[CFD] Pseudo-Transients for Steady State CFD (Part 3) - Fluid Timescale

[@nicotrivium]

Dude, amazing explanations... I really love them. I couldn't find a better one so far, related to TSF calculations. Unfortunately, channels like this are so underrated... I am really glad I found it, thanks a lot man!!

[@Michallote]

As always a pleasure to listen to your lectures Dr Aidan

[@edgarfernandoreytorra7106]

didn't expect that neat piece of information about characteristic lenght and velocity, thank you

[@carlosmoreno4682]

Right now my PC dying in a 76h simulation VOF (3 phases) in Pseudo-transient regime, just to add to the importance of this nugget of knowledge to me. Thank you Aidan!

[@fluidmechanics101]

76h ..... That's a long one!

[@saravananvenkatesh5717]

The lecture was very informative and clear. Than you Dr. Aidan.

[@sergniko]

Very useful talk! I think a better understanding is reached :)

[@MrLote83]

Perfect! as always. Thank you Dr. Aidan for such very very informative video.

[@Sam12347398]

Thanks for those presentation, great and clear so far 👍

[@anirbandas4201]

Great Clarification. Thnks for such informative content.

[@jessbuildstech]

Perfect, part 3 is just what I needed! Thank you Dr Aidan (:

[@julioalvespereirafilho6564]

J usar

Nice series! Thank you very much.

[@guidofranceschini9931]

Love you Aidan!

[@ganeshyng5403]

Thank you so much

[@forestshaner8994]

What a wonderful lecture series. Once again, well done! I do have one ambiguity that I'm trying to work through: if we're considering compressible flow, and our Mach number is greater than 1, shouldn't our characteristic speed be either the maximum velocity in the cells or the maximum velocity at the boundaries, rather than the speed of sound? For example, for a Mach number of 2, shouldn't the local velocity be twice the speed of sound? If we're aiming to be conservative by taking the greatest representative speed, shouldn't we use the local velocity (U_cell) in this situation?

[@thirumech6970]

Thank you very much for all your crisp and informative series Dr. Aidan!!! Will it be possible for a content covering segregated and coupled solvers? Thank you!!

[@fluidmechanics101]

Yes! Coupled solvers is going to be a tricky one and is definitely on my list. I want to do all of the big topics in CFD eventually 🙂

[@thirumech6970]

@@fluidmechanics101 Thank you! 🙂

[@angirekulavenu4028]

Hi Aidan, thanks for the wonderful lecture. 🙏 Does OpenFOAM support Pseudo transient simulations?

[@fluidmechanics101]

I think it does in the more recent versions of PimpleFOAM

[@حسنخرمي-ف2ز]

Thank you very much for your all lectures. They are really helpful. Can you elaborate on how f is chosen (fraction) of time? Another thing, can you make another video on courant number. how does it differ from steady to transient, and from pressure based to density based. Thanks again for your helpful videos

[@fluidmechanics101]

f is normally set to 1 and is then left as a variable for the user to adjust. If the calculation is unstable then the user can reduce f, to reduce the pseudo time step and get a more stable calculation. Courant number should be the same, regardless of your type of solver or solution algorithm 👍

[@saisreechandras9295]

Amazing video explanation. Thank you Dr. Aidan. I have a question. You have explained pseudo transients based on solving for temperature profile in the domain. I am currently trying to solve a fluid problem (external flow) to visualize the velocity profiles. Does only the convection and compressibility effects hold for my case in calculating U0 or do I need to use a different approach?

[@daronisaac3226]

Great content. Thanks for creating this great video series. Point of clarification about characteristic velocity for compressible flows. I'm not understanding your statement at 19:23. The speed of sound depends on temperature, a=√(gamma R T), and not on Mach number. So it seems to to me that the characteristic velocity based on the speed of sound would always dominate when computing characteristic velocity for compressible and other low speed flows. What am I misunderstanding? Thanks

[@Jialei-dw3li]

Thank you for the fantastic lectures. I am a little confused: is the coupled solver in Fluent or StarCCM+ use the same algorithm? Looking for your explanation. Thank you very much.

[@fluidmechanics101]

As these are proprietary codes, we can never really know how their algorithms work exactly. However, they should be using these ideas in their algorithm

[@Jialei-dw3li]

@@fluidmechanics101 I see that in coupled solver the Courant number should be set rather than URF. In this case, what's the relationship between Courant number and the pseudo-time step and why Courant number can be so large (e.g. 100 or 500)? Thank you.

[@fluidmechanics101]

Ah ok, yes they are using local Courant number instead of pseudo time step as they are related to eachother. As you are aiming for a steady state solution you can go with Co > 1, and should probably use the values that they recommend. If the solver is a bit unstable then try reducing Co a bit until you are happy

[@minjoong6687]

Thank you sir! I just have one question. The density and viscosity are a property of a certain fluid and should be constant for given initial & boundary condition. However, the maximum fluid velocity along the cell should be 'calculated' prior to determining the pseudo timestep. Which needed to be determined 'before' calculation. I wonder how maximum fluid velocity is calculated when we don't even started the calculation. Once again, thanks for your excellent lecture. I appreciate it alot.

[@fluidmechanics101]

It will be based on the initial condition 👍

[@minjoong6687]

@@fluidmechanics101 Then, I think the initialization should be done by hybrid of FMG intitialization cuz the standard initialization often assume that velocity or other conditions are all the same for given case. Am I right?

[@fluidmechanics101]

Yep, you've got the right idea. Normally this is something to experiment with and see what happens

[@ShakeelAhmad-zh6ol]

Dear Aidan, how we can implement pseudo-transient solution for multiphaseInterFoam solver in OpenFOAM?

[@oskarelmgren]

Pseudo-transient should then on average be able to reach a converged solution faster, and be more stable than true steady state solving with a fixed relaxation factor? Even in the case of a single domain fluid?

[@fluidmechanics101]

That's quite a big claim! I think the best we can say is that they are different methods for achieving a steady state solution. Different CFD codes may be optimised for different approaches, so I don't think we can really know for sure. My hope with the videos is that you can at least see how they are related ☺️

[@Maxakch]

Thanks Aidan! You make great videos, I love it. It helps me look at certain aspects of CFDs more broadly. But you have error in this video at slide kzbin.info/www/bejne/bneZgIl3iqiWjqs . The error is related to the speed of sound. Formula 16 doesn't work for a flow with Mach number (M) low than 1, it gives as reference speed equal to speed of sound. And that is wrong. Also, by defenition M=U/c, so if M > 1 than c < U.