Learning Forum

[harshabharadwaj1]

Hello,nI had a question. I want to compare two simulations for the same flow time (4s). I use the Total Time option in the Duration Specification Method box during Calculation.nBut one of the simulations converges in the specified time (4s) whereas the other does not. Does it make sense to compare the two even if one of the simulations has not reached steady state?nIf not what do you suggest I do?nThanks for the help in advance.n

[Rob]

It depends. What are you trying to compare? n

[harshabharadwaj1]

Hello,nThanks for the reply. I am trying to compare the shape and size of the flow at the outlet for different flow velocities and gap distance. nI created a plane near the outlet to get the shape as seen in the image (OutletPlane2.PNG) . This shape changes depending on the inlet velocity and the gap distance (Gap-0.4MM)nn

[Karthik Remella]

Are you running a transient simulation and are you attempting to compare the velocities at t = 4 s? nIs your end goal to compare the two steady state solutions?nThanks.nKarthiknn

[harshabharadwaj1]

Hello Karthik,nI am a bit confused with this myself. I am running a transient simulation and I want to compare the different simulations. So from what I understood it made sense to compare the simulations for the same time frame (4s). nBut some of the simulations does not reach convergence within 4 seconds. It takes longer. Does it make sense to compare the converged solutions for different times or does it make sense to compare the flow for same time (4s) with unconverged solutions?n

[Karthik Remella]

I'd suggest that you let all your simulation reach a steady state and then compare results. This way you are sure to compare apples and apples.nThanksnKarthikn

[harshabharadwaj1]

Okay thanks for the help Karthik. But the different flow times will not matter is it?. Under what circumstances are different flow times comparable?n

[Karthik Remella]

This will not matter if you let your simulations achieve a steady state. After this, your results will be time independent.nThanks.nKarthikn

[harshabharadwaj1]

Okay thanks for the help. I just had one more question. What does it mean when a solution converges at an iteration, but it does not in the next consequent iterations?nSome of the solutions converge intermittently. n

[Karthik Remella]

What do you mean? Are you talking about a transient simulation run?nThanks.nKarthikn

[harshabharadwaj1]

Yes, I am talking about a transient simulation. nAlso when do I know if a solution has reached steady state?. If the residuals converge does it mean the solution has reached steady state?nIs there a way I can verify whether steady state has been reached or not?n

[Karthik Remella]

Please create monitors such as 'Average velocity at the outlet boundaries' or 'Average Temperature at outlet' or Average inlet pressure'. Please plot these against time. When you have a steady solution during your transient run, you should see that these values becomes more or less constant (in an ideal scenario). In most cases, you might be able to achieve a periodic oscillating solution. You should be able to take this solution as approximately (time-averaged) steady state.nI hope this helps.nThanks.nKarthikn

[harshabharadwaj1]

Thanks for the help. What if it is already a steady state simulation. Do I still have to check by creating monitors OR as the name 'steady' itself indicates, once the residuals are converged should I just assume it to be steady?nAnd also a quick clarification. My steady and transient simulations do not match. So I assume that the transient simulations will eventually match my steady results at some point ( 100, 1000 seconds..etc). Is my understanding correct?nSince transient simulations take a long time, is it better off doing a steady simulation for my case?nI really appreciate the help that you have provided me over the month.nThank you.n

[Karthik Remella]

You should always create monitors and make sure that your converged solution is steady.nYes, most transient solutions finally reach a steady state solution. However, there might be cases where you might only get an oscillatory solution.nSteady state solutions are must faster than transient. If you are confident about your modeling and if the physics you are solving has a steady solution, you could work by solving a steady model.nThanks.nKarthikn