[CFD] The SIMPLE Algorithm (to solve incompressible Navier-Stokes)

[@kendalmalone705]

Hey i'm a Master Student from Germany and i just wanted to say that this is by far the best channel on youtube in terms of CFD and everything thats comes with it. Thanks for your work

[@ana.scarabino]

Congratulations for the best CFD channel I have seen. You not only do teach CFD, but also are a teaching model, clear and concise. Ansys Fluent´s black box becomes transparent thanks to your videos. Chapeau! and Thanks!

[@fluidmechanics101]

Thank you so much for your kind words, they really mean a lot to me 🙂

[@souravduttagupta4082]

Sir, your contents and explanations are really useful and easy to understand. I've completed my B.Tech and M.Tech in Aerospace engineering from well known institute, IIT Bombay without having basic understanding of some CFD equations like N-S equations, algorithms and various schemes. The way Professors taught this subject was only helpful to those who were very much intelligent and advanced. Students like me suffered the most. At such a young age the intellectual capabilities and teaching qualities that you have are truly far better than many professors. I must say I've seen most of your lectures and understood the concepts with much better clarity. Thanks again for such quality content at free of cost. Keep sharing your knowledge, Sir👍🙂.

[@pranab1234]

Sourav, don't worry. You are in the same boat as many others. I have done my PhD from a reputed US university and still my fundamentals are not very strong, especially in programming. I do understand the physics very well, but implementing it myself in code was somewhat challenging for me. But one thing I have noticed is that when I started coding in small chunks (simplifying the NS equation and solving for one particular problem) I was able to overcome a lot of hesitancy. Still a long way to go, but making good progress on it. Hope you are able to do the same and much more. Never doubt your capabilities. - Best, Pranab

[@souravduttagupta4082]

@@pranab1234 thanks for responding. Yes, I'm doing little better than earlier by myself. Cool! U have completed PhD.

[@jaronsherwood1501]

You are the freaking man, I have yet to have a question involving CFD that you have not covered, and covered well! I'm undergraduate learning CFD for a research project I just got funding for and your videos have been a monumental benefit to me!

[@haseeblodhi6161]

I want to say that what you do is really really great and deserves a lot of appreciation ! I did my fyp on computational aeroacoustics of jet and faced lot of problems understanding some concepts. Your channel is the best way of helping people interested in this field and I am eager to help.

[@fluidmechanics101]

Thank you so much Haseeb! Im glad your fyp went well and glad that i could help out 😊 well done!

[@kolibogi6951]

never understood this algorithm till i found your video. thank you so much. keep up the great work.

[@MsJuanelogodinez]

Man, your work is amazing! Many thanks!

[@vai_-cn9br]

Thanks Aiden. Keep doing what you're doing brother. Much love

[@realez2806]

As a new learnor,I must say it is hard to understand,but very attractive as well.I'll keep learning.Thank you Aidan!

[@dustin3291]

Your videos are extremely helpful! I'm your CFD fan :D

[@MohammedAli-ic2wt]

Nice explanation. Quite clear to me as i have some previous basic knowledge. But for a beginner would be difficult to get.

[@ahmet8790]

If you can't explain it simply, you don't understand it well enough. thx a lot mate keep up the good work

[@shubhamphysicist]

Amazing way of explaining things

[@killua9369]

Great work as usual, if you can explain the PISO algorithm too that will be great! I am studying from Jasak thesis too which is very good.

[@fluidmechanics101]

Jasaks thesis is excellent 👍i havent quite worked out a good explanation for the piso algorithm, but will definitely be doing this is future, as it is such an important method in CFD!

[@ibragim_on]

Спасибо за видео. Продолжай вебинары)

[@brunokassar]

Great work, Aidan! You've just got a new subscriber. Best regards from Rio de Janeiro, Brazil

[@skankhunt-zw6gg]

Thank you, I always wanted know how SIMPLE works.

[@fluidmechanics101]

Me too! Im so glad i finally made this video. I hope you found it useful :)

[@peterirvin7121]

Damn, I really needed this video in 2017 fall semester lol

[@kadrikocer5021]

Thank you. It is very simple to understand thanks to you.

