when I try to figure out non-
linear "stuff".

COSMOS, proved to be the most popular with Solidworks and offered 
two levels of power: linear and non linear. Historically, the 
linear package proved much more popular. It was significantly 
less expensive and satisfied 90% of the analysis needs that 
mechanical designers ask for. The non- linear software only 
becomes necessary when the material properties and the loads are 
such that the material will deform significantly beyond any given 
modulus and there is a risk of permanent deformation and/or 
catastropic failure. Unless you have constructed your entire OTA 
out of Salt Water Taffy or licorice sticks, there is NO SUCH 
CONDITION in a telescope and thus no demonstrated need for non-
linear analysis. The load distribution, which is what we are 
concerned with in a mirror, will be accurately portrayed in a 
spectrographic pattern at each point in the cell where stresses 
are present. The beauty lies in the fact that a person with very 
little knowledge of the complexities of FEA can run a very 
capable stress test on the part and make an informed decision as 
to how the part or assembly should be altered to lower the 
concentrated stress. And it will cycle the assembly through 
stress cycles, capture the images and then play them back as an 
animation.

The COSMOS license ($5,000) on my my 2001 version of  that 
software (linear and non linear) expired long ago and with no 
client needs to justify the expense I chose NOT to order an 
upgrade to a later version. I have, however, ordered a seat of 
Solidworks 2005 which I understand has a very rudimentary FEA 
program which comes for free. Is THAT the one you used? Something 
that is free is hardly likely to perform with the degree of power 
and sophistication of it's $5,000 older brother!

For you to suggest that Civil Engineering software is better 
suited to calculating telescope stresses is so inconsistent with 
25 years of CAD experience and so far "out of ANY box" that I've 
come to suspect you are telling us that Tongue In Cheek.

Art Bianconi

> --- Peter <peter@kitgear.com> wrote:
> 
> > I've tried modeling a cell with SolidWorks/COSMOS
> > and discovered that this particular package cannot perform
> > the simulation successfully because it cannot perform non-
> > linear math. You'll need a much more expensive package
> > like one used in large scale Civil engineering work.
 
This post and any attachments were tested before being sent and 
found to be virus free.
Our Norton-Symantic virus list is updated daily.



_______________________________________________
ATM mailing list http://www.atmlist.net/