Note: Running version 6 and ofcourse R command doesn't work so had to go for simpleFoam -postProcess -func R. Also, file turbulenceproperties:R gets created instead of R in 0 directory. No worries, just rename it to R. While running simulation I faced an error saying keyword div((nu*dev2(T(grad(U))))) undefined so added it and it worked.
Hi Paras :) Thank you from 2 years later. I just wonder where did you find the already changed command for later versions of OpenFOAM. I got stuck here for a while and did not find any explanation about the missing R file online. Did you find this command on the OpenFOAM user mannual?
@@OpenFOAMJozsefNagy how is this R thing done these days ? How is this done these days. The postprocess R doesn't work anyymore and R command doesn't work either.
@@OpenFOAMJozsefNagy hey thanks for the reply and all the great tutorials and links. They have been very helpful- Howabout this R variable that is in the pitz daily tutorial? is this needed only in models with reynolds stress solving? the transient les and keqn simulations ran fine iwithout it but still wondering. Cheers
For anyone trying to run this tutorial on Windows 10 through bash and in a Windows 10 filesystem folder, the function R was replaced by the postprocessing function turbulenceFields, which tries to create a folder named turbulenceModel:R but fails to do so because of the two dots that Windows can't process. To solve this problem just copy the case folder to home through the terminal and run the command there, then change the name of the folder to just R and copy it back to the windows case folder. Thanks a lot Jozsef for your amazing tutorials :)
Hi, I tried to copy the case directory (backWardFacingStep_LRR) in home (which for me is "MIGW64 OpenFOAM-5.x /home") and tried to type "simplefoam -postProcess -func turbulenceFields", but the following error results: "--> FOAM FATAL ERROR: cannot find file "C:/Program Files/blueCFD-Core-2017/msys64/home/backwardfacingstep_LRR/0/R" From function virtual Foam::autoPtr Foam::fileOperations::uncollatedFileOperation::readStream(Foam::regIOobject&, const Foam::fileName&, const Foam::word&, bool) const in file global/fileOperations/uncollatedFileOperation/uncollatedFileOperation.C at line 522. " any suggestion?
It still has an issue with "Program Files". It doesn't like the space in the path. I thought they reaired that. Back in the days before bash on ubuntu on windows I installed blueCFD core on mx c drive directly instead of the suggested location in Program Files.
Dear Sir, your explanations are superb about the openFOAM . Because it was very messy for me at the beginning. Now I can handle it than previous after seeing your tutorial videos. I thankful for you lot for your kind effort you had.
Hi Jozsef! Thank you for all the effort you put in to al your tutorials, the are helping me a lot! At this moment i'm running you tutorials with the v1606+-version of OpenFoam, and there i had to replace the RASProperties file with a turbulenceProperties file. In this file i had to put this text to run the calculations smoothly: simulationType RAS; RAS { RASModel kEpsilon; turbulence on; printCoeffs on; } Thanks again for all you work!
Hi Pieter! You are absolutely right! In this video i used back then version 2.3.0. In the newer versions some changes were made in the turbulence syntax. If you check out the tutorial collection I talked about in one of my recent videos, I do list these changes there: wiki.openfoam.com/Stationary_turbulence_modeling_(RAS)_by_Jozsef_Nagy Of course I cannot redo all the videos for each release, but it is possible for me to give you this information through the tutorial collection. Thanks for your comment, I hope others find it also useful!
Just in case someone has the issue with yPlus function using higher version of openfoam. In order to postProcess yPlus, please try this command 'simpleFoam -postProcess -func yPlus', then you get the yPlus value for each iterations you saved. It works for my case. If you dont add simpleFoam before '-postProcess', then you will get the error of 'Unable to find turbulence model in the database'.
the solution for error encountered @8:00 is to rename frontAndBack faces in 0 folder to defaultFaces before blockMesh and runnig simplefoam .. thanks very much for your fantastic work
But shouldn't y+ have to be either y+>200-300 for a high-Re-treatment, since we're not able to approximate the buffer-layer with a linear- or a log-fuction correctly?
Hello József, I am using LRR for a vertical pipe flow (using a wedge geometry with axis-symmetry BC on front and back of the wedge). I get convergence for p, U (x,y,z), R(xx,yy,zz,xz) with tolerance of 1e-7. However for R(xy, yz) I get convergence only with a tolerance value >= 2e-3. PIMPLE: iteration 729 ... ... time step continuity errors : sum local = 1.00909e-12, global = -6.24038e-13, cumulative = -1.22627e-08 DILUPBiCG: Solving for epsilon, Initial residual = 9.31912e-08, Final residual = 9.31912e-08, No Iterations 0 DILUPBiCG: Solving for Rxx, Initial residual = 9.89784e-08, Final residual = 9.89784e-08, No Iterations 0 DILUPBiCG: Solving for Rxy, Initial residual = 0.00114164, Final residual = 7.47445e-07, No Iterations 1 DILUPBiCG: Solving for Rxz, Initial residual = 9.89262e-08, Final residual = 9.89262e-08, No Iterations 0 DILUPBiCG: Solving for Ryy, Initial residual = 9.70953e-08, Final residual = 9.70953e-08, No Iterations 0 DILUPBiCG: Solving for Ryz, Initial residual = 0.000545211, Final residual = 3.55113e-07, No Iterations 1 DILUPBiCG: Solving for Rzz, Initial residual = 9.93673e-08, Final residual = 9.93673e-08, No Iterations 0 ... ... Can you suggest any option?
Yeah, you have to watch out with LRR, it is a 3D model, so in 2D you have to keep this in mind. This tutorial should only show, how it works in general, actually you'll have to use a 3D model.
Thanks for the reply. I'll try it with a 3D mesh. Can you tell me how does OpenFOAM calculate R knowing the type of RAS model (eg. SST)? Does it use the Boussinesq hypothesis?
Well I don't exactly know, but you can check out the equations in src/turbulenceModels/incompressible/RAS/LRR. There you also find all the available models.
After searching the OpenFOAM code, I finally found the piece of code that indicates how R is calculated. It is in the file named eddyViscosity.C and can be consulted at this link: www.openfoam.com/documentati...8C_source.html At line 117: ((2.0/3.0)*I)*tk() - (nut_)*dev(twoSymm(fvc::grad(this->U_))) When calculating R with the k- ε model, the code for the calculation of the value of nut can be found in kEpsilon.C, line 42 (www.openfoam.com/documentati...8C_source.html) this->nut_ = Cmu_*sqr(k_)/epsilon_;
I am trying to generate R matrix for the LLR model. Using the last OpenFOAM version, and the code "R" does not work. I tried typing: "simplefoam -postProcess -func R" and something happens, but finally it appears the following message: "No MRF models present functionObjects::turbulenceFields R writing field: turbulenceProperties:R End" and no "R" file is created in the "0" folder. What should I do?
Wow, I have no experience with the R model in recent versions. Sorry. Maybe google helps or post this issue on cfd-online. Maybe someone can help you there.
I have the same issue. Looks like OpenFOAM demands file MRFProperties in constant folder for proper R execution. Example of such a file could be found in simpleFoam/mixerVessel2D tutorial.
Even I tried to rename turbulenceProperties:R to R in 0 folder then, openFOAM found error in fvSolution, I guess I cannot follow until 21:12. The LRR model might not working properly here
Hi, I'm trying to do the "homework" and you seem to suggest that the results of the k-epsilon model are the best and modifications should be done to the other two models to get similar results. Do I understand you correctly? If that is so, what indicates that the results of the k-Epsilon model are those that can be trusted? Is it because the initial residuals are the lowest? I'm asking because I'm trying to do a similar case and I would like to know what are my criteria to determine the best result, other than using experimental results as a benchmark.