[@paulstephen7225]

Thank you for the incredibly helpful content on CFD principles! I've gained a lot from your explanations. Could you also cover the COUPLED algorithm in one of your future videos? I'd really appreciate your insights on this topic! Thanks again!

[@fluidmechanics101]

... coming soon 😄

[@mleonganem]

Excelent. These videos are very very helpful. Thanks!

[@refikalpertuncer8512]

Hey Aidan, Great explanation. I just wanna leave few remarks. 1. It would be amazing if you could share an insight on how these alghorithms are morphed with compressible flows. 2. Also a short explanation or demonstration on on how to track residuals would be nice as well. Im sure little detail on forms of momentum and pressure equations and their respecive residuals would be appriciated by many. Keep up the great work!

[@parthsavyasachi9348]

Quite convoluted explanation to simple algorithm probably because of learning coming from openfoam side.

[@FreeThiinker]

Excellent Work, I just may add a hint to clarify that in Order to construct the H Matrix which is also dependent from the unknown U Field, H(U) uses the U Field from the last iteration

[@fluidmechanics101]

Yes!

[@golnooshmorshedi3859]

It was really helpful for me, thank you.

[@arshan721]

Dr You are a great teacher. I would request you to please do a video on Artificial Compressibility Method for solving the NS equation. It would be a huge favor. TIA

[@StudySessionYT]

Awesome video!

[@AJ-et3vf]

Awesome video! Thank you!

[@goodday276]

Very good and helpful course. I wish you make a courses about coding of this algorithm in matlab or openfoam

[@abdullahidada8434]

Hi, Fluid Mechanics 101 Excellent video, I would love it if you did a video on the matrix of coefficients M and how it is computed thanks

[@fluidmechanics101]

Hi Abdullah, yes I will be putting out a video soon for how to compute the matrix of coefficients M. In the meantime, you can always check out the textbook by Versteeg & Malalasekeera (you can find it online here: ftp.demec.ufpr.br/disciplinas/TM702/Versteeg_Malalasekera_2ed.pdf). Chapters 4 and 5 have some simple examples (pages 115-124 are probably the best) for how to compute the matrix of coefficients for a simple diffusion problem. Later when I post my video, I will show you how to do it for all terms in the Navier Stokes equations. Hope this helps!

[@adaikhsan]

hi, I also waiting about how to make m matrix, thank you

[@fluidmechanics101]

If you check out my ‘CFD fundamentals’ course (links in the description) you can find out how to make the M matrix from first principles 👍

[@alex07guerrero]

Fantastic content man!

[@harikply08]

Can you make a video on structured and unstructured grid🧐

[@dewis2967]

hi, its a good video and I have learn to much from this video.

[@fluidmechanics101]

That's great to hear! I'm glad you found it useful

[@himanshubanait9166]

Great Video 💓💓👏👍. Sir, can we get a video on density based solver. The equations are in vector form and includes calculation of flux vector with different methods. It makes bit difficult to understand...your explanation will be very helpful for us. 🙂

[@harikply08]

Very informative....thank you👍😊

[@extradounia3027]

thank you so much, that was very helpfull.... keep going

[@fluidmechanics101]

Thanks! Ive got loads more lectures planned 😄

[@Wow_1991]

Thank you for the video

[@dimi3mi]

Hi Aidan, first off I'd like to say that I think you are one of the best resources for CFD online. I have two questions: 1. How does the non-linear convection term in equation 4 get treated in equation 5? Would it appear semi-implicitly where the volume flux F_fi would be calculated from the last iteration in the M matrix? 2. Does the updated velocity field from equation 23 affect equation 21 of the next iteration (say by updating the M matrix) , or does the new iteration only use the updated pressure gradient from equation 22? Thank you for your answer! Kind regards, Dimitri

[@fluidmechanics101]

Hi Dimitri, in answer to your questions: 1) yes you are correct! The convection term is treated semi-implicitly, with the face flux F_fi calculated from the velocity field in the previous iteration. 2) Yes, the updated velocity field (from equation 23) is also transferred to the next iteration, along with the pressure gradient. The reason for this is the velocity calculated with equation 23 satisfies the continuity equation, so the face flux F_fi (which is treated semi-implicitly) will always be conservative.

[@soroushsepahyar]

GREAT VIDEO!

