r********o
发帖数: 1423 | 1 就是laminate这个单元，在分层的时候，每一层nastran是当作什么处理的？是不是就是
把每一层相当于看作是板单元（假设是Quadrilateral)里面的CQUAD4或者CQUADR?我想如
果把每层的材料性质都输成一样的isotropic材料的话，那么laminate单元应该就相当于
同厚度的板单元。
如果这个假设成立的话，那么由于CQUAD4的节点自由度是5，CQUADR的节点自由度是6，
laminate单元的节点自由度应该和自己选的Quadilateral的自由度是一致的。（CQUAD4或
者CQUADR可以在run model之前的advanced option里面选），由于CQUADR是考虑in plane
loading的，我就选了CQUADR,结果run出来以后，后处理项里面就没有laminate各层的
data了。。。
俺正在给公司里面写FEA manual,才疏学浅，万望各位大牛指教。 | p**e
发帖数: 126 | 2 the laminate theory assumes the classic plate strain distribution across the
thickness. therefore for the the entire laminate, there is no separate
DOF for each layer. only that the integration is done for each layer to get
the appropraite K. after getting the solution, then back calculate the
stress/strain at each integraion point at each layer.
read through theory manual, and you will see how it is treated. the laminate
theory incorperated in all finite element codes (ansys, abaqus, nastran) i
【在 r********o 的大作中提到】
: 就是laminate这个单元，在分层的时候，每一层nastran是当作什么处理的？是不是就是
: 把每一层相当于看作是板单元（假设是Quadrilateral)里面的CQUAD4或者CQUADR?我想如
: 果把每层的材料性质都输成一样的isotropic材料的话，那么laminate单元应该就相当于
: 同厚度的板单元。
: 如果这个假设成立的话，那么由于CQUAD4的节点自由度是5，CQUADR的节点自由度是6，
: laminate单元的节点自由度应该和自己选的Quadilateral的自由度是一致的。（CQUAD4或
: 者CQUADR可以在run model之前的advanced option里面选），由于CQUADR是考虑in plane
: loading的，我就选了CQUADR,结果run出来以后，后处理项里面就没有laminate各层的
: data了。。。
: 俺正在给公司里面写FEA manual,才疏学浅，万望各位大牛指教。
| r********o
发帖数: 1423 | 3 Thanks for yuor reply, hehe, I was dealing with Laminate theory and thin plate
/shell theory previously for my research. Though sandwich structure it is
treated, the one layer(mid-plane) assumption is taken inside FEA. My question
is for the DOF of laminate, when I select the CQUAD4(5 dof) or CQUADR(6 dof)
for the solution, the data for the failure index for each layer are only
available for CQUAD4, while not available for CQUADR. Therefore, I was
confused about the way how Nastran treats the La
【在 p**e 的大作中提到】
: the laminate theory assumes the classic plate strain distribution across the
: thickness. therefore for the the entire laminate, there is no separate
: DOF for each layer. only that the integration is done for each layer to get
: the appropraite K. after getting the solution, then back calculate the
: stress/strain at each integraion point at each layer.
: read through theory manual, and you will see how it is treated. the laminate
: theory incorperated in all finite element codes (ansys, abaqus, nastran) i
| p**e
发帖数: 126 | 4 classic laminate theory does not support in of plane torsion. therefore the
6th degree of freedom has to be zero. I don't know exactly how nastran is
implemented. but that might be the reason. in abaqus such problem does
not exist. in both abaqus and nastran, laminate is just add add-on to the
classic shell theory. no special treatment necessry just have to integrate
across the layers.
also in nastran manual, it seems only the cquad4 supports composite.
abaqus is much stronger in composite than
【在 r********o 的大作中提到】
: Thanks for yuor reply, hehe, I was dealing with Laminate theory and thin plate
: /shell theory previously for my research. Though sandwich structure it is
: treated, the one layer(mid-plane) assumption is taken inside FEA. My question
: is for the DOF of laminate, when I select the CQUAD4(5 dof) or CQUADR(6 dof)
: for the solution, the data for the failure index for each layer are only
: available for CQUAD4, while not available for CQUADR. Therefore, I was
: confused about the way how Nastran treats the La
| r********o
发帖数: 1423 | 5 Thanks for your reply, it's very helpful. I did run several additional simple
models with in-plane torsion of plate element and laminate element with same
isotropic material on each layer with CQUAD4 and CQUADR individually, the
results are the same. But it seems the CQUADR doesn't support the laminate
element, though the displacement I got is the same with the CQUADR plate
element, no data available for the laminate layers.
【在 p**e 的大作中提到】
: classic laminate theory does not support in of plane torsion. therefore the
: 6th degree of freedom has to be zero. I don't know exactly how nastran is
: implemented. but that might be the reason. in abaqus such problem does
: not exist. in both abaqus and nastran, laminate is just add add-on to the
: classic shell theory. no special treatment necessry just have to integrate
: across the layers.
: also in nastran manual, it seems only the cquad4 supports composite.
: abaqus is much stronger in composite than
| p**e
发帖数: 126 | 6 Eif you use anisotropic material, might see the difference. for isotropic
material layers do not make any difference as the result would be the same
if you integrate over three points across the thickness or three points in
each layer. however, if the layers are not isotropic material, then the diff
might surface. i suspect the 6 dof element did not implement the laminate
theory. Natran is a nasty old car, patched here and there so it can tumble
ahead, eventually, it is going to break down unles
【在 r********o 的大作中提到】
: Thanks for your reply, it's very helpful. I did run several additional simple
: models with in-plane torsion of plate element and laminate element with same
: isotropic material on each layer with CQUAD4 and CQUADR individually, the
: results are the same. But it seems the CQUADR doesn't support the laminate
: element, though the displacement I got is the same with the CQUADR plate
: element, no data available for the laminate layers.
|
|
