zf_geongineer 发表于 2006-12-6 03:37:46

Stifness matric问题!

计算中老是出现如下问题：Stiffness matrix not positive definite,boundary conditions or model collapsed
请高手指点！！谢谢！！

tonnyw 发表于 2006-12-7 02:13:44

刚度矩阵非正定，检查边界条件是否正确。＝＝＝＝＝＝＝＝＝

zf_geongineer 发表于 2006-12-7 16:54:26

我检查过了,边界应该没什么问题,而且我用同样的边界条件做了一个验算的例子是可以做的,所以还是请高手多多指教呀!!

混天涯 发表于 2006-12-7 18:44:35

检查一下材料定义。看看是不是弹性模量或者其他什么参数有问题.

zf_geongineer 发表于 2006-12-21 23:08:45

楼上的gg们，以上问题都检查了，我把材料全部换成弹性的也不行，高手们，多多指点亚！！

tonnyw 发表于 2006-12-21 23:30:57

save your model in *.IN file and send it to me at featalk@gmail.com. I will take a look.

loveala 发表于 2006-12-23 17:07:25

你把问题描述一下也好啊，是不是在定义材料的属性的时候你没有定义材料的密度（density）啊？这样也会在一些问题中出现这种提示的。也可能是你加了不够合理的约束方程。

[ 本帖最后由 loveala 于 2006-12-23 18:27 编辑 ]

zf_geongineer 发表于 2007-1-3 00:06:15

下面是我的命令流，欢迎各位大侠指点！！:handshake

tonnyw 发表于 2007-1-3 04:18:37

Coud you upload the *.idb file too?

zf_geongineer 发表于 2007-1-3 14:14:15

有.in文件就可以了呀！为什么还要那个文件？？？
那么你看我的.in文件了吗？

tonnyw 发表于 2007-1-4 04:15:46

I did see you *.in file. But there is no elements, no element type. I need to know what type of elements you are using.

mushengyuan 发表于 2007-1-4 11:14:40

Continue Even when Non-Positive Definite Stiffness Matrix ...

If this option is not selected (default),

In linear analysis:
ADINA stops if the stiffness matrix is not positive definite, except when potential-based fluid elements are in use.
In nonlinear analysis:
ADINA stops if the stiffness matrix is not positive definite, unless:
the automatic load-displacement (LDC) option is being used, or
the automatic time-stepping (ATS) option is being used, or
the element birth/death option is used, or
potential-based fluid elements are being used, or
a contact analysis is being performed.
If this option is selected (i.e. IPOSIT=CONTINUE), ADINA will always continue execution. If an exact zero pivot is encountered, ADINA assigns a very large number to the diagonal term, effectively attaching a very stiff spring to the degree of freedom.

If the stiffness matrix is not positive definite in linear analysis, this usually means that the problem is not well defined (e.g. insufficient restraint). You should first check for modeling errors. Selecting this option to force a solution in such a case can give misleading results.

tonnyw 发表于 2007-1-5 13:36:20

For truss element, one node just has one degree of freedom. But for two-dimensional solid element, one node you have two degrees of freedom. You need couple the two types of element. That's what your problem is.

zf_geongineer 发表于 2007-1-7 10:21:13

To tonnyw:
How can I couple the element ?Could you give some details ? Thank u !

zf_geongineer 发表于 2007-1-7 10:22:55

Give my gratitude to all of you !

zf_geongineer 发表于 2007-1-7 10:34:25

To mushengyuan :
Could you give some adivice about the caculation results ? If I choose the opition you said theresults may bemisleading or wrong !
And how can I judge it ?

loveala 发表于 2007-1-7 21:43:27

大家都来练习外语阿？
几个问题想问你，为什么你的模型在土与混凝土之间没有接触单元，但却用独立节点（土体之间也有这个现象）？
这样的话应该是不会正定的，节点之间是不传力的，怎么能正定阿。
你的锚杆和护壁都是没有定义生死关系的是不是在研究不正定的时候给关掉的？
把你的模型改了一下，在使用mor-col不加护壁的时候可以计算。

zf_geongineer 发表于 2007-1-8 18:21:58

To loveala :
能不能麻烦哥哥你把我的命令流改过后发到我的信箱里？？
zf_geongineer@126.com
谢谢！！
也可以与我QQ联系：727737585

hejt2004 发表于 2007-1-8 21:23:42

问一下
网格化分不好会不会导致Stiffness matrix not positive definite,boundary conditions or model collapsed？

zf_geongineer 发表于 2007-1-9 10:21:18

是有那种可能,不过好象一般的这种问题是由于材料属性、划分网格时设置等问题造成的，我检查过！！
页: [1] 2
查看完整版本: Stifness matric问题!