[@vangelissilver4]

Your videos are very informative and thorougly presented, can you make a video about the "behind the scene stuff" of the ansys coupled solver?

[@fluidmechanics101]

I am going to do the coupled solver soon. Just thought i would cover SIMPLE and PISO first

[@safooh94]

Thank you for your great work. Can you do a video about MAC algorithm too? It would be very useful for me as I'm doing a course on CFD this semester.

[@fouadayad8008]

good work sir

[@Miao_zai]

Very good video thank you!!

[@fluidmechanics101]

Great! Glad you found it useful

[@luorisluo3634]

top helpful! thanks so much, please keep on

[@fluidmechanics101]

Thanks Luoris 😊

[@nofugz8939]

In what cases is SIMPLE algorithm recommended over it's counterparts such as SIMPLEC, COUPLED, PISO etc.?

[@fluidmechanics101]

SIMPLE is usually a good first attempt for steady state solutions, because it is faster and has lower memory requirements than Coupled. If you are struggling with mass conservation then switching to Coupled is slower but usually gives slightly better convergence. PISO is best for transient cases. SIMPLEC can sometimes give better performance than SIMPLE (but the differences are usually small). These are just my first hand experiences. Performance may vary considerably depending on the case, so best to try them out for yourself!

[@nofugz8939]

@@fluidmechanics101 Thanks for the reply Aidan. Love your videos, you are a legend! The CFD professor we didn't deserve but needed.

[@fluidmechanics101]

😊

[@sherifahdy908]

Thank you for the great content and your effort. Would you please make a video about numerical schemes, its properties and how to check the accuracy, stability and consistency? My regards

[@fluidmechanics101]

Hi Sherif. Yes, I am planning on doing some videos for numerical schemes. In the meantime, you can check out the thesis by Prof. Hrvoje Jasak from Imperial College London. There is lots of useful information in there (just do a google search)

[@sherifahdy908]

@@fluidmechanics101 Thank you.

[@신승철-e6r]

Thanks for such an amazing lecture. I learned a lot about CFD and the fluid mechanics itself. I just have one question to ask you. you said convection term (u dot gradient) u is equal to gradient dot (uu) by product rule but isn't (u dot gradient) u equal to 2 times gradient dot (uu)?

[@fluidmechanics101]

Yes, my lecture isn't quite correct. You actually have to use the product rule on nabla dot uu and then combine this with the transient term (du/dt). Subtracting the continuity equation then gives you the correct result

[@biaohuang7957]

Nice video. And very nice Beamer style as well.

[@fluidmechanics101]

❤️ beamer

[@wangjia2451]

Great lecture! ! Just feel a bit confused about Eq.23. Doesn't it come from Eq.21 which is momentum equation? Why do we say the velocity field is corrected to satisfy continuity instead of momentum equation?

[@abhinavaggarwal5488]

Sir thanks a lot.. it was very helpful

[@fluidmechanics101]

Thanks Abhinav, glad you found it useful!

[@jamattos9957]

Congrats buddy, your videos are really amazing!! But I got a question for you: After discretizing the Navier-stokes we end up in a non-linear set of equations, I suppose that there is a method to transform them into linear equations (and them resolving using a SIMPLEX alg. for example), or they do use a linear solver in a non-linear set of equations? Does it relate to the SIMPLE Algorithm? I´ve been modeling a torque converter using CFD, and it has been awesome to see a theoretical approach (math and physics) working in a real-world application, I think CFD is a very promising field!

[@fluidmechanics101]

Thanks 😄 so all CFD solvers go for a linear solution by ‘lagging’ the convection term (the U squared term). One of the velocities is taken from the previous iteration, so you can effectively linearise the equation. The SIMPLE algorithm (or the PISO) algorithm is then used to deal with the pressure - velocity coupling, which is a separate issue to the non-linearity of the convection term. Oh yes, and all non linear source terms are lagged and evaluated using the values from the previous iteration as well. Hope this helps!

[@jamattos9957]

@@fluidmechanics101 thanks a lot !

[@parthsavyasachi9348]

@@fluidmechanics101 I think polyflow does not really do it. It applies non linear solver on navier Stokes. So there might be some exceptions.

[@gangadharvenkataramanap7651]

Thank you so much

[@lihongxie9402]

Excellent !

