Search questions and share your knowledge with the world here on Questionpoint.
Home > Mesh Analysis Interview Questions > Node Coordinates in FEM (Patran, Nastran and Matla...

:: Categories ::

Interview Question in Mesh Analysis


 

Interview Question :: Node Coordinates in FEM (Patran, Nastran and Matlab) Need Help


Hi, I am stuck in FEA/FEM and BEM coupling and need FEM/BEM expert's help. Thanks
1.How to calculate out nodes coordinates that meshed by pre-processor in Finite Element Analysis? Is there any formula for this calculation?

2. Can patran or nastran communicate with Matlab via .xls or anything?

3. What is exact solution? How should I calculate out exact solution if compared with FEM solution?

4. What are the characteristics of paver mesher? Does it follow certain rule in its numbering of nodes and elements ?
Answers to "Node Coordinates in FEM (Patran, Nastran and Matlab) Need Help"
RE: Node Coordinates in FEM (Patran, Nastran and Matlab) Need Help?

I'll try to answer these in the order asked.



1. If you have a simple structure (i.e. a flat plate) and you know what the mesh size is, then, yes, you could determine what the nodal coordinates are for the mesh. But, why not just write out the nodes & coordinates for the mesh from your pre-processor?

2. You can write out stiffness matrices from Nastran that could be imported to Matlab. What type of data are you interested in porting to Matlab?

3. An exact solution is a closed form solution (i.e. stress = M*c/I for bending in a beam). Finding a closed form or exact solution will depend on the complexity of your model. Although, you generally find a subset of your model that you can solve for displacements or stresses with a closed form solution.

4. Node numbering in Patran is just done consecutively and the direction of nodes i, j, & k will specify the direction of the element normal for shell elements. Patran doesn't worry about nodal numbering in order to minimize matrix bandwith.



Updating based on your new info concerning question 1:

First, I would determine the half perimeter of the sphere based on the formula s = r*theta, where theta equals 180 degrees. Then, I would divide this length by my required nodal spacing. This will give you the number of element edges or segments (if you add one to that value and it should give the total # of nodes along that half perimeter). This will be your loop counter. Then, knowing your nodal spacing, calculate the angle between nodes based on s = r * theta. Now, nodal locations in x,y space can be calculated using the conversion from rectangular to cylindrical coordinates (I'm viewing this as a 2-D problem right now) - r = x*cos(theta) + y*sin(theta). Once you get the nodes laid out on this half perimeter, then you can just transform them (in spherical coordinates) about the center of the sphere one half perimeter at a time until the whole sphere is addressed.
 
Vote for this answer ::  
Update Alert Setting
Advertisement2

Top Tags



- Sharepoint Developer Interview Questions - Java Interview Questions - J2EE Interview Questions - ASP.NET Interview Questions - MOSS 2007 Interview Questions - WPF Interview Questions - Microsoft CRM Interview Questions - Technical Interview Questions for Programmers - SSRS Interview Questions - SSAS Interview Questions
Home | Terms Of Service | Contact Us | Site Blog | Investors | Advertise With Us | Disclaimer | Privacy Policy | Questionpoint Widget
© CERVO.BIZ 2010 User-posted content, unless source quoted, is licensed under a Creative Commons Public Domain License