I changed the grid 4x ( 2 x 2 x 1 ) & got y+ as 19 & 22 for upper wall & lower wall respectively But Y max value is 35 which is out of range ( is it because of mesh grading scheme ? which i used 0.e near the step area.?)
Hello joseph, i started a new reserch project and your videos are amazing to have the hand on openfoam. I was trtying to reproduce this tutorial with the actual version of openfoam (V12). For the velocity i was able to obtain relativeley similar results. Not for the pressure. And for k epsilon and nut the values does not change after 2000 iterations. I verified all the parameters and everthing is ok. Is it a version thing ? Thanks for your response and continue with the videos without stoppin ..'that's what she said' :)
Jozsef Sir, first of all thanks for this tutorial. I have some doubts: when you check for convergence, you look at initial residual, not final residual; why so? 2nd thing, we have set the no. of iterations, this simulation will go through, under controlDict-> endTime=2000(say); at the same time, we set tolerance limit to 10^-6 , what is the use of the later settings as programme will end after 2000 iterations, no matter whether the tolerance limit is achieved or not? Kindly explain to me. Thanks.
It is up to you, where you define your simulation to be convergent. It is always depending on the flow. Sometimes it is 1e-3 sometime 1e-6 sometimes 1e-12. You have to check always the flow itself. Sometime you need 100 iterations, sometimes 20000.
Hello Prof. Nagy, I have some problem regarding convergence. There are two stopping mechanisms for the simulation, one is the convergenslce condition and the other is defined as the number of time steps. Which one is prioritized by the program and how can we change it? My simulation (when using a relaxation factor of 0.9) converges after 18 time steps and wont reach the 2000th timestep, I was able to reach the 1980th time step by setting the relaxation factor to 0.2 I guess there should be an option to tell the program to continue the simulation till the end, although the convegence has reached and continuing seems pointless! Could you assist me about this problem?
Thank you so much Mr Nagy, But would you please make a tutorial about the aerodynamically generated noise in an incompressible flow (AAC) and thank you again.
This is a rather special case. On this channel my main focus is to post videos for my polymer engineering students, so I will finish the videos on turbulence modeling, multiphase modeling, meshing, CHT before I start working on special cases. I cannot promise anything, but I can put this topic on my list.
Why large initial residue number is considered to be bad, if the final residual achieves the desired accuracy? Or the point is, that in this problem it is not a serious issue to wait a bit for achieving stationarity, but in more realistic cases, large initial resildual will lead to long convergence, or even failure of it?
It gives the change you get by using a new e.g. velocity value in the pressure equation. This means, that your simulation still changes a lot and is not convergent yet.
In my experience I've had better results visualizing separation bubbles by looking for where the wall shear stress equals zero as opposed to tracking Ux, Uy, or Uz. On complex, 3D geometries, the velocity alone may under-predict the true scale of the separation.
Of course, however this is an introductory tutorial, I cannot go crazy. If you are interested in these flows, you should check out all the features in the source code.
@@OpenFOAMJozsefNagy Actually they aren't until I used foamToVTK to open VTK version any idea why this is happening?? Ps- I am using OpenFOAM8 and paraview 5.6.0
@@mahashayjee5545 Create an empty .foam file. The command is "touch open.foam". Then you can open up this file in Paraview and it will visualize the results not the VTK files.
@@OpenFOAMJozsefNagy hello, when I tried running the simulation, I received a floating point exception error and I think because the solution started to diverge. I fixed it by reducing the under-relaxation factors but I am not sure if the under-relaxation factors could impact the final solution. What do you think?
Hi Josef, First off your tutorials are absolutely fantastic! It has been immensely useful to me so far. I noticed that my 'Time' at convergence on my machine is significantly smaller than yours (mine is 375 vs 2000 on yours) I seem to have a faster machine, but shouldn't the endtime be the same as we specified it at 2000? I followed your instructions exactly. Just hoping I didn't do something wrong...
I'm using OF v2.3.x (not sure what the x is referring to but I think it's v2.3.1). My settings are the same as yours, although I am running my simulation with residual control while you are not. I still would have thought that my solution would still extend to the specified time step though?
I have a Floating Point exception (core dumped) when I run the simulation kEpsilon (Time=19s). What should I do to correct this? I did it step by step like the tutorial. Thank you so much! OpenFoamVersion v7.
I had this problem as well. In the file k, I have changed its value from 0.2 to 0.3, and the simulation has worked. I am figuring it out, because I still don`t know the reason of this happens.
@@murilobrunazzo7053 Glad to found this comment, it solved my problem in OpenFOAM 8 , but the graph in ParaView differs much as what Joszef was explained in video right?
[SOLVED] Whenever I run the "R " command or try to find the y+ value (yPlusRAS) I get the following error : ImportError: /opt/OpenFOAM/ThirdParty-v1912/platforms/linux64/gcc-6.3.0/lib64/libstdc++.so.6: version `GLIBCXX_3.4.26' not found (required by /lib/x86_64-linux-gnu/libapt-pkg.so.6.0) I tried downloading libstdc++ but it doesn't get updated as the newest version has already been installed. Can someone please help me out ? SOLUTION: "In newer versions of OpenFOAM the commands yPlus, yPlusLES and/or yPlusRAS are not used. Instead AFTER your simulation run the command simpleFoam -postProcess -func yPlus" wiki.openfoam.com/Stationary_turbulence_modeling_(RAS)_by_Jozsef_Nagy
Hello Everyone, first case(k-epsilon) results is very different in OpenFOAM 5 (i checked repeteadly.) and even the simulation crashes in 5000 iteration with 2nd order upwind scheme for turbulence kinetice energy. The initial residuals are very high at last time step. Are these differences in solution between software versions usual? And the convenient discretization schemes differ from the version to version? *Mr. Nagy thank you for the tutorials.
Hello József, all these video tutorials are very beneficial for the openFoam learner. i just started using openfoam, i want to ask you how can i plot averaged Reynolds stress(u'u', v'v', w'w', u'v') and wall shear stress in ke, kw and RAS model thank you
Dear Sir, it seems that the post-processing commands, including "yPlusRAS", do not work anymore in OpenFoam 7. I would be grateful if you would have any suggestions.
You are correct. The video was created for version 2.3.0. I cannot redo the videos for each version, but updates can be found on the tutorial wiki: wiki.openfoam.com/Stationary_turbulence_modeling_(RAS)_by_Jozsef_Nagy There you can find, how you can calculate yPlus in recent versions.
Hi Sushrut if you are haveing issue with calculating y Plus in newer version simply run the following command in terminal : simpleFoam -postProcess -func yPlus . hope this help you and everyone in future
Hi Mr. Nagy Thank you very much for your tutorial as always. I am having doubt, if you kindly help me to clarify those, it will be very useful. I have spherical mesh domain and at the center I have a plate and I wish to compute aerodynamic coefficients at different conditions. I have to use compressible flow. I am using freestram as BC for p and U, but not sure as I am getting diverged. Also I see negative k and epsilon (which is wrong). How can I calculate k and epsilon for farfield. Thank you very much. And happy new year. Bye
Hello, I'm running openFOAM v9, and when i execute the command turbulenceFields (former R) to generate R folder, I cannot make it works. Can someone help me? Below i reported the reports on my terminal after the execution of simpleFoam -postProcess -func turbulenceFields Create time Create mesh for time = 0 SIMPLE: No convergence criteria found Time = 0 Reading field p Reading field U Reading/calculating face flux field phi Selecting incompressible transport model Newtonian Selecting turbulence model type RAS Selecting RAS turbulence model kEpsilon RAS { model kEpsilon; turbulence on; printCoeffs on; Cmu 0.09; C1 1.44; C2 1.92; C3 0; sigmak 1; sigmaEps 1.3; } No MRF models present No fvModels present No fvConstraints present turbulenceFields turbulenceFields: no fields requested to be stored End
@@OpenFOAMJozsefNagy there are other commads of text files I have to add to avoid this interruption without results? I have inserted the file turbulenceProperties in /constant but it seams not effective.