[@achyuth_rj]

Thanks Aidan for an extremely informative and succinct video. A few clarifications if I may: 1. After step 1 in the solution process, do we use the new velocity field to compute H and A again before step 2? 2. After step 3, once we realize U does not satisfy momentum equation, do we compute H and A again before solving for pressure in Step 2? Thank you.

[@AstroEditss]

Did you ever get an answer to this? I'm assuming that one would correct the H(U) every time a new velocity is computed for example: 1) Compute H(U) from previous value (or initial guess U) 2) Solve momentum equation U*=A^(-1)H(U)-A^(-1)P (Where P is from previous iteration) 3) Update H, (H(U*)) = AU*-MU* 4) Update Pressure field to find P* using H(U*) 5) Correct velocity field, U**= A^(-1)H(U*)-A^(-1)(delP*) 6) Step 1 using now U** Does this make any sense?

[@azimarshad]

From what I understand, you should not update the H and A with the new velocity. The velocity and pressure coefficients should be calculated using the velocity values from the previous loop. The values are updated at the beginning of the new loop. This is based on the derivation from Patankar.

[@Ħamir]

Hi! Thank you very much for these videos. When will you explain "couple" algorithm?

[@fluidmechanics101]

Maybe after finishing the LES series? It is definitely on my 'to do' list 👍

[@חמוסיםמתוקים]

If I have body forces (like gravity) the discertization will be MU=-gradP+g ? And If I have a situation that the density and dynamic viscosity change as a function of temperature, for example in a transient natural convection model, the continuity and N.S will be extended because the term drho/dt. The solution algorithm will be the same?. The SIMPLE algorithm can handle with that? Thank you 😊

[@jamattos9957]

What about the residuals Aidan? The solver guess values for the flow variables and uses them on the transport and continuity equations. The residuals that we usually see in CFD software are the difference between those guessed values in the transport and continuity equation? In a way that for every cell in the mesh, there is a residual? so CFD software normalizes them using a specific method?

[@fluidmechanics101]

Yes, the method used for normalisation is different for every CFD code, which makes comparisons quite difficult. You should be able to find details of the normalisation used in the user guides / source code 👍

[@ahmedhafez8522]

I think if we have initial velocity is given we must start process with poisson eqn to calc pressure then we used momentum eqn to calc velocity then we correct it by correction eqn

[@fluidmechanics101]

Usually we have an initial guess for velocity and pressure (and all other fields), so it doesnt really matter what we start with. If you use ‘standard initialisation’ in ANSYS Fluent or look in the 0/ folder in OpenFOAM, that is where all the initial fields are 👍

[@김뇸뇸-o4p]

Before asking it is good video and thank you. Can you explain how simple algorithm can work in unsteady state? Or SIMPLE algorithm can not apply to unsteady flow case?

[@fluidmechanics101]

Yep, just apply SIMPLE again every time step. Normally we limit the number of iterations to around 10-15 each time step or it would get really slow!

[@김뇸뇸-o4p]

​@@fluidmechanics101 Thank you for the answer. I try to apply SIMPLE to the numerical model but it seems like it cannot solve unsteady term. SIMPLE algorithm iterates the U and P so that both momentum and continuity equations are satisfied. But in unsteady case, both equations are satisfied by compensation term = unsteady term. What I'm saying is.. if we solve the momentum equation and it does not satisfy the continuity with the current velocity, the imbalanced part of continuity becomes the change of density in time. I got stuck here. haha

[@nomanyousuf4290]

Hi @Fluid Mechanics101 I have a question, suppose the time level you explained is nth level. when we go to n+1 time level, does the momentum equations is solved for velocity is based on pressure gradient evaluated from poisson equation in n th time level?

[@fluidmechanics101]

Yes. The first iteration within a time step will use the pressure gradient from the previous time level. The second iteration within the time step will then use the pressure gradient from the first iteration within the time step and so on

[@nomanyousuf4290]

@@fluidmechanics101 Thanks for the reply. Another thing I wanted to ask is that in this video, you are solving steady Navier stokes equation... so this iteration is associated with evolution of velocity with respect to space. what happens when there is unsteady part in the equation (temporal evolution of velocity).. does SIMPLE works in iteration in space only and there is another algorithm that handles the temporal part?

