Sabir,
My example was done as a quick test 4 years ago and most packages are now
able to get within 5%.
I have more severe example which approximate some of our problems better,
but this simple problem has been tested on several systems.
OD= 85mm ID=6 CrossHoleDia=24 filletRad=2 Symm X,Y and Z H=50
Press=200MPa : Stress MaxPrinc MPa
Boundary Element and ProMech got within 1% (tho' the latest Mechanica
default settings give -2%)
I have mainly looked at automated meshing as a guide to what a designer can
expect.
Procision, gives the worst results, even with iterative solution technique.
(even worse than the -21% given by the DesignSpace5 worst setting)
Regards
John
-------------------------------------------------------------
John Milroy
Woodward Diesel Systems
(was Lucas Bryce, then Delphi Diesel Systems)
Fax +44 ( 0) 1242 27 7277
Hatherley Lane, Switchboard Tel 01242 27 7000
CHELTENHAM, W J Milroy Tel 01242 27 7149
Glos. ENGLAND EMail [log in to unmask]
GL51 0EU NOTE TEMPORARY ^^^^^^^^^^^^^^^^^
//||/||________________________________________________
Opinions expressed are not necessarily those of my employer.
----------
From: Sabir Ahmed
To: [log in to unmask]
Subject: Re: Procision evaluation
Date: Monday, August 27, 2001 7:44AM
Hi John,
I agree that one will be severely limited on the choice of the CAD tool
since I am looking at it from a designer's & a service provider's point of
view. I understand that Rand is planning to release a CAD independent
version of Procision - though their release schedules are not clear yet.
Did you have a chance to validate your thick walled cylinder problem with
another FEA solver. How far off were the results?
TIA
Sabir
On Tue, 21 Aug 2001 18:25:00 +0200, MILROY, John
<[log in to unmask]> wrote:
>It only works with (inside) ProEngineer
>
>I tried a simple thick walled cylinder with cross hole and intersection
>fillet with internal pressure and the results were terrible, even with the
>new iterative solution technique.
>
>Apparently I may have got better results if I split the geometry into
>sub-parts, but my ProE does not allow that as the Procision function to do
>that calls up a surfacing license (which we do not have).
>
>Even if we could do it this way this seems like a hard way to do meshing
(by
>hand).
>
>Why mesh manually when ProE has the "perfect"? analysis partner in
>ProMechanica which has good automeshing and adaptive P elements to give
>accurate results (mostly?)
>
>Regards
>John
|