When i put the kEpsilon model to run, the simulation only runs till the 1710th iteration with initial residuals on the order of 1e-1. I can't find where am i messing this up. Any guess?
It seems that is actually the other way around. But anyways, the whole things was really my bad,i set the relaxation factor for the field "p" as 0.03, when it maybe was originally set as 0.3 as you teach on the video. But even though i'm getting 2000 iterations, the initial residual for Uy is on the order of 1e-3 and the number of iterations vary from 3 to 8. Finally, i want to know if you have a suggestion of course, but also an explanation on why that happens. And thank u so much for the replies. @@OpenFOAMJozsefNagy
Actually, for the sake of the community who watch these videos, i should come back and say that the actual reason why the number of iterations was beeing limited was due to a line right below the "nNonOrthogonalCorrectors" in SIMPLE, "consistent yes;". Apparently , when you delete the line, which is found, i guess, in the latest versions of OpenFoam the simulation runs completely till the end. Nevertheless, the problems with the number of iterations still exist. Now, when i'd said that the problem for the number of iterations was the relaxation factor of the field "p", i was running the simulation on my home computer, which i had already deleted the "consistent yes;" line and had only to fix the relaxation factor value from "p". In this case, the number of iterations was going merely till the 51 iteration. It was only for the case where i had the relaxation factor already properly set and the mysterious line "consistent yes;" added that the number of iterations was going to the 1710th. I must say that all this is too confusing to me. Could you help me with this?? I still don't know why the number of iterations are that high and why the initial residualfor Uy is on the order of 1e-3. @@OpenFOAMJozsefNagy
Jozsef, thank you for your nice work, which is informable. By the way, when I try to run a similar case with LRR using v8, an error"attempted assigment to self" poped up. Do you know what on earth it is?
Hi, I'm a student writing my bachelor thesis with OpenFoam. I'd like to do your backstep simulation with the rhoSimpleFoam solver(compressible flow). Unfortunatelly, I couldn't find any valid ressources and the case tutorial is unhelpful. What would you recommend?
+Vlad Chalapco Do this tutorial, do a rhoSimpleFoam tutorial and change the dictionaries from this one to the needed ones for rhoSimpleFoam. Step by step.
@@chensongtao9244 Might be the difference in OpenFOAM version. I do not know, why you only get the results for the first time. Maybe you ran the simulation in parallel?
Hello Mr. József, great tutorials! This is the first time I got stuck and i need to ask this question. I used this command to create R values: -simpleFoam -postProcess -func R I got turbulenceProperties:R file in my 0 folder but instead of getting: internalField nonuniform List
József Nagy I cannot stop watching your videos József :-) Thank you for this effortJust another simply question. You use the same grid for kepsilon and komega model. I though they required different meshes as the don't use the same wall functions, and they don't need the same y+ don't they? I am a newbie to OpenFOAM
Rubén B. Don't forget, that this is a tutorial and not a scientific work. You should be able to complete it within a reasonable time. Of course you have to think about the mesh. And thanks for your kind words!
Your tutorial videos have taught me a lot about OpenFOAM. Since this is my first time dealing with CFD, it is very useful. Can I seek your assistance in the "homework" that you mention in the video though? 1) My idea for making nut = 1e-3 for the k-omega model is to use nut as 1e-3 and reverse enginer to get the omega value. Is this acceptable? 2) For LRR model, any hints on how to change or correct the nut to 1e-3?
1) My idea for making nut = 1e-3 for the k-omega model is to use nut as 1e-3 and reverse enginer to get the omega value. Is this acceptable? Try it. 2) For LRR model, any hints on how to change or correct the nut to 1e-3? I know, this is rather difficult. Try with upwind, use low underrelaxation factors and really wait for the settings to be convergent, then change the discretization.
in the version 2012 at time = 11 i am getting error Time = 11 smoothSolver: Solving for Ux, Initial residual = 1, Final residual = 7.34729e-07, No Iterations 39 smoothSolver: Solving for Uy, Initial residual = 1, Final residual = 7.34771e-07, No Iterations 39 #0 Foam::error::printStack(Foam::Ostream&) at ??:? #1 Foam::sigFpe::sigHandler(int) at ??:? #2 ? in /lib/x86_64-linux-gnu/libpthread.so.0 #3 Foam::scalarProduct::type Foam::sumProd(Foam::UList const&, Foam::UList const&) at ??:? #4 Foam::PCG::scalarSolve(Foam::Field&, Foam::Field const&, unsigned char) const at ??:? #5 Foam::GAMGSolver::solveCoarsestLevel(Foam::Field&, Foam::Field const&) const at ??:? #6 Foam::GAMGSolver::Vcycle(Foam::PtrList const&, Foam::Field&, Foam::Field const&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::PtrList&, Foam::PtrList&, unsigned char) const at ??:? #7 Foam::GAMGSolver::solve(Foam::Field&, Foam::Field const&, unsigned char) const at ??:? #8 Foam::fvMatrix::solveSegregated(Foam::dictionary const&) at ??:? #9 Foam::fvMatrix::solveSegregatedOrCoupled(Foam::dictionary const&) at ??:? #10 Foam::fvMesh::solve(Foam::fvMatrix&, Foam::dictionary const&) const at ??:? #11 ? in /usr/lib/openfoam/openfoam2012/platforms/linux64GccDPInt32Opt/bin/simpleFoam #12 __libc_start_main in /lib/x86_64-linux-gnu/libc.so.6 #13 ? in /usr/lib/openfoam/openfoam2012/platforms/linux64GccDPInt32Opt/bin/simpleFoam Floating point exception (core dumped)
Thank you for the information. I just tested this tutorial with v2012 and it nicely converges for me. There may be a case setup problem on your end. Maybe you forgot something?
Well this is a tutorial, it should give you a first idea about the workflow of OpenFOAM. For an actual simulation you should by between 5 and 30, however for a complicated geometry this will be a challenge. The question is always, how big your error is, which you create by being outside of this interval.