[@fluidmechanics101]

Yes, you are correct. There is another part of the algorithm that handles the time stepping. The most common methods are Euler Implicit, Euler Explicit and 2nd Order backwards differencing, if you wanted to know what to look for 👍

[@amirrezamohammadi7421]

hi I have to use buoyant Boussinesq simple foam in open foam and my question is what is the difference in natural convection problems best regards

[@fluidmechanics101]

It all depends on the temperature differences, 'how strong' the buoyancy forces are and what fluid you are working with. Do you have air? Water? A liquid metal? Argon? Are the temperature differences 10 degrees, 100 degrees or 1000 degrees?

[@cmojr1]

Thanks for that!!

[@chaoyan6097]

Thanks for the video, it is very clear explanation! Don't know if you familiar with DG method, which seems to be very porpular recently for CFD, will you have chance to make a video for that?

[@fluidmechanics101]

Hi Chao, by the DG method i assume you mean the discrete galerkin finite element method? Its not currently on my list of videos as most mainstream CFD codes (which will provide the most value to the most people) use the finite volume method. What code do you use and is it the discrete galerkin approach?

[@chaoyan6097]

@@fluidmechanics101 Hi, yes, what I meant DG method is the Discontinuous Galerkin method. Right now I am using my code, but the CFD code that my research group using includes this FEM solver. Also, as far as I know the CFD and CAA code in DLR (German Aerospace Center) and NASA also includes this solver. The reason I am interested into this method is the higher order accurancy and due to the algorithm structure it is well-suited to be parallelled. I studied the fundamental of this method, but really want to see how other researchers understand it in math space.

[@ramkumars2329]

could u plz explain on similar for compressible flow solutions?.. with OpenFOAM terms like in this one... i am tired of searching on literature where i couldnt get a clear idea on how to solve one..

[@mercurial152]

So, after the first iteration we go back to momentum equation and use grad p from the first poisson equation like a guess and calculate U field again ?

[@fluidmechanics101]

Yep 👍

[@kartiniganesha6741]

Hello professor, I have a question. For solving the NS equation, there is one algorithm which is an explicit predictor-corrector . What's that? And how to apply to the formulation? And if you don't mind, maybe make a video example also. Thanks

[@PhilippGeorgiou]

Hello, I do not understand the step from equation 8 to 9. If you decompose M, H must also be multiplied with U? What am I getting wrong? Thanks in advance!

[@pradyumnchiwhane2577]

Remarkable explaination. Can you help me develop own code for direct numerical simulation?

[@fluidmechanics101]

DNS codes are difficult to write, as the finite volume method often isnt good enough and techniques like the spectral element method are needed. I can't really help you write these codes, but you might want to look into codes that are already available?

[@mortensilcowitz5698]

around 1.29 you mention "leaving out the unsteady term for simplicity". What is that unsteady term? (viscosity i guess) and does that mean that the solution is alway "steady" when that term is absent?

[@fluidmechanics101]

Hi Morten, my apologies, i forgot to talk about the unsteady term. The unsteady term is ‘d U / d t’. This the rate of change of the solution with time. Yes, you are right, when we do not include the unsteady term we are assuming that the solution does not change with time. This is not always the case. If you try and solve for flow around a cylinder at high reynolds number for example, the solution is always unsteady. If you use a steady solver (or dont include d U / d t), you will see that the residuals do not converge and they oscillate. This is because the solution is unsteady and needs d U / d t to be properly resolved. I hope this helps Aidan

[@mortensilcowitz5698]

@@fluidmechanics101 Oh I see, thanks. I was in a state of mind where I thought there was some implicit time step, so once we had solved for U, we could go on to the next "time step" and solve for U again, with the previous U as our starting point or "state". Generally I'm confused about why or when a flow is "steady" or not changing over time. Is it correct to think, that if no external forces are acting on the fluid, and there is no dissipation of energy, (zero viscosity?) then the flow will just remain the same and go on forever?

[@understandingconcepts4516]

Hi. Thanks for the video. I would like to know how you implement boundary conditions such as velocity inlet, outlet, slip and noslip wall, in the context of a staggered grid for SIMPLE algorithm. I have referred material from literature, but nothing seems to work for me.

