Variable Fixity Boundary Conditions 1

[Lyster]

I'm fairly inexperienced at modelling in Patran and I'm trying to model a plate with fixities varying from fully fixed to simply support and everything in between.
What I had though of doing is using spring elements between the plate and a fully fixed support then modifying the spring stiffness accordingly.
The problem I have is how to define the fully fixed and the simply supported situations without using arbitrary guesses of 1e8 for fully fixed and 1e-8 for simply supported rotational stiffnesses (for example).
The next issue (of course) is how to define everything in between, i.e. what number between 1e8 and 1e-8 is classed as 50% fixity for example?

Any assistance on this matter is greatly appreciated.

 

[40818]

Although used in the industry (aero) as an interpolation between fully fixed and simply supported when using hand calcs, the idea of say 50% fixity is a bit vague. How can something have a fixity value halfway between zero edge rotational stiffness and infinite edge rotational stiffness?
Depending upon what your modelling will guide you as to what dof's to use in your constraints.
If you give some explanation of your model outlay then maybe i can help.

 

[Lyster]

Thanks for your reply.
Basically I'm examining the effects on the stresses in a sandwich panel for different edge fixities. I'm using a/b=1 at the moment and defining all of my sandwich properties within the Patran laminate area (but that's not really part of the question).
I'm aware in patran that you can define boundary conditions on nodes by placing 0's in for constrained and null values for unconstrained. What I thought is that Patran is still a program and as such would need to represent constrained (null) as a number. So what I intended to do was create some nodes on the periphery of the panel (but not connected to the panel) and fully constrain those nodes. I would then connect some springs to those nodes and to the panel and vary the spring stiffness (hence boundary conditions) that way... I'm just a little unsure of how to do that and whether there is way of working out what the stiffness numbers should be...

 

[40818]

If all you want is to is basic sandwich panel stress analysis, then you dont need any FE. In fact your probably better off doing it by hand.
How are you modelling the sandwich panel?
Fully fixed condition is never really achievable in structures (apart from some civil uses etc). Therefore we tend to use the simply supported approach, as it will give conservative deflections,strains and stresses in the panel. Also a fully fixed analysis is sometimes carried out to obtain conservative edge stresses.
As far as your use of spring elements is concerned, your idea seems reasonable, if your going to use celas elements then they need to be coincident, but also make sure your nodes are in the correct coordinate systems. Dont constrain all the of the edge lengths as you will get unrealistic effects, i.e the corners wont be able to move the oposite way.
The best way is to use trial and error to learn, but if you keep your model simple so that you can use classical methods to know the results you should expect. This way your model will be more accurate.
You can calculate a "fixity" value, based upon its rotational displacement. You can work out what rotation is achieved under simple supports and you know you have zero rotation under fixed condition. Therefore, if it rotates 20deg (as SS) then a 50% fixity would give 10deg. Adjust the stiffnesses to suit.
Better off by hand though.