Hi, Mr. Nagy, thank you very much for your lectures on CFD , they have helped me a lot on CFD. Nagy, if you could, I have a doubt about this lecture of yours. At the moment of the first simulation using simpleFoam, this message appeared in my terminal: "Time = 28 smoothSolver: Solving for Ux, Initial residual = 0.578291, Final residual = 1.55951e+24, No Iterations 1000 smoothSolver: Solving for Uy, Initial residual = 0.578291, Final residual = 1.03625e+25, No Iterations 1000 #0 Foam::error::printStack(Foam::Ostream&) at ??:? #1 Foam::sigFpe::sigHandler(int) at ??:? #2 ? in "/lib/x86_64-linux-gnu/libc.so.6" #3 double Foam::sumProd(Foam::UList const&, Foam::UList const&) at ??:? #4 Foam::PCG::solve(Foam::Field&, Foam::Field const&, unsigned char) const at ??:? #5 Foam::GAMGSolver::solveCoarsestLevel(Foam::Field&, Foam::Field const&) const at ??:? #6 Foam::GAMGSolver::Vcycle(Foam::PtrList const&, Foam::Field&, Foam::Field const&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::PtrList&, Foam::PtrList&, unsigned char) const at ??:? #7 Foam::GAMGSolver::solve(Foam::Field&, Foam::Field const&, unsigned char) const at ??:? #8 Foam::fvMatrix::solveSegregated(Foam::dictionary const&) at ??:? #9 Foam::fvMatrix::solve(Foam::dictionary const&) in "/opt/openfoam6/platforms/linux64GccDPInt32Opt/bin/simpleFoam" #10 Foam::fvMatrix::solve() in "/opt/openfoam6/platforms/linux64GccDPInt32Opt/bin/simpleFoam" #11 ? in "/opt/openfoam6/platforms/linux64GccDPInt32Opt/bin/simpleFoam" #12 __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6" #13 ? in "/opt/openfoam6/platforms/linux64GccDPInt32Opt/bin/simpleFoam" Exceção de ponto flutuante (imagem do núcleo gravada) // in English, Floating Point exception (core dumped) " However, reading some comments, I have changed the value of k from 0.2 to 0.3, and the simulation has worked well. I still do not know the reason of it happens I have followed your steps to do the simulation, such as setting U = 7.333. I use openFoam version 6. Mr. Nagy, would you have any suggestion about it? Thank you a lot!
Thanks Mr.Jozsef Nagy !! Priceless Tutorials !! I am getting started with OpenFOAM !! Have one question though !! Are quality of OpenFOAM results on par with commercial codes !! Thanks again !! Can u just improve the fonts size on screen.. or color also!! White would be real good
If it is comparable... If you ask me or other scientists/engineers working with OpenFOAM, we would say yes. If you ask someone using a commercial tool, they might give you a different answer. I think it depends on the problem. I would say yes, because in OpenFOAM, you can implement everything, if it is missing from the official release. You just have to take your time.
Whenever i try to execute R as per the video, I end up with this message ""R has been superseded by the '-postProcess' solver command-line option, e.g. simpleFoam -postProcess -func R"" ANYONE with a solution!!
Hi~ Thanks to your video. It's really helpful to me. by the way, I got a error message when I followed this video. can you let me know the reason? Please see below message. boundaryField FrontAndBack not found in object U at time = 0 actually, I got this message for all case such as U, p, epsilon, v2 etc. I think I need to put boundary condition for each factor.
I documented my reading of your tutorial at www.cfd-online.com/Forums/openfoam-solving/227418-simplefoam-tutorial-pitzdaily-using-reynolds-stress-tensor-lrr-rasmodel.html . I hope this contributes to your work. Thank you.
hello, Jozsef. Your videos really have helped me a lot when I turned to OpenFOAM. However, recently, I encountered another problem which puzzled me a lot. in the same case, with same mesh, same discretization schemes, with kEpsilon family models, it works well(mean converge well), however, with kOmegaSST, it always diverges. I have posted my problem in CFDOnline:www.cfd-online.com/Forums/openfoam-solving/160957-sonicfoam-komegasst-cannot-converge.html#post568569 Would you please have a look at it when you are avaliable? Thanks a lot in advance.
+Shiqi Wang Hey, if you are using sonicFoam, you are solving a transient problem. I don't understand your definition of convergence. Do you mean, that the flow converges with time towards a stationary solution?
Usually in this case you get a limitation of one of the turbulent quantities, or maybe both. If this is the case, try and rethink the (inlet) bounadry conditions and the value on that boundary for that quantity. Hope this helps.
Thank you Mr.Nagy for these very awesome tutorials!
I am glad I can help!
Note: Running version 6 and ofcourse R command doesn't work so had to go for simpleFoam -postProcess -func R. Also, file turbulenceproperties:R gets created instead of R in 0 directory. No worries, just rename it to R. While running simulation I faced an error saying keyword div((nu*dev2(T(grad(U))))) undefined so added it and it worked.
Well done.
Hi Paras :)
Thank you from 2 years later. I just wonder where did you find the already changed command for later versions of OpenFOAM. I got stuck here for a while and did not find any explanation about the missing R file online. Did you find this command on the OpenFOAM user mannual?
@@OpenFOAMJozsefNagy how is this R thing done these days ? How is this done these days. The postprocess R doesn't work anyymore and R command doesn't work either.
@@jeromepowelltoecrum6044 Forget this model. Especially in the .org version.
@@OpenFOAMJozsefNagy hey thanks for the reply and all the great tutorials and links. They have been very helpful-
Howabout this R variable that is in the pitz daily tutorial? is this needed only in models with reynolds stress solving? the transient les and keqn simulations ran fine iwithout it but still wondering. Cheers
For anyone trying to run this tutorial on Windows 10 through bash and in a Windows 10 filesystem folder, the function R was replaced by the postprocessing function turbulenceFields, which tries to create a folder named turbulenceModel:R but fails to do so because of the two dots that Windows can't process. To solve this problem just copy the case folder to home through the terminal and run the command there, then change the name of the folder to just R and copy it back to the windows case folder.
Thanks a lot Jozsef for your amazing tutorials :)
Excellent point! Thank you!
Hi, I tried to copy the case directory (backWardFacingStep_LRR) in home (which for me is "MIGW64 OpenFOAM-5.x /home") and tried to type "simplefoam -postProcess -func turbulenceFields", but the following error results: "--> FOAM FATAL ERROR:
cannot find file "C:/Program Files/blueCFD-Core-2017/msys64/home/backwardfacingstep_LRR/0/R"
From function virtual Foam::autoPtr Foam::fileOperations::uncollatedFileOperation::readStream(Foam::regIOobject&, const Foam::fileName&, const Foam::word&, bool) const
in file global/fileOperations/uncollatedFileOperation/uncollatedFileOperation.C at line 522.
"
any suggestion?
It still has an issue with "Program Files". It doesn't like the space in the path. I thought they reaired that. Back in the days before bash on ubuntu on windows I installed blueCFD core on mx c drive directly instead of the suggested location in Program Files.
Thank You for reply! I Will try To reinstall It
Great!
Congratulation Mr Nagy
Fantastic tutorials
I hope you have time to continue with this work
Thank you very much
Alberto
Dear Sir,
your explanations are superb about the openFOAM . Because it was very messy for me at the beginning. Now I can handle it than previous after seeing your tutorial videos.
I thankful for you lot for your kind effort you had.
I am glad.
Hi Jozsef!
Thank you for all the effort you put in to al your tutorials, the are helping me a lot!
At this moment i'm running you tutorials with the v1606+-version of OpenFoam, and there i had to replace the RASProperties file with a turbulenceProperties file. In this file i had to put this text to run the calculations smoothly:
simulationType RAS;
RAS
{
RASModel kEpsilon;
turbulence on;
printCoeffs on;
}
Thanks again for all you work!
Hi Pieter! You are absolutely right! In this video i used back then version 2.3.0. In the newer versions some changes were made in the turbulence syntax. If you check out the tutorial collection I talked about in one of my recent videos, I do list these changes there:
wiki.openfoam.com/Stationary_turbulence_modeling_(RAS)_by_Jozsef_Nagy
Of course I cannot redo all the videos for each release, but it is possible for me to give you this information through the tutorial collection. Thanks for your comment, I hope others find it also useful!
Just in case someone has the issue with yPlus function using higher version of openfoam. In order to postProcess yPlus, please try this command 'simpleFoam -postProcess -func yPlus', then you get the yPlus value for each iterations you saved. It works for my case. If you dont add simpleFoam before '-postProcess', then you will get the error of 'Unable to find turbulence model in the database'.
