Steady state and transient solution.

  • 31 Views
  • Last Post 11 March 2019
  • Topic Is Solved
ari003 posted this 09 March 2019

What is the difference between steady state and transient solution?
Based on the application how can I figure out that which solution to use?
It will be really helpful if someone helps.

Order By: Standard | Newest | Votes
abenhadj posted this 10 March 2019

All flows are of unsteady nature and do have time history and even some if them memory. This us a specially the case where flow properties material properties or BC are time dependent. Some flows however do result in a time invariant state independent of initial stage.

So whenever the final state depends on the initial state then you will probably require the transient solver. This is even required for flows which are stated as steady but the path to that state might require some temporal resolution.

You question is basic stuff that is why please read some text books and theory.

Here a good link for you: http://www.ecourses.ou.edu/cgi-bin/ebook.cgi?topic=fl&chapsec=03.2&page=theory

Best regards,

Amine

ari003 posted this 11 March 2019

Thanks a lot sir for your reply but what I tried to convey through my question is that how this transient and steady state solution is related to the convergence of the solution in fluent?

I had performed one transient simulation of an airfoil where after a certain time-step the solution converged and it doesn't change velocity(for example) with time, so in this case what can I predict from this solution?

And last thing, suppose I have an airfoil at whatever AOA which kind of simulation is perfect is it transient or staedy?

abenhadj posted this 11 March 2019

Transient needs to converge every time step as if you have N steady state runs where N the number of time steps. The first transient could be neglected. Transient run needs to cover the required physical time or periods.

You can just deduce that it ended with quasi steady state solution.

Best regards,

Amine

ari003 posted this 11 March 2019

Actually converge means that the difference between two successive residual has gone below the tolerance value right?

If it is yes then in steady solution the convergence is a must while in transient it shouldnt be converging because it is unsteady and changes with time-step but in the above comment you mentioned that transient converge every time-step. This is where I am getting confused.

 

abenhadj posted this 11 March 2019

Residual convergence plus reports are not changing anymore towards the end of the transient. For transient the convergence has to be achieved per time step

 

Best regards,

Amine

ari003 posted this 11 March 2019

Residual convergence plus reports are not changing anymore towards the end of the transient. For transient the convergence has to be achieved per time step

 Please can you elaborate the first sentence?

 

abenhadj posted this 11 March 2019

You create some reports for some important and key parameter in your simulation like pressure drop, average of temperature at certain boundary or any other integrated quantity. The transient evolution of that quantity when plotted versus the flow iteration (and not flow time or time step) should depict a constant (at least near) constant behavior towards the end of time step.

 

Best regards,

Amine

ari003 posted this 11 March 2019

What I ve understood are as:-

1. In steady state simulation if the solution converge within the number of iteration then it is appropriate and doesnt have any transient behaviour.

2. If the simulation doesn't converge within the number of iteration then it will have transient behaviour for sure. If so then what is the best number of iteration?

3. In transient simulation there remains inner iteration within each time-step and that is supposed to converge in every time-step. Then what happen in pseudo          steady condition where the simulation initially shows transient and then changes to steady?

abenhadj posted this 11 March 2019

1+2/Is not must perhaps you have wrong boundary conditions and that is why the steady state run does not converge: do not generalize!

3/Ignore the first transients where the flows tries to find itself. For statistic averaging one would ignore at least one flow throughout time. (L/U) before one starts getting mean  values and other statistics.

 

Best to do some tutorials when it comes to this core basics!

Best regards,

Amine

Close