单元缺陷？

feasky

我采用MITC4单元，但是平面应力部分含有转角，并采用连续介质力学关于旋转度的定义。

这是否有问题？龙驭球院士曾指出这个定义的问题，但是应该不会有如下那么大的误差。

建立一个剪力墙，固定底部，平面内将一个水平集中力作用于顶部，结果与单纯用四结点平面应力单元（无转角）相差比较大，更为恼人的是MITC4无法使结构破坏，当混凝土受压破坏（图中数值不稳定的地方）后居然刚度还能增加，所以平面应力Quad单元可能更能反应结构的实际物理状态。混凝土开裂（图中顶点）前吻合很好，开裂后有一段还不错，接着误差逐渐增大。 说明：我在平面应力和MITC4里的膜部分采用相同的本构关系。应该不会是本构关系的问题，此外，程序代码方面不会出错，已经检查N遍了。我想可能在壳单元出问题了。

怎么改进呢？会不会在平面内也产生类似于平面外剪切闭锁的问题？

tonnyw

MITC4 shell element is very popular which is indeed a very popular element nowadays used by almost all the commercial finite element codes. Its reliability is guaranteed.

The effective way to figure out this problem is that you can put your derivation here.

feasky

2# tonnyw

Thank you for reply.

Yes, it is very popular and reliable for plate bending analysis.

I am just wondering for the membrane part.

hillyuan

MITC4 do nothing on membrane strain. Therefore, I am wondering if you do something wrong in 平面应力部分含有转角(?) I don't understand what you mean. Is it drilling degree of freedom?

feasky

Thank you hillyuan.

The drilling degree is defined as shown in the following figure.


I code the program using the theory devoloped by E. Love, which was implemented into Opensees. His code can be found at
http://opensees.berkeley.edu/WebSVN/filedetails.php?repname=OpenSees&path=%2Ftrunk%2FSRC%2Felement%2Fshell%2FShellMITC4.cpp

feasky

MITC4 do nothing on membrane strain. Therefore, I am wondering if you do something wrong in 平面应力部分含有转角(?) I don't understand what you mean. Is it drilling degree of freedom?
hillyuan 发表于 2011-7-31 09:36

Yes, 转角 means "drilling degree of freedom"

pasuka

我采用MITC4单元，但是平面应力部分含有转角，并采用连续介质力学关于旋转度的定义。

这是否有问题？龙驭球院士曾指出这个定义的问题，但是应该不会有如下那么大的误差。

建立一个剪力墙，固定底部，平面内将一 ...
feasky 发表于 2011-7-31 02:30

膜单元的转角自由度，好像有3种定义方式，似乎没有哪一种方法是通吃所有情况的，不妨按照其它方式定义转角自由度，对比一下结果

feasky

回复 7# pasuka

谢谢，龙老师那个定义好像合理一些。

hillyuan

It seems that your test problem is a plane stress probelm. In this case, your implementation is definitely wrong. I think you can confirm it letting all drilling degree of freedom to zero. It you get the same result to that of plane stress analysis, the problem would be your implementation dealing with drilling degree of freedom.

worry

这个也不懂，来学习一下

feasky

It seems that your test problem is a plane stress probelm. In this case, your implementation is defi ...
hillyuan 发表于 2011-8-1 08:28

I agree with you. Something is wrong in the drilling degree of freesom.

However, I don't think that we can solve this issue by setting the rotaion zero.

If the rotation is zero, the translational components of displacment at the point is affected. Please see the definition of the drilling degree of freedom. Actually, the plane stress problem doesn't constrain the rotation. It is totally free.

Yes, I just want to test it plane stress behavior under a inplane load. The out-of-plane response is OK now.

tonnyw

I agree with you. Something is wrong in the drilling degree of freesom.

However, I don't thin ...
feasky 发表于 2011-8-3 00:54

How about making zeroes the columns and rows corresponding to the drilling dofs and adding perturbation to the diagonal terms?

My guess is that there might be a bug somewhere else.

A couple of questions:
1. Did you use shear correction factor?
2. The constitutive law is defined with respect to Cartesian coordinate. The tensorial interpolated strain is defined in the local natural coordinates which are usually not orthogonal if the element is twisted. Did you do the transformation on the constitutive law, i.e. the matrix D?
3. Did you code it yourself or take it from the Opensees?

feasky

How about making zeroes the columns and rows corresponding to the drilling dofs and adding pertu ...
tonnyw 发表于 2011-8-3 04:52

Thank you for your guidance.

1. shear factor is used for out-of-plane shear. I used it.

2.I just consider the inplane case. No twist.
》What I mean is the twist in geometry. The element is not rectangle.

3.I coded the consititution law. The element part is taken from Opensees.
>You didn't answer my question.

feasky

已经OK了：）

问题就处在转角上。