If you run the OpenFoam with higher version, you could try the command "R" with "simpleFoam -postProcess -func R"
the solution for error encountered @8:00 is to rename frontAndBack faces in 0 folder to defaultFaces before blockMesh and runnig simplefoam .. thanks very much for your fantastic work
+Hani Ceaser Thanks for the input.
But shouldn't y+ have to be either y+>200-300 for a high-Re-treatment, since we're not able to approximate the buffer-layer with a linear- or a log-fuction correctly?
This is a tutorial. You are correct, for a real simulation, you have to modify the mesh in order to fulfill the correct y+ values.
@@OpenFOAMJozsefNagy Okay, I just wasn't sure. Thank you very much for the great tutorials !
Hello József,
I am using LRR for a vertical pipe flow (using a wedge geometry with axis-symmetry BC on front and back of the wedge). I get convergence for p, U (x,y,z), R(xx,yy,zz,xz) with tolerance of 1e-7. However for R(xy, yz) I get convergence only with a tolerance value >= 2e-3.
PIMPLE: iteration 729
...
...
time step continuity errors : sum local = 1.00909e-12, global = -6.24038e-13, cumulative = -1.22627e-08
DILUPBiCG: Solving for epsilon, Initial residual = 9.31912e-08, Final residual = 9.31912e-08, No Iterations 0
DILUPBiCG: Solving for Rxx, Initial residual = 9.89784e-08, Final residual = 9.89784e-08, No Iterations 0
DILUPBiCG: Solving for Rxy, Initial residual = 0.00114164, Final residual = 7.47445e-07, No Iterations 1
DILUPBiCG: Solving for Rxz, Initial residual = 9.89262e-08, Final residual = 9.89262e-08, No Iterations 0
DILUPBiCG: Solving for Ryy, Initial residual = 9.70953e-08, Final residual = 9.70953e-08, No Iterations 0
DILUPBiCG: Solving for Ryz, Initial residual = 0.000545211, Final residual = 3.55113e-07, No Iterations 1
DILUPBiCG: Solving for Rzz, Initial residual = 9.93673e-08, Final residual = 9.93673e-08, No Iterations 0
...
...
Can you suggest any option?
Yeah, you have to watch out with LRR, it is a 3D model, so in 2D you have to keep this in mind. This tutorial should only show, how it works in general, actually you'll have to use a 3D model.
Thanks for the reply. I'll try it with a 3D mesh. Can you tell me how does OpenFOAM calculate R knowing the type of RAS model (eg. SST)? Does it use the Boussinesq hypothesis?
Well I don't exactly know, but you can check out the equations in src/turbulenceModels/incompressible/RAS/LRR. There you also find all the available models.
After searching the OpenFOAM code, I finally found the piece of code that indicates how R is calculated.
It is in the file named eddyViscosity.C and can be consulted at this link: www.openfoam.com/documentati...8C_source.html
At line 117: ((2.0/3.0)*I)*tk() - (nut_)*dev(twoSymm(fvc::grad(this->U_)))
When calculating R with the k- ε model, the code for the calculation of the value of nut can be found in kEpsilon.C, line 42 (www.openfoam.com/documentati...8C_source.html)
this->nut_ = Cmu_*sqr(k_)/epsilon_;
Thank you for this video. Could you make more videos to explain how calculation is done in solvers in detail?
Congratulations on your tutorials mr.Nagy.I would like to ask you why it can't find the command yPlusRAS?
Hi, I also have this kind of problem, did you solve that?
Maybe its not built-in inside OpenFOAM for windows...I also have this problem
@Anshul Singh Tomar am17s018 It works
Hello Prof. I am using openFOAM v-7 and I can't run the command 'R'. can you please tell me what are the other options to generate R file? Thanks.
Got the same issue !
Yes, this command was discontinued.
maybe you could try with the command "simpleFoam -postProcess -func R"
I am trying to generate R matrix for the LLR model. Using the last OpenFOAM version, and the code "R" does not work.
I tried typing: "simplefoam -postProcess -func R" and something happens, but finally it appears the following message:
"No MRF models present
functionObjects::turbulenceFields R writing field: turbulenceProperties:R
End"
and no "R" file is created in the "0" folder.
What should I do?
Wow, I have no experience with the R model in recent versions. Sorry. Maybe google helps or post this issue on cfd-online. Maybe someone can help you there.
I have the same issue. Looks like OpenFOAM demands file MRFProperties in constant folder for proper R execution. Example of such a file could be found in simpleFoam/mixerVessel2D tutorial.
Even I tried to rename turbulenceProperties:R to R in 0 folder then, openFOAM found error in fvSolution, I guess I cannot follow until 21:12. The LRR model might not working properly here
In later versions the LRR model changed a lot. This is correct.
I had the same issue :(
Hi,
I'm trying to do the "homework" and you seem to suggest that the results of the k-epsilon model are the best and modifications should be done to the other two models to get similar results. Do I understand you correctly? If that is so, what indicates that the results of the k-Epsilon model are those that can be trusted? Is it because the initial residuals are the lowest? I'm asking because I'm trying to do a similar case and I would like to know what are my criteria to determine the best result, other than using experimental results as a benchmark.
I changed the grid 4x ( 2 x 2 x 1 ) & got y+ as 19 & 22 for upper wall & lower wall respectively
But Y max value is 35 which is out of range ( is it because of mesh grading scheme ? which i used 0.e near the step area.?)
Hello joseph, i started a new reserch project and your videos are amazing to have the hand on openfoam. I was trtying to reproduce this tutorial with the actual version of openfoam (V12). For the velocity i was able to obtain relativeley similar results. Not for the pressure. And for k epsilon and nut the values does not change after 2000 iterations. I verified all the parameters and everthing is ok. Is it a version thing ? Thanks for your response and continue with the videos without stoppin ..'that's what she said' :)
Jozsef Sir, first of all thanks for this tutorial. I have some doubts: when you check for convergence, you look at initial residual, not final residual; why so? 2nd thing, we have set the no. of iterations, this simulation will go through, under controlDict-> endTime=2000(say); at the same time, we set tolerance limit to 10^-6 , what is the use of the later settings as programme will end after 2000 iterations, no matter whether the tolerance limit is achieved or not? Kindly explain to me. Thanks.
It is up to you, where you define your simulation to be convergent. It is always depending on the flow. Sometimes it is 1e-3 sometime 1e-6 sometimes 1e-12. You have to check always the flow itself. Sometime you need 100 iterations, sometimes 20000.
Hello Prof. Nagy,
I have some problem regarding convergence. There are two stopping mechanisms for the simulation, one is the convergenslce condition and the other is defined as the number of time steps. Which one is prioritized by the program and how can we change it?
My simulation (when using a relaxation factor of 0.9) converges after 18 time steps and wont reach the 2000th timestep, I was able to reach the 1980th time step by setting the relaxation factor to 0.2
I guess there should be an option to tell the program to continue the simulation till the end, although the convegence has reached and continuing seems pointless!
Could you assist me about this problem?
Thank you so much Mr Nagy, But would you please make a tutorial about the aerodynamically generated noise in an incompressible flow (AAC) and thank you again.
This is a rather special case. On this channel my main focus is to post videos for my polymer engineering students, so I will finish the videos on turbulence modeling, multiphase modeling, meshing, CHT before I start working on special cases. I cannot promise anything, but I can put this topic on my list.
József Nagy OK .. I'm counting minutes till your next video
Why large initial residue number is considered to be bad, if the final residual achieves the desired accuracy? Or the point is, that in this problem it is not a serious issue to wait a bit for achieving stationarity, but in more realistic cases, large initial resildual will lead to long convergence, or even failure of it?
