NASTRAN - RBE2 usage

Mar 08, 2005 3 Replies

Folks-



Please be gentle on the newbie...



I have been asked to provide a customer with "interface loads" for a water tank in an aircraft application. We generally know the shape of the tank and the attachment points and methods are all well defined. Being statically indeterminant, I have been using MSC-NASTRAN to perform a rigid body analysis. (This is the approach our customer has specified as well.)



My approach has been to define nodes at each of the attachment points and the CG. I created a rigid (RBE2) with the attachment points as dependant nodes and the CG as the independant node. All okay so far...



There are several tie-rods with spherical bearings that support the tank assembly. These I modelled with bars (CONROD) and gave them artificially high stiffnesses.



Where I run into a problem is with two attachment points where the bearings are pressed into the tank struture. I figured that I would just apply the correct boundary conditions at these two nodes and be off to the races.



What I have run into is that NASTRAN will not allow you to apply single point constraints (SPC) directly to an RBE2. [I suppose this is because an RBE2 is a multi-point constraint (MPC).] To get around this, I created an additional node along the line connecting the CG with the problematic node (1% from attachment point) and created a short (~.25") beam (CBEAM) with artificially high stiffness between the RBE2 and the attachment point.



TWO SIMPLE QUESTIONS:



1) How is anyone else dealing with this type of problem?
2) Is the solution that I chose going to provide accurate answers?

Your comments are appreciated.



Mark Hampson Sr. Product Engineer GOODRICH Corp.



Yep.

CBARs with pins flags work as well.

SPCs can not be applied to the dependent DOF associated with MPCs, which include RBEXs.

Stiff spring (CELAS) elements in the right DOF work for that as well. It's best to use coincident nodes with springs.

Similar, but not exactly. That'll work fine as long as it is statically determinate.

You should go through a model checkout procedure to ensure no mechanisms exist. Also, watch for singularity warnings. Don't blindly use Param Autospc Yes.

Jeff Finlayson Structural Analyst

To me it is unclear what you want to achieve with your model. When you connect the CG of your structure to the support points using a rigid element (RBE2), you do indeed just that: you a create a rigid element interconnecting all your support points, effectively bypassing the stiffness of the modeled structure (which by definition is not rigid). Effectively you can just throw away your model and leave the RBE2, which I assume is not the aim of the analysis (since you say that the structure is indeterminate, I assume that you want to find the effect of this on your support load distribution).

I am afraid that to model anything like a realistic load distribution you will have to use the actual model without rigid element, and search for a method to apply the internal load of the water (which I assume is more or less what you intent to model).

Methods for applying an internal pressure load:

- create the internal pressure by applying the correct hydrostatic load distributions as pressure to the shell elements

- model the interior fluid by means of the virtual mass method (MFLUID)

- mesh the interior volume with fluid elements (MAT10)

The first method is by far the simplest. Normally your pre-processor will allow you to create a distributed hydrostatic pressure by applying element pressures over the boundary elements, with the pressure defined as a function of the coordinates of the element nodes.

Timo

schreef in bericht news: snipped-for-privacy@l41g2000cwc.googlegroups.com...

You get any of this Mark?

Join the Discussion

Have something to add? Share your thoughts — no account required.

Didn't find your answer?

Ask the community — no account required