Hi Peter

The answer is in the the question!  "Membranes", by definition, have no
out-of-plane stiffness of their own.  What happens is any in-plane tension
in the membrane causes stress-stiffening in the out-of-plane direction -
just like a tensioned cable.  This in-plane tension may be as a result of
initial stress (eg from manufacturing process), or it may only be developed
as the membrane is deformed.  In your case, a thickness of .003mm over a 4mm
span (i.e. span/depth ~ 1000), is likely to be controlled entirely by stress
stiffening, with bending stiffness contributing very little.

As for analysis approach, that depends on the amount of initial stress.  If
there is significant initial tension, this can be modelled by many FE codes
with an initial (non-linear) static step, prior to the (linear) eigenvalue
solution.  This idea is to "lock in" the stress stiffness terms, then
calculate the frequencies of the stiffened structure.  If you can quantify
the initial stress regime, this should work well.  On the other hand, if
initial stress is small, and stress-stiffeneing does not get going untill
the deformations become significant, you have a seriously non-linear
problem - probably not "newbie" territory!

BTW - your name sounds familiar.  I think I did some analysis work for you
in my days at FEA Ltd.  Give me a call!

Regards

Philip Cooper
K W Ltd.
Fetcham Park House
Lower road
Fetcham, Leatherhead
Surrey KT22 9HD
http://www.kwltd.com
mailto:[log in to unmask]
Tel. +44 (0) 1372 371122
Mob. +44 (0) 7850 140538
Fax. +44 (0) 1372 371123

-----Original Message-----
From: FEUSERS is a list serving the researchers and groups active in
using and de [mailto:[log in to unmask]]On Behalf Of Peter Lovelock
Sent: 23 August 2001 12:13
To: [log in to unmask]
Subject: Membrane and Shell Elements


Hello All

First of all I must apologise in advance for the complete newbie dumb
question that I am about to ask.

I have a very simple frequency analysis (I initially want the first 20
modes) of a 4mm square silicon membrane which is 0.003mm thick - this
membrane is clamped on all 4 edges. I have meshed it using membrane
elements (3 DOF per node)and also shell elements (6 DOF per node) and
uderstanably I get completely different results for both.

What I want to know is which type of element would give me the more
accurate answer?

Thanks in advance

Peter