It gives the change you get by using a new e.g. velocity value in the pressure equation. This means, that your simulation still changes a lot and is not convergent yet.
Sir, my simulation is stopping once it is getting converged. So, what changes should I make to extend the simulation behind the convergence?
In my experience I've had better results visualizing separation bubbles by looking for where the wall shear stress equals zero as opposed to tracking Ux, Uy, or Uz. On complex, 3D geometries, the velocity alone may under-predict the true scale of the separation.
Of course, however this is an introductory tutorial, I cannot go crazy. If you are interested in these flows, you should check out all the features in the source code.
@József Nazy When I ran this simulation k, epsilon, nuT aren't appearing in paraview where I went wrong?
Are the present in the results folders?
@@OpenFOAMJozsefNagy Actually they aren't until I used foamToVTK to open VTK version any idea why this is happening??
Ps- I am using OpenFOAM8 and paraview 5.6.0
@@mahashayjee5545 Create an empty .foam file. The command is "touch open.foam". Then you can open up this file in Paraview and it will visualize the results not the VTK files.
@@OpenFOAMJozsefNagy Is there any problem with results in VTK format I am asking will there be any errors or not?
@@mahashayjee5545 If the command runs it is an alternative to the .foam approach, but if it fails the .foam approach is always good.
In the fvSolution, under the class SIMPLE after nOrthogonalcorrectors, if we put "consistent yes", the solution doesn't converge. Why is that?
Difficult to say.
I am getting a floating point exception error.
At which point?
@@OpenFOAMJozsefNagy hello,
when I tried running the simulation, I received a floating point exception error and I think because the solution started to diverge. I fixed it by reducing the under-relaxation factors but I am not sure if the under-relaxation factors could impact the final solution. What do you think?
@@ashishtiwari1912 No they don't, they just slow down the simulation and make it more stable. No worries.
Hi Josef,
First off your tutorials are absolutely fantastic! It has been immensely useful to me so far. I noticed that my 'Time' at convergence on my machine is significantly smaller than yours (mine is 375 vs 2000 on yours) I seem to have a faster machine, but shouldn't the endtime be the same as we specified it at 2000? I followed your instructions exactly. Just hoping I didn't do something wrong...
Oscar Wilsby Hi, hmm so your OpenFOAM converges faster. Which version are you using? Do you use the same settings in fvSolution?
I'm using OF v2.3.x (not sure what the x is referring to but I think it's v2.3.1). My settings are the same as yours, although I am running my simulation with residual control while you are not. I still would have thought that my solution would still extend to the specified time step though?
If you reach your desired value of residuals, the simulation stops. Try a lower value, the simulation should run further.
I have a Floating Point exception (core dumped) when I run the simulation kEpsilon (Time=19s). What should I do to correct this? I did it step by step like the tutorial. Thank you so much!
OpenFoamVersion v7.
decrease the value of K in 0 folder
I had this problem as well. In the file k, I have changed its value from 0.2 to 0.3, and the simulation has worked. I am figuring it out, because I still don`t know the reason of this happens.
@@murilobrunazzo7053 Glad to found this comment, it solved my problem in OpenFOAM 8 , but the graph in ParaView differs much as what Joszef was explained in video right?
[SOLVED]
Whenever I run the "R " command or try to find the y+ value (yPlusRAS) I get the following error :
ImportError: /opt/OpenFOAM/ThirdParty-v1912/platforms/linux64/gcc-6.3.0/lib64/libstdc++.so.6: version `GLIBCXX_3.4.26' not found (required by /lib/x86_64-linux-gnu/libapt-pkg.so.6.0)
I tried downloading libstdc++ but it doesn't get updated as the newest version has already been installed.
Can someone please help me out ?
SOLUTION: "In newer versions of OpenFOAM the commands yPlus, yPlusLES and/or yPlusRAS are not used. Instead AFTER your simulation run the command simpleFoam -postProcess -func yPlus"
wiki.openfoam.com/Stationary_turbulence_modeling_(RAS)_by_Jozsef_Nagy
Great!
I love you！
Hello Everyone, first case(k-epsilon) results is very different in OpenFOAM 5 (i checked repeteadly.) and even the simulation crashes in 5000 iteration with 2nd order upwind scheme for turbulence kinetice energy. The initial residuals are very high at last time step. Are these differences in solution between software versions usual? And the convenient discretization schemes differ from the version to version?
*Mr. Nagy thank you for the tutorials.
You are welcome, and yes there were major changes to the turbulence models since this version of OpenFOAM.
Hello József,
all these video tutorials are very beneficial for the openFoam learner. i just started using openfoam, i want to ask you how can i plot averaged Reynolds stress(u'u', v'v', w'w', u'v') and wall shear stress in ke, kw and RAS model
thank you
Dear Sir,
it seems that the post-processing commands, including "yPlusRAS", do not work anymore in OpenFoam 7. I would be grateful if you would have any suggestions.
It was moved into the postProcessing function. On wiki.openfoam.com/Tutorials the syntax updates for newer versions are mentioned.
mine gives error for finding yplus yPlusRAS:command not found what to do?
You are correct. The video was created for version 2.3.0. I cannot redo the videos for each version, but updates can be found on the tutorial wiki: wiki.openfoam.com/Stationary_turbulence_modeling_(RAS)_by_Jozsef_Nagy
There you can find, how you can calculate yPlus in recent versions.
Thank you sir
Hi Sushrut if you are haveing issue with calculating y Plus in newer version simply run the following command in terminal : simpleFoam -postProcess -func yPlus . hope this help you and everyone in future
Hi Mr. Nagy Thank you very much for your tutorial as always. I am having doubt, if you kindly help me to clarify those, it will be very useful. I have spherical mesh domain and at the center I have a plate and I wish to compute aerodynamic coefficients at different conditions. I have to use compressible flow. I am using freestram as BC for p and U, but not sure as I am getting diverged. Also I see negative k and epsilon (which is wrong). How can I calculate k and epsilon for farfield. Thank you very much. And happy new year. Bye
This can be numerics. Did you change your discretization schemes?
@@OpenFOAMJozsefNagy no I didn't change. I shall take your advice and try again. Thanks for your suggestion.
Hello, I'm running openFOAM v9, and when i execute the command turbulenceFields (former R) to generate R folder, I cannot make it works.
Can someone help me?
Below i reported the reports on my terminal after the execution of simpleFoam -postProcess -func turbulenceFields
Create time
Create mesh for time = 0
SIMPLE: No convergence criteria found
Time = 0
Reading field p
Reading field U
Reading/calculating face flux field phi
Selecting incompressible transport model Newtonian
Selecting turbulence model type RAS
Selecting RAS turbulence model kEpsilon
RAS
{
model kEpsilon;
turbulence on;
printCoeffs on;
Cmu 0.09;
C1 1.44;
C2 1.92;
C3 0;
sigmak 1;
sigmaEps 1.3;
}
No MRF models present
No fvModels present
No fvConstraints present
turbulenceFields turbulenceFields: no fields requested to be stored
End
It was discontinued as far as I know.
@@OpenFOAMJozsefNagy there are other commads of text files I have to add to avoid this interruption without results?
I have inserted the file turbulenceProperties in /constant but it seams not effective.
@@ale.manzo.95 Sorry, but I am not familiar about the recent changes in v9. Maybe check the function object folder for more info.
maybe you could try with the command "simpleFoam -postProcess -func R"
How do you solve the problem? I met a similar problem when I use v10.
When i put the kEpsilon model to run, the simulation only runs till the 1710th iteration with initial residuals on the order of 1e-1. I can't find where am i messing this up. Any guess?
