stability&convergnece
 

Hi what is stability and convergenec? I am trying solve an unsteady problem.please help me I need your help. my time step is 0.002 and number of itreation at every time step is 35. i solve it whit segregated solver . flow and energy equations should be solved. I receaved this message after 20 or 26,...itrations at every time step.(solution is converge!) is this message enough? .i mean that realy solution is converge and results is reliable. is there another criterion for cheking correctnec of results? please explaine for me about convergnece and stability criterion in FLUENT.please help me. BEST REGARDS, Amir

Re: stability&convergnece
 

Hi Amir. Are you new to Fluent and CFD? If so, I highly recommend that you read through the manual, or at least use the manual in conjuction with tutorial files.

Convergence is seen with the gradual decrease in residuals. If your residuals decrease, and your solution appears to stop changing (you can perhaps monitor a surface or force and see when they level off). If the residuals are increasing and the solution is not coming to a final set of values, it can be considered diverging. If the solution is converging with consistency over many iterations, it can be called stable.

The default criteria for convergence is set at 1e-3 for all variables (except energy which is at 1e-6). For most of my work, this is not usually enough. Go to the residuals panel and change the covergence criteria if needed. Again, I suggest monitorning a point of force (you can choose a value relevant to your flow, for example: if you are studying an airplane wing, Cl may work)

I won't be able to go into more detail, but the manual should answer most of your questions.

Re: stability&convergnece
 

Hi Chetan, tank you very much.

you are right . how know that my results is correct? is this message enough : solution is converge! or not i should exam another criteria for convergence Regards, Amir

Re: stability&convergnece
 

The short answer is "You never know whether your solution is correct or not". But you can make sure that solution of algebraic equations that fluent solves converged to the solution of the partial differential equations that approximate your physical model. Here is my the check list:

1. Make sure that your residuals level off, and this level should be below 1e-3 for all variables and less than 1e-6 for energy. Small oscillations al very low levels are acceptable (they are usually due to round off and truncation errors).

2. Monitor the most important variables for your model like flow rates, heat transfer rates, concentration, pressures, drag, lift etc. and make sure they level off as well.

3. check you mesh using obtained solution a. y+ on the wall boundaries (is the mesh fine enough to resolve BL?) b. does your mesh resolve major vortexes and flow features in the volume?

4. Check that the solution is consistent with your model assumptions by critically evaluating the absolute values of velocity, temperature and pressure..etc.. Let say if your temperature varies a lot you need to specify how the fluid properties changes with temperature. if velocity is high maybe you have to switch from laminar to turbulent and vice versa. etc...

Re: stability&convergnece
 

Alex's suggestions of checking convergence by monitoring the values you care about are really necessary. Based on the message you give, I think you need to use a smaller value for convergence criteria so that the calculation can continue.