why does none care about meshfree methods?

jing-qian

Methods based on element have been developed more than half century. Do you still believe FEM will keep in use for next fifty hundred years? Do you still want to spend  80% or 90% of your time on generating mesh? At least, I do not believe that. Where is the future direction for next generation of CAD/CAE? Although Meshfree methods may not be prefect, it provides us a hint and clue -  what we can do without element or shape functions. Nurbs might be the other direction. Let's start a discussion and think about the future.

绿色的农民

没看到要表达什么意思

caoer

to lz,
yes, we care, but need some pioneers to talk about it.
I am assuming u r the one of them.

jing-qian

I am definitely not such a big guy. But this can not prevent me to think about such topics. Since FEM has reached the limit somehow, what will happen in next 50 years? We keep working on a square element forever? I doubt it. So far, I just want to start a thread and let more persons to join this discussion.

caoer

good to know that.
you know, it is why I create this section.
everyone like you are more than welcomed.
and we still need a moderator here, think about it dude.

nongda

1.现阶段做毕业论文的时候，倒是参考了不少无网格伽辽金的论文。感觉确实是很有生命力的样子。
感觉meshfree methods对coupling是友好，不用担心不同网格的之间的mapping。
2.另外我觉得Nubrs或者B-spline或者 T-spline应该也很有发展的，如果真的像它们最终的目标一样，做完CAD之后直接拿过来就可以做FEA，不用重新做网格，这样至少设计模型->分析模型之间转换造成的误差可以避免，不过貌似nurbs在CAD中是用来表达物体的表面surface，而FEA需要三维实体模型（除了板壳结构）。虽然通过有些方法可以将nurbs surface转化为nurbs solid，但是是有条件限制的的。
3.另外还有Fictitious Domain类的一些方法。用简单正几何体的单元，如长方体或者立方体，近似的包括整个求解域（扩展了的），这样原不规则曲面边界通常是穿越正几何体的单元，边界条件要特别处理。

这些都只是我做论文时候的了解到的，应该有很多错误，希望批评指正。

有点跑题了，不过大家讨论的都是非classic有限元法，算是属于一个阵营的，呵呵。

jing-qian

That is exactly what I am thinking now!!!   Meshfree and Nubrs are two poles in this view. See, meshfree or particle methods are fully discrete and neglect the limitation of geometry. However, Nubrs is completely geometry based and everything relies on original CAD. FEM is in the middle. It needs geometry but discrete geometry meanwhile.  That's why FEM make things a little bit complicit. In my mind, I believe the breakthrough will be either particle method or Nubrs, even both of them. When the geometry can be described by CAD, Nubrs plays well. If the geometry can not be represented by CAD such as organs and tissue, let particle method handle it. It is not necessary to impose an element structure on  non-geometry object.

nongda

7# jing-qian

仁兄对Nitsche's method 中的penal项的取值是否了解。
可否解释解释

zhengzhou60

请问无单元网格法能运用到电磁场中去吗，现在还没怎么见。

paleoshrimp

nongda 发表于 2011-1-23 15:56
7# jing-qian

仁兄对Nitsche's method 中的penal项的取值是否了解。

There is a brief introduction to this method in Shaofan Li's book, Meshfree Particle Methods. Reference to Wchweitzer(2003) on how to determine your so-called 'penalty term' is also provided.

nongda

paleoshrimp 发表于 2011-9-15 09:59
There is a brief introduction to this method in Shaofan Li's book, Meshfree Particle Methods. Refe ...

感谢楼上的回复，我研究研究去

wang61761980

mesh free  is  a new way to solve such as  geotechnical problem
one day abaqus  will  integrate this method!

hoop247

OK, here I'm doing something in LS-dyna using EFG to simulate the crack propagation. So I think the commercial software company has already cared about meshfree method, but now only Dyna released the EFG.

hoop247

nongda 发表于 2011-1-21 22:29
1.现阶段做毕业论文的时候，倒是参考了不少无网格伽辽金的论文。感觉确实是很有生命力的样子。
感觉meshfre ...

对于你的第二点，个人感觉Nubrs等适合于边界元，张见明教授就有一个现成的UG模块，直接用UG的三维CAD的模型做分析，他自己称作边界面法

muyefu

需要共享这方面的资料啊