Check my answer below the other video.
It seems that is actually the other way around. But anyways, the whole things was really my bad,i set the relaxation factor for the field "p" as 0.03, when it maybe was originally set as 0.3 as you teach on the video. But even though i'm getting 2000 iterations, the initial residual for Uy is on the order of 1e-3 and the number of iterations vary from 3 to 8. Finally, i want to know if you have a suggestion of course, but also an explanation on why that happens. And thank u so much for the replies. @@OpenFOAMJozsefNagy
Actually, for the sake of the community who watch these videos, i should come back and say that the actual reason why the number of iterations was beeing limited was due to a line right below the "nNonOrthogonalCorrectors" in SIMPLE, "consistent yes;". Apparently , when you delete the line, which is found, i guess, in the latest versions of OpenFoam the simulation runs completely till the end. Nevertheless, the problems with the number of iterations still exist. Now, when i'd said that the problem for the number of iterations was the relaxation factor of the field "p", i was running the simulation on my home computer, which i had already deleted the "consistent yes;" line and had only to fix the relaxation factor value from "p". In this case, the number of iterations was going merely till the 51 iteration. It was only for the case where i had the relaxation factor already properly set and the mysterious line "consistent yes;" added that the number of iterations was going to the 1710th. I must say that all this is too confusing to me. Could you help me with this?? I still don't know why the number of iterations are that high and why the initial residualfor Uy is on the order of 1e-3. @@OpenFOAMJozsefNagy
@@Dmplivemail Thank you for your input on the consistent setting!
Jozsef, thank you for your nice work, which is informable. By the way, when I try to run a similar case with LRR using v8, an error"attempted assigment to self" poped up. Do you know what on earth it is?
The syntax changed over the years, so LRR does not work.
hi Jozsef, my velocity at the inlet in paraview is showing to be 2.98 and not 7.33 and even my results also differ than yours too so what should I do?
That is difficult to say from a distance. At which point do you have 2.98? On the patch itself or in the cells?
Hi,
I'm a student writing my bachelor thesis with OpenFoam. I'd like to do your backstep simulation with the rhoSimpleFoam solver(compressible flow). Unfortunatelly, I couldn't find any valid ressources and the case tutorial is unhelpful. What would you recommend?
+Vlad Chalapco Do this tutorial, do a rhoSimpleFoam tutorial and change the dictionaries from this one to the needed ones for rhoSimpleFoam. Step by step.
Thanks
Can i ask why my initial residual is around 1e-03 or 1e-04 which is much higher than urs in the tutorial? Will it has influence on the further result?
one more question:when i run the commond: simpleFoam -postProcess -func yPlus, it only shows the result of TIme=0. what's wrong with that?
@@chensongtao9244 Might be the difference in OpenFOAM version. I do not know, why you only get the results for the first time. Maybe you ran the simulation in parallel?
Hello Mr. József, great tutorials!
This is the first time I got stuck and i need to ask this question. I used this command to create R values:
-simpleFoam -postProcess -func R
I got turbulenceProperties:R file in my 0 folder but instead of getting:
internalField nonuniform List
Yes, this changed in recent versions.
Hi Josef!
Thank you for this priceless tutorials!
I have a question for you: Can wall funcionts be used for rotating walls?
Rubén B. Hi! Thank you. I have never tried it before.
József Nagy I cannot stop watching your videos József :-) Thank you for this effortJust another simply question. You use the same grid for kepsilon and komega model. I though they required different meshes as the don't use the same wall functions, and they don't need the same y+ don't they?
I am a newbie to OpenFOAM
Rubén B. Don't forget, that this is a tutorial and not a scientific work. You should be able to complete it within a reasonable time. Of course you have to think about the mesh. And thanks for your kind words!
Your tutorial videos have taught me a lot about OpenFOAM. Since this is my first time dealing with CFD, it is very useful.
Can I seek your assistance in the "homework" that you mention in the video though?
1) My idea for making nut = 1e-3 for the k-omega model is to use nut as 1e-3 and reverse enginer to get the omega value. Is this acceptable?
2) For LRR model, any hints on how to change or correct the nut to 1e-3?
1) My idea for making nut = 1e-3 for the k-omega model is to use nut as 1e-3 and reverse enginer to get the omega value. Is this acceptable? Try it.
2) For LRR model, any hints on how to change or correct the nut to 1e-3? I know, this is rather difficult. Try with upwind, use low underrelaxation factors and really wait for the settings to be convergent, then change the discretization.
in the version 2012
at time = 11 i am getting error
Time = 11
smoothSolver: Solving for Ux, Initial residual = 1, Final residual = 7.34729e-07, No Iterations 39
smoothSolver: Solving for Uy, Initial residual = 1, Final residual = 7.34771e-07, No Iterations 39
#0 Foam::error::printStack(Foam::Ostream&) at ??:?
#1 Foam::sigFpe::sigHandler(int) at ??:?
#2 ? in /lib/x86_64-linux-gnu/libpthread.so.0
#3 Foam::scalarProduct::type Foam::sumProd(Foam::UList const&, Foam::UList const&) at ??:?
#4 Foam::PCG::scalarSolve(Foam::Field&, Foam::Field const&, unsigned char) const at ??:?
#5 Foam::GAMGSolver::solveCoarsestLevel(Foam::Field&, Foam::Field const&) const at ??:?
#6 Foam::GAMGSolver::Vcycle(Foam::PtrList const&, Foam::Field&, Foam::Field const&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::PtrList&, Foam::PtrList&, unsigned char) const at ??:?
#7 Foam::GAMGSolver::solve(Foam::Field&, Foam::Field const&, unsigned char) const at ??:?
#8 Foam::fvMatrix::solveSegregated(Foam::dictionary const&) at ??:?
#9 Foam::fvMatrix::solveSegregatedOrCoupled(Foam::dictionary const&) at ??:?
#10 Foam::fvMesh::solve(Foam::fvMatrix&, Foam::dictionary const&) const at ??:?
#11 ? in /usr/lib/openfoam/openfoam2012/platforms/linux64GccDPInt32Opt/bin/simpleFoam
#12 __libc_start_main in /lib/x86_64-linux-gnu/libc.so.6
#13 ? in /usr/lib/openfoam/openfoam2012/platforms/linux64GccDPInt32Opt/bin/simpleFoam
Floating point exception (core dumped)
Thank you for the information. I just tested this tutorial with v2012 and it nicely converges for me. There may be a case setup problem on your end. Maybe you forgot something?
Hi! I just had a doubt! Is it sufficient for the average yplus to between 5 and 30 or should it be the minimum yplus?
Well this is a tutorial, it should give you a first idea about the workflow of OpenFOAM. For an actual simulation you should by between 5 and 30, however for a complicated geometry this will be a challenge. The question is always, how big your error is, which you create by being outside of this interval.
Got it! Thank you very much for your tutorials ! :-)
Hi, Mr. Nagy, thank you very much for your lectures on CFD , they have helped me a lot on CFD.
Nagy, if you could, I have a doubt about this lecture of yours. At the moment of the first simulation using simpleFoam, this message appeared in my terminal:
"Time = 28
smoothSolver: Solving for Ux, Initial residual = 0.578291, Final residual = 1.55951e+24, No Iterations 1000
smoothSolver: Solving for Uy, Initial residual = 0.578291, Final residual = 1.03625e+25, No Iterations 1000
#0 Foam::error::printStack(Foam::Ostream&) at ??:?
#1 Foam::sigFpe::sigHandler(int) at ??:?
