CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   is it necessary to d ofirst steady analysis ? (http://www.cfd-online.com/Forums/fluent/39347-necessary-d-ofirst-steady-analysis.html)

ravi January 21, 2006 12:28

is it necessary to d ofirst steady analysis ?
 
Is it necessary to first do a steady state analysis and from that conserved results as a intial files to do a transient analysis in fluent?

Ahmed January 21, 2006 17:37

Re: is it necessary to d ofirst steady analysis ?
 
Not always

kharicha January 22, 2006 04:25

Re: is it necessary to d ofirst steady analysis ?
 
Strictly speaking you have to perform it, in order to verify that your system do not lead to numerical instabilities.

Usteady results means "we model instabilities", and physically usual instabilities are created within region of high gradients(boundary layers and shear layers). A steady state analysis is necessary so that those area are well defined. Then you can switch to the unsteady solver, perturb or not the solution, the instabilities will devellop correctly...

If you do not verify this, then your transient solution might be the sum of numerical and physical unstabilities...this occurs very often.

Mathematically, your transient solution contains the steady state solution. This is what you can observe on many results where the flow simply oscillates arround the steady solution.

But I agree with Ahmed, in some particular problems it is not necessary. Typically it is the case of unsteady boundary conditons (moving walls) or multiphase flows(dripping droplets with VOF)...

So it depends on your problem, and on the time you have for your project...


ravi January 22, 2006 07:37

Re: is it necessary to d ofirst steady analysis ?
 
Thanks sir.

Luca January 23, 2006 04:15

Re: is it necessary to d ofirst steady analysis ?
 
To me it is senseless to study a problem without getting a steady solution before transient analysis. This is because your system response depends on the (initial) status of your NON-LINEAR system. What you do is then study how your system responds to an excitation signal. Depending of the initial status you'll get a different response. When the transient is over you'll get an another steady-value (depending of what kind of excitation you used). Luca

kharicha January 23, 2006 04:35

Re: is it necessary to d ofirst steady analysis ?
 
to Luca,

You have come to the point!!

You are right if the system is highly non-linear, i.e chaotic, i.e strong dependance on the initial conditions.

And before studying a non-linear regime, we should verify first that our model (grid,equations, discretization of the convectif term..) is able to cath the frequencies occuring in the linear regime. And before that it is necessary to perform a steady state analizis....

Strictly speaking, the transient evolution of the system depends on the perturbations introduced added to the steady solution.

But unfortunatly, I know a lot of people that try to model chaotic system without the necessary care. This is either due to a lake of knowledge or a lake of time....


Luca January 23, 2006 06:49

Re: is it necessary to d ofirst steady analysis ?
 
I agree with you. Luca

Marie-Anne January 27, 2006 10:28

Re: is it necessary to d ofirst steady analysis ?
 
Hi,

I read your comments and have a question arising from this even though I agree mostly with what you say. I medelize the onset of natural convection in a chemney. Therefore, if I run a steady state analysis and apply the result as initial conditions to the transient one, the convection is already happening and I cannot see the onset. How do I get to verify my transient computations with no steady state initiation ?

Thank you,

Marie Anne


All times are GMT -4. The time now is 01:34.