CFD Online Discussion Forums

CFD Online Discussion Forums (http://www.cfd-online.com/Forums/)
-   FLUENT (http://www.cfd-online.com/Forums/fluent/)
-   -   steady/unsteady (http://www.cfd-online.com/Forums/fluent/32133-steady-unsteady.html)

zwdi September 14, 2003 12:05

steady/unsteady
 
Hi everyone, I am not clear about the criteria for steady solution. I mean when I use unsteady solver, how do I know that the state has arrived at steady state, from unsteady state? How do I know for some cases, there is no steady state solution? Thanks a lot. Zwdi

martin September 15, 2003 04:35

Re: steady/unsteady
 
hi,

you can create a point/line/surface. then you can activate a monitor in solve/monitor/surface. there you can monitor all values at point/line/surface. the solution is "steady state" when there is no change in the selected value. e.g. for heat transfer problems temperature is a good value.

martin

zwdi September 15, 2003 15:16

Re: steady/unsteady
 
Thanks Martin. I understand what you mean. For my case, the changes of variables between two time steps keep constant(time steps=100, step size=10s),around 3.7% of the value in previous step and the change trend still goes on. Could I say it arrives at steady state or there is no steady state for my case? Thanks again. Zwdi

martin September 16, 2003 01:43

Re: steady/unsteady
 
hi, if the value oscillate, i think it is "steady state". otherwise i would wait more time, to see how the absolute value changes.

zwdi September 17, 2003 12:19

Re: steady/unsteady
 
Thanks again, Martin. I increased the runing time from 100 to 200 time steps(time step size=10s). The vaiable changes between two steps are still keeping constant(3.7%). I checked the flux value of the inlet. Only 0.16% difference between the true value(I set at the velocity inlet boundary) and the calulated value. The runing time now is 2000s. I think it is enough. So, now could I say there is no steady solution for my case? Could you please give me other suggestions? Thanks. Zwdi

thomas September 18, 2003 06:31

Re: steady/unsteady
 
Hi, Let me express an Idea. I would effectively say there is no 'instantaneous' steady state solution. But as Martin said a regular oscillating solution could involve your system has achieved a steady state. Therefore to get rid of these oscillating values use the 'Data Sampling for time statistics' option available with the unsteady solver. A last comment, I do not know what is your case but be aware some case can give unsteady results in 3D and steady results in 2D (i.e. : Bubble column simualtion)

Hope this help, Best regrads Thomas.

zwdi September 18, 2003 20:23

Re: steady/unsteady
 
Thanks, Thomas. I think maybe I am still not clear about the logical relationship between steady solution and unsteady solution. In my case,

1)I compare the unsteady solution and steady solution to see if they are the same(or very close) when time is enough long. Unfortunately, the two difference is large, around 27%. I think they should be the same if solutio is converaged.

2)About the time step size, FLUENT says "A good way to judge the choice of time step size is to observe the number of iterations Fluent need to converage at each time step. The idea number of iterations per time step is 10-20.If Fluent needs substantially more, the time step size is too large". Could you please explain to me why this criteria can judge time step size? In my case, I set the time step size=10s, I need 2000 iterations (>>20 iterations)to converage (residual:1e-04). Do you think 10s is not proper? Could you please tell me what is your criteria for setting up time step size?

3)I active the 'Data Sampling for time statistics' option to collect the mean value and root mean square value. Do you think if this mean value got from unsteady calculation should be very close to the value got from steady calculation? For a example, in multiphase flow, the variable is volume fraction of sencond phase. I mean from unsteady calculation, the mean volume fraction for phase-2 should very close to the volume fraction value got from steady calculation. But I checked my case, there is much differences between these two values. Could you please point where I am wrong?

Thank you very much. Best regards,Zwdi

thomas September 19, 2003 03:48

Re: steady/unsteady
 
Hey man, I think I can clarify some of your question.

1 - I would be really interesting to know where you find this method to set your time step size. Trust me this is not the method that Fluent recommand. The most general method is based on courant number. In fact to choose the time step size you need to make sure the time step is not too big in order to catch/calculate enought variation (of a particle for example) during this time step. To give you an idea you should check what is the maximum velocity in your domain (axial or radial), and also check the cell size in the velocity direction (axial or radial). There you are gonna have a time characteristic for your flow adapted to your mesh. This is approximatively what you should set as a time step size. Of course some tests could tell you if you need a smaller time step or if you can choose a bigger one (Less cpu time).

2 - Concerning the adequacy between unsteady results and steady results I would say it depend. As I do not know your case let me take an example. Take a bubble column with an initial volume of water. At t=0s we inject air at the column bottom. There the water height will increase till stabilization. In that case steady state simulation cannot give good results when obsviously unsteady does. This is due to the fact the final solution depends on the initial condition, here the initial height of water. You should check in your case if the final solution depends on the initial condition. Think in some case the hystory of the convergence is very important to achieve realistic and good results, your case might follow this rule.

So i think this will clarify some stuff for ya and I hope my english has been clear.

Now just a question for me, Where do you come from in Canada? or Univeristy in Canada?

good luck thomas.


zwdi September 22, 2003 10:34

Re: steady/unsteady
 
Hi Thomas,

According to your guide, I tried different initial VOF values in Unsteady solver. I got two different final results(the difference is about 25%).In this case, I can say that there is no steady solution. Am I right?

Thank you very much for your kind to my ignorance.

Zwdi

thomas September 23, 2003 05:07

Re: steady/unsteady
 
Hey, I am sorry I cannot answer, could you simply explain to me what is your case, geometry, fluid, inlet and outlet conditions. With those informations we could evaluate what is going on with your results and if your system is steady, quasi-steady, or completely unsteady.

thomas


All times are GMT -4. The time now is 00:47.