#2 ? in "/lib/x86_64-linux-gnu/libc.so.6"
#3 double Foam::sumProd(Foam::UList const&, Foam::UList const&) at ??:?
#4 Foam::PCG::solve(Foam::Field&, Foam::Field const&, unsigned char) const at ??:?
#5 Foam::GAMGSolver::solveCoarsestLevel(Foam::Field&, Foam::Field const&) const at ??:?
#6 Foam::GAMGSolver::Vcycle(Foam::PtrList const&, Foam::Field&, Foam::Field const&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::PtrList&, Foam::PtrList&, unsigned char) const at ??:?
#7 Foam::GAMGSolver::solve(Foam::Field&, Foam::Field const&, unsigned char) const at ??:?
#8 Foam::fvMatrix::solveSegregated(Foam::dictionary const&) at ??:?
#9 Foam::fvMatrix::solve(Foam::dictionary const&) in "/opt/openfoam6/platforms/linux64GccDPInt32Opt/bin/simpleFoam"
#10 Foam::fvMatrix::solve() in "/opt/openfoam6/platforms/linux64GccDPInt32Opt/bin/simpleFoam"
#11 ? in "/opt/openfoam6/platforms/linux64GccDPInt32Opt/bin/simpleFoam"
#12 __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6"
#13 ? in "/opt/openfoam6/platforms/linux64GccDPInt32Opt/bin/simpleFoam"
Exceção de ponto flutuante (imagem do núcleo gravada) // in English, Floating Point exception (core dumped) "
However, reading some comments, I have changed the value of k from 0.2 to 0.3, and the simulation has worked well. I still do not know the reason of it happens I have followed your steps to do the simulation, such as setting U = 7.333.
I use openFoam version 6.
Mr. Nagy, would you have any suggestion about it?
Thank you a lot!
Stability changed over the years. You might have dome issues in newer versions. Underrelaxation might improve the problem.
I'm using version 9, and had the same error. I'd to change k to 0.4 and also change the tolerance to 10^-4.
Thanks Mr.Jozsef Nagy !! Priceless Tutorials !! I am getting started with OpenFOAM !! Have one question though !! Are quality of OpenFOAM results on par with commercial codes !! Thanks again !! Can u just improve the fonts size on screen.. or color also!! White would be real good
If it is comparable... If you ask me or other scientists/engineers working with OpenFOAM, we would say yes. If you ask someone using a commercial tool, they might give you a different answer. I think it depends on the problem. I would say yes, because in OpenFOAM, you can implement everything, if it is missing from the official release. You just have to take your time.
Thanks
@vijaya I guess you've gotten into OpenFOAM! I see you now have tutorials of your own. Congrats!
Whenever i try to execute R as per the video, I end up with this message
""R has been superseded by the '-postProcess' solver command-line option, e.g.
simpleFoam -postProcess -func R""
ANYONE with a solution!!
Yes, R was the way to do it in version 2.3.0. Now you use the suggested postProcess command.
Thanks for the response sir,
But I am currently using openFOAM V5 , so what should be the postProcess command (did u mean simpleFOAM?)
simpleFoam -postProcess -func R
Hi~ Thanks to your video. It's really helpful to me.
by the way, I got a error message when I followed this video. can you let me know the reason?
Please see below message.
boundaryField FrontAndBack not found in object U at time = 0
actually, I got this message for all case such as U, p, epsilon, v2 etc. I think I need to put boundary condition for each factor.
you used an uppercase F instead of a lower case f
I documented my reading of your tutorial at www.cfd-online.com/Forums/openfoam-solving/227418-simplefoam-tutorial-pitzdaily-using-reynolds-stress-tensor-lrr-rasmodel.html . I hope this contributes to your work. Thank you.
Time = 18
smoothSolver: Solving for Ux, Initial residual = 4.12274e-06, Final residual = 6.37679e-07, No Iterations 1
smoothSolver: Solving for Uy, Initial residual = 1.11335e-08, Final residual = 1.11335e-08, No Iterations 0
GAMG: Solving for p, Initial residual = 5.32991e-12, Final residual = 5.32991e-12, No Iterations 0
time step continuity errors : sum local = 1.21668e+20, global = -8.72718e+14, cumulative = -2.99379e+15
smoothSolver: Solving for epsilon, Initial residual = 4.84391e-14, Final residual = 4.84391e-14, No Iterations 0
smoothSolver: Solving for k, Initial residual = 3.24229e-06, Final residual = 8.00548e-07, No Iterations 4
bounding k, min: -18405.2 max: 1.48162e+39 average: 1.84105e+36
ExecutionTime = 0.6 s ClockTime = 1 s
Time = 19
smoothSolver: Solving for Ux, Initial residual = 0.610723, Final residual = 9.80804e-07, No Iterations 31
smoothSolver: Solving for Uy, Initial residual = 0.196477, Final residual = 7.73829e-07, No Iterations 26
#0 Foam::error::printStack(Foam::Ostream&) at ??:?
#1 Foam::sigFpe::sigHandler(int) at ??:?
#2 ? in "/lib/x86_64-linux-gnu/libc.so.6"
#3 Foam::GAMGSolver::scale(Foam::Field&, Foam::Field&, Foam::lduMatrix const&, Foam::FieldField const&, Foam::UPtrList const&, Foam::Field const&, unsigned char) const at ??:?
#4 Foam::GAMGSolver::Vcycle(Foam::PtrList const&, Foam::Field&, Foam::Field const&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::Field&, Foam::PtrList&, Foam::PtrList&, unsigned char) const at ??:?
#5 Foam::GAMGSolver::solve(Foam::Field&, Foam::Field const&, unsigned char) const at ??:?
#6 Foam::fvMatrix::solveSegregated(Foam::dictionary const&) at ??:?
#7 Foam::fvMatrix::solve(Foam::dictionary const&) in "/opt/openfoam7/platforms/linux64GccDPInt32Opt/bin/simpleFoam"
#8 Foam::fvMatrix::solve() in "/opt/openfoam7/platforms/linux64GccDPInt32Opt/bin/simpleFoam"
#9 ? in "/opt/openfoam7/platforms/linux64GccDPInt32Opt/bin/simpleFoam"
#10 __libc_start_main in "/lib/x86_64-linux-gnu/libc.so.6"
#11 ? in "/opt/openfoam7/platforms/linux64GccDPInt32Opt/bin/simpleFoam"
Floating point exception (core dumped)
hello sir , what should i do in this stua.?=
hello, Jozsef. Your videos really have helped me a lot when I turned to OpenFOAM. However, recently, I encountered another problem which puzzled me a lot. in the same case, with same mesh, same discretization schemes, with kEpsilon family models, it works well(mean converge well), however, with kOmegaSST, it always diverges. I have posted my problem in CFDOnline:www.cfd-online.com/Forums/openfoam-solving/160957-sonicfoam-komegasst-cannot-converge.html#post568569 Would you please have a look at it when you are avaliable? Thanks a lot in advance.
+Shiqi Wang Hey, if you are using sonicFoam, you are solving a transient problem. I don't understand your definition of convergence. Do you mean, that the flow converges with time towards a stationary solution?
hi, my case is definitely a transient problem. Cannot converge means that the calculation is crashed after a number of iterations.
Usually in this case you get a limitation of one of the turbulent quantities, or maybe both. If this is the case, try and rethink the (inlet) bounadry conditions and the value on that boundary for that quantity. Hope this helps.
+József Nagy I am trying to modify the wall conditions first. Is the choice between lowRe and HighRe wall condition for k and omega very important?
I would say as a first try not, but if you want to go for detailed physcally correct results yes.