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

 rajni June 9, 2006 11:48

I am running a transient simulation ( transient due to a heat source inside a room). How to decide that a steady state has been attained after seemingly sufficient number of iterations ? What's the significance of "convergence" in such a situation ? I hope it doesn't indicate steady state.

 Swarup June 10, 2006 01:06

Steady state denotes "uniformity in time" at a position (of your choice) while spacial homogeneity may or may not exist.

Convergence is an indication of numerical accuracy of solution. You need numerical accuracy per time step for truly "time accurate" solution.

Some times, you just need a snapshot in time say a duartion of 1 hour (physical) and may not be concrened about steady state. I am not sure but natural convection is such a situation.

Swarup.

 rajni June 10, 2006 13:12

Thanx Swarup. But how can I check in my case that steady state conditions have reached ? I have a room with two windows in which a heat source is kept. Cold air enters the room from one window and leaves from the other. I can obtain profiles alright. But is there no other means of ascertaining that ?

Regards, rajni

 Swarup June 10, 2006 23:52

Try to track air temperature with time at out window. If you have reached a steady state, it should settle to a nominally constant vale. Use monitor panel to track this temperature/its average value. Steady state is a physically realizable case to some extent. Hence, it is most appropriate that you look at variable/s which attain a time invariant value. If you are not interested in time acccrate values, then you may afford to have somewhat large time steps (within the stability limit); so called false transient approach. The solution may converge numerically and then may reach a steady state value eventually. I guess natural convection is inherently transient phenomenon.

Swarup.

 rajni June 11, 2006 05:11

Thanx swarup for your reply ! I won't try your suggestion but I like the way you send detailed replies. What do you do, by the way ? Regards, Rajni

 Rajeev Kumar Singh June 20, 2006 04:03

In any unsteady state calculation the best way to find whether steady state has been reached is by monitoring certian parameters at the point of interest. In your case certainly the temperature fits the bill as Swarup pointed out. Convergence criteria only looks after accuracy of the solution within each time step. So convergence only cannot be the true indicator. So it is better to use monitors also even when you are solving in steady state. Good convergence does not mean correct solution.

 Amr June 28, 2006 15:15