[@fluidmechanics101]

Staggered grids are a bit more tricky than cell centred grids as most of the popular CFD codes (OpenFOAM, Fluent) use cell centered formulations. Are you writing your own code or are you using a custom code that someone else has written? Do they provide you with source code or a help manual?

[@ilhantalih9949]

Hello sir , can you explain Rhie & Chow interpolation in your videos or would you advise any video that explain it?

[@fluidmechanics101]

I am hoping to do a video on this soon, as it is quite a tricky one. I haven't quite worked out the best explanation for it yet, but hopefully this one is coming soon!

[@ilhantalih9949]

@@fluidmechanics101 Thank you , best regards.

[@kasuora6233]

Hi, I cant get how can MU=-Del(p) be solved? Do we know Del(p) ?

[@parasghumare5502]

What happens after step 3 when the momentum equation is left unsatisfied? Do we then make a guess for pressure again based on previous loop?

[@sabinanton5574]

Hi! How do you solve the Poisson equation in matrix form for the pressure field (Eq. 22)? Is there an iterative procedure, or is it solved directly?

[@fluidmechanics101]

Any interative flow solver will do. Most CFD codes use an algebraic multi grid solver for the pressure equation

[@jesushernanjimenez6471]

Thank you very much for the video. I have a quick question. I am trying to figure out the matrix M, however, there is the non-linear term that complicates things. Any ideas on how to overcome this? I get products of velocities and that can't be represented in the matrix-vector multiplication. I was guessing we could use velocities from previous iterations but I don't know if that would work. I am also using finite differences, I don't know if the problem does not show up when using FV instead.

[@fluidmechanics101]

Yep, you are correct. One of the velocities in the convection term is taken from the previous iteration. You will still have the non linear term, regardless of whether you use the finite volume or finite difference methods

[@idreeskhan1629]

Hi sir in Eq 12 how you are finding H while U is also unkwon?

[@fluidmechanics101]

U is the velocity from the previous iteration (or the initial condition if this is the first time step) 👍

[@bernardocoelho2045]

Hi, Aidan If I have an incompressible periodic flow driven by a pressure gradient, will the SIMPLE method or any other pressure-velocity algorithm be used or not? Because the pressure will be known in Equation 5. Thank you!

[@fluidmechanics101]

Hi Bernardo, yes you can still use a pressure-based algorithm, as you only know the pressure on the boundaries of the domain. The SIMPLE algorithm can be used to calculate the pressure and velocity in all the cells in the domain and the velocity field develops in response to the pressure gradient 👍

[@AhmedTaha-ij8xj]

hello .. can you make a video for the coupled algorithm too ?

[@fluidmechanics101]

Yes, i do plan on doing the coupled solver in future. I would also like to have a look at SIMPLEC and SIMPLER.

[@RCB0B]

What about Boundary conditions for velocity? Is that included in matrix M?

[@ME-et2iy]

Great video! I want to ask you if I use the SIMPLE algorithm in a transient case in Fluent and the time step is converging from the first iteration, does it affect the solution statistics like the moments? I forced the solver to do more than 1 iteration and the moment was changing quite a bit within each time step but I am not sure if the overall average is affected. (I am using LES dynamic Smagorinsky with the default residuals) Thank you!

[@fluidmechanics101]

The best way to check is to do a comparison of a the same case with different time step settings and see if you get a difference in the results you care about 👍 1 iteration per time step doesn't seem like a lot, even if your courant number is less than 1. Try 3 or 4 and see if your results change

[@ME-et2iy]

@@fluidmechanics101 I can't increase the time step because I need to keep CFL low, and reducing it will still take 1 iteration per time step. I can lower the residuals but it is a large case and I can't wait long enough for LES to sample... My main concern or question is: does the SIMPLE scheme do the pressure corrections inside every iteration including the first, or are the corrections made in iteration n+1 after solving the momentum in iteration n? In other words, if the time step only does 1 iteration, does any correction occur at all? Thank you very much for your help!

[@fluidmechanics101]

Yep, even if you have '1 iteration per time step' you will still solve the momentum predictor, pressure corrector and flux correction once. It just means that you don't loop back and do the stages again because your 'momentum predictor' was a good prediction

[@newlenin]

Hi thy for the work, I have one question. Do the H matice (and the other) has to be updated and when yes, where does it happen?

