qausi-static VS. transient
any difference between these methods?I was told that the transient method compute the both displacement and velocity but the qausi-static doesn't.
Any other gives?
how about the difference in solver? 本帖最后由 tonnyw 于 2010-7-9 09:55 编辑
I don't know much about quasi-static. My conjecture is that quasi-static problem is basically elliptic problem while transient problem is hyperbolic problem.
For quasi-static problem, we just need spatial discretization while for transient we need temporal and spatial discretizations. We may use Wilson-theta, Newmark, discontinuous Galerkin method, and so on to do the temporal discretization. We use finite element to handle the spatial part.
But I guess you know all that.
1# caoer tonnyw provides a good description about the difference of those two.
Because transient ones need to take care of temporal discretization, it is more difficult in programming and harder to convergent when using real implicit scheme. In practice,few softwares use static explicit FEM scheme because it is time-consuming due to its need of much small step increment. On the other hand, few software adopts implicit dynamic sheme. When using explicit dynamic approach, FEM code maybe completely different with the static implicit ones because quite different algorithm may be adopted. For the dynamic problem, basically it has something do to with what types of time integration scheme one should use: explicit or implicit (forward integration scheme or backward integration scheme). The advantage part of forward integration scheme is that no need to do the factorization of the stiffness matrix. However it has requirement on the size of the time step size which I think can be determined by solving an eigenvalue problem. It is conditional stable scheme.
Backward integration scheme is unconditoinal stable scheme which involves the factorization of the stiffness matrix. But I guess with the development of new iterative solver. Backward integration scheme might be appealing now. I am not an expert on this. This is just my guess. 本帖最后由 caoer 于 2010-7-9 14:20 编辑
Thanks you guys nice answers.
This question comes from one application in COMSOL, when I looked at the shell element
section of COMSOL, there are several options including static, eigen, transient, quasi-static,
....etc, then I found both transient and quasi-static are able to the solve the time-dependent
problem. So this question above popped up in my mind.
In addition, I was thinking why COMSOL develop a quasi-static solver, does
it have any application?is it exclusively belong to the shell element ?
many question follows that.....
So do you guys believe this difference lies in the explicit or implicit scheme?
I was thinking the the transient method consider the velocity and algorithem takes
the current velocity to the next step computation. Otherwhile the Qausi-static doesn't
take the current velocity in the next time step consideration. However I am unsure about
this.
As far as I know, the explicit method emerged early rather than the implicit one, because
it is easy to be coded. while the drawback of it is about the time step should be small enough
to get precise solution. The implicit method has a couple of advantages over the explicit, such
as correspondingly large time step, saving memory usage, but hard to be paralleled. nowadays
developers prefer to the implicit method is for the memory saving I think, because the explicit
solver require much more data space, when the project comes to be giant with millioms of
elements, some data have to be stored in the swap section or the hard driver, which may
terribly increase the cost of computation time because the operation on the hard disk.
one thing about the unconditional stable, I remember the Crank-nickson and Newmarks for
1st-order and 2n-order respectively are unconditional stable, and have 2nd order accruarcy.
but don't know the backward does for sure. From your above answer, I think there are some misunderstanding bwteen us.
1. What you describe about quasi-static maybe something about viscoelasticity and viscoplasticity. They do use real time and consider velocity. About transient analysis, it takes consider about acceleration, not just velocity.
2. Take care about the difference about implicit and explicit linear equation solver and implicit and explicit FEM scheme. I does think there exists much difference in memory usage and parrallel compution bewteen implicit and explicit FEM scheme, although explicit or direct solver uses much more memory and hard to parrallelize. 本帖最后由 caoer 于 2010-7-10 07:30 编辑
What you describe about quasi-static maybe something about viscoelasticity and viscoplasticity
proberly....Thanks
Take care about the difference about implicit and explicit linear equation solver and implicit and explicit FEM scheme.
hillyuan 发表于 2010-7-9 19:03 http://forum.simwe.com/images/common/back.gif
what's the difference I would know it, any textbook or paper about it?
I appreciate that if you notice us excat difference.
页:
[1]