本帖最后由 wdhd 于 2016-3-23 14:09 编辑
(循环对称结构模态计算及结果察看)
Product Line: MSC.Patran Product Name: MSC.PATRAN Customization/Other (0501)
Product Version: 9 Product Feature: Utilities/Shareware Menu
Article Type: FAQ Publish: Y
--------------------------------------------------------------------------------
Extended Description Problem Summary:
A few hints for using rotational cyclic symmetry in a normal mode analysis with patran-nastran (SOL 115):
- Define the cyclic symmetry in patran via:
- Create a cylindrical system where the z-axis is the axis of symmetry.
Let's call it Coord 1.
- Define the cyclic symmetry:
Finite elements - Create - Mpc - Cyclic Symmetry
Connect the independent nodes on one side of the segment to dependent nodes which are revolved around the z-axis over an angle of 360/N, where N is the number of segments.
The dependent nodes need not be part of the segment. They do not even need to be attached to elements. They are just used to define the number of segments in the model .
In most situations the dependent nodes are also part of the segment when the segment exactly represents a "piece of cake".
Note that here the Mpc has nothing to do with a rigid element, but is only used to define the cyclic symmetry in the nastran input deck.
The Mpc results in a CYSYM,N,ROT entry and two CYJOIN entries in the nastran input deck, one for side 1; the independent nodes and one for side two; the dependent nodes.
If a node lies exactly on the z-axis, choose it both as dependent and as independent.
This results in one CYAX entry.
Make sure that the independent nodes have lower theta angle than the dependent nodes,t.i. nastran will create the new segments using the right hand rule by rotating around the z-axis in positive direction.
- In order to use the patran cyclic symmetry utility (see below), make sure all the nodes are defined in this cylindrical system, otherwise the in-plane eigenvectors are wrong!
Finite Elements - Modify - Node - Edit: Analysis Coord system = Coord 1 -
Reference Coord System = Coord 1
- Analysis - solution type - normal modes - solution parameters - enable: cyclic symmetry
- write out a nastran input deck
- in the nastran input deck add above the subcase:
HARMONICS = ALL
NOUTPUT = ALL
and change:
VECTOR(SORT1,REAL) = ALL
in
VECTOR(SORT1,REAL,PUNCH) = ALL
- For post-processing in patran:
Utilities - Results - MSC/N Cyclic Symmetry Tool
Select your punch file - select a group - and select the coordinate system used to define the mpc's - apply
Your modes will be extended to your whole model subdivided in harmonics.
The number of harmonics will be N/2 where N is the number of segments. If N is an odd number N/2 will be rounded off to the lower integer.
(If you had defined independent and dependent in the wrong way, the results can look discontinous as the utility creates the model by rotating in positive direction around the z-axis while nastran used the negative direction in that case).
If you still notice discontinuous results over the segments, it is probably caused by the fact that some nodes are still defined in the global system. The utility simple reads the eigenvectors from the punch file. Eigenvectors are defined in the coordinate system of the grid point.
Furthermore, the punch file represents the vectors with respect to segment 1. So they are not rotated over an angle (360 /N) * (i-1) for segment i. That means the magnitude of the vector is correct but not its direction. So a fringe plot in patran of the displacement is correct but a deformation plot is very likely not correct for the in plane components.
The z-component is correct. A solution is to define all your nodes in Coord 1, then the eigenvectors are expressed in Coord 1 which is a cylindrical system. Now the nastran representation of the modes on segment 1 is the same for all segments as the coordinates are cylindrical.
|
发表于 2006-5-24 22:16
提升卡
变色卡
显身卡
楼主