[@prabup3114]

Hi Aiden, Thank you for your videos. Can you also say how the VOF equation goes with it. Thanks

[@idreeskhan1629]

kindly sir upload a video for PISO algorithm as well

[@fluidmechanics101]

Yes, i really need to do a video on the PISO algorithm 😅 i havent got round to it yet but it will be definitely coming soon!

[@idreeskhan1629]

@@fluidmechanics101 thanks sir. i have also another question i am studying the algorithm in OpenFoam i have confusion

[@常亮-y4r]

i just want to know how to prove the converge of P.

[@jamattos9957]

Hey Aidan! I´ve seen the PRESTO! scheme for pressure coupling, does it relate to the SIMPLE algorithm?

[@fluidmechanics101]

It does relate to SIMPLE. But PRESTO! is an interpolation scheme that is used specifically in the pressure correction equation, rather than an overall algorithm (like SIMPLE). The best thing to do would probably be having a good read of the Fluent manual to get an idea of what PRESTO is doing 👍

[@bobreece]

could you cover some under-relaxation factors for simple, piso and pimple

[@fluidmechanics101]

Will do 👍 the default values used by most CFD software are usually fine. If you are having difficulty, it is probably because of your mesh or boundary conditions ...

[@dewis2967]

Please share MATLA/AYNSIS code for incompressible Navier-Stokes equations with algorithm. I will be very thankful to you.

[@fluidmechanics101]

Hi Dewis, unfortunately the ANSYS codes are proprietary and so i dont have access to them and cant share them. However, you can always check out the OpenFOAM code, which is open source and free for you to look at. You can find it on the OpenFOAM git repository through a google search :)

[@mateusalves9558]

Hello, thank you for that excellent video, may I do some questions: 1) When we set the inlet boundary condition to a fixedvalue for velocity, we still have to obtain the velocity field from the momentum equation? 2) The way the algorithm is build, declaring a matrix M and later an diagonal and off-diagonal matrix to obtain a equation for pressure and a velocity corretor (since the velocity have already been predicted from the momentum equation), is a implicit or explicit coupling? The latest question came to me while I was looking the slide 20 from the www.tfd.chalmers.se/~hani/kurser/OS_CFD_2013/KlasJareteg_CoupledSolvers_20130917.pdf Hope to see your reply and more videos because they are fantastics!

[@fluidmechanics101]

Im not sure about 2) but inlet boundary conditions are applied to cell faces. The momentum equations are solved for thr velocity at the cell centroids 🙃

[@mateusalves9558]

@@fluidmechanics101 But to achieve the velocity field from the momentum equations the algorithm get's the gradient of the pressure field , right? Do this gradient come from the boundary conditions ? Thinking that the boundary condition from the outlet is zeroGradient , so it's a interpolation from the cell centers values. I am getting cofused on how do the hole thing starts exactly. Thank you for the attention and hope to see your reply!

[@fluidmechanics101]

We normally use a special interpolation scheme for the pressure gradient (body force weighted or PRESTO for example) to ensure that we dont get checkerboard oscillations. I will be covering this in another video soon 👍

[@mateusalves9558]

@@fluidmechanics101 I Will try to reformulate my question. How and when the boundary and initials conditions enter in the loop of the SIMPLES and PISO algorithm? I have saw ALL the vidros of your channel about this subject and haver studied from others lectures too, but I really have to confirma that because I am getting confused. Thanks Aidan for all the help!

[@fluidmechanics101]

Hi Mateus, the boundary conditions and initial conditions enter into the matrices themselves (the M, A and H matrices). You can always check out my fundamentals course if you would like to see a simple worked example 😊

[@lizarettflavour]

👍

[@serrareabderrahime4528]

Tooooooooooooooop

[@HasanAli-uz5un]

I see, this is the SIMPLE algorithm, not the simple algorithm

[@jadetaylor6630]

Simple algorithm doesn't look simple!

[@jadetaylor6630]

Hey Aidan 😛

[@fluidmechanics101]

Haha i know! I wish they picked a different acronym 😂

[@jadetaylor6630]

I feel like next you need a flat Earth video 😂