
[Sponsors] 
October 31, 2012, 06:49 
pisoFoam with kepsilon turb blows up  Some questions

#1 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Hello,
I am fairly inexperienced when it comes to OpenFOAM and CFD in general. Slowly I am getting more familiar with the structure of OpenFOAM and I know the basics of CFD, but so far I have not been able to successfully apply my current knowledge and get my case running. To give you an idea about the case I am currently working on: in its most simple form, I am trying to simulate the flow through a tilted pipe in freestream. Flow enters the domain in +x direction from 2 different inlets with known velocity. The flow can leave the domain through my pipe (in x/+y direction) with a fixed velocity or simply by flowing out of the domain (p = 0). Now there are a few problems: 1) Finding the right conditions for p for the fixed velocity outflow seems to be very hard. I cannot apply zeroGradient, but p = 0 does not seem like a good solution either (due to 0 internalField and p = 0 for the freestream outlet). I am not really familiar with other possibilities. 2) I have given my 0/U file an internalField value of (v 0 0), but of course this also gives the fluid initially inside the pipe that same velocity. This velocity is obviously equal to my inlet velocities. 3) For the kepsilon model I am using l = 0.07*L where L is my pipe diameter or the hydraulic diameter of my inlets. The domain is bounded by slipwalls and a symmetry plane perpendicular to my zaxis causing my inlets and outlets to be half their actual surface. If for instance I have an inlet of 2 by 10 meters (yz plane), should I use a width of 10 meters or of 20? The same goes for the pipe. Due to the symmetry plane its surface is half a circle, should I use my actual diameter or something else? 4) I read a lot about using upwind schemes for k and epsilon. I have not tried this yet as I am a little afraid to apply changes in my fvSchemes and fvSolution without knowing why I would have to do so. 5) What should I do with my turbulent viscosity field 0/nut and with my Reynolds stress 0/R? Right now I have them "calculated" for in&outflows, *wallFunction for my walls and symmetryPlane for my symmetry plane. When I start running a simulation, my Co number seems to be very constant for the first few timesteps. All values actually do converge. But then suddenly my k, epsilon and Co blow up and U and p also start diverging again. When I notice this, I usually stop my simulation, but I've had it crashing a few times as well. My mesh consists of about 4.5M cells. A summary of checkMesh allGeometry allTopology is included as an attachment. I have also included my current fvSchemes and fvSolution. Some input and advice would be greatly appreciated. I would love to get more familiar with OpenFOAM without having to sacrifice hours of waiting for my simulation to fail Thanks! edit: I thought I mentioned it, but it seems like I did not. The mesh I am working with has been used for simulations in Ansys CFX with success. The mesh is not created by me, nor are the Ansys CFX simulations. I just need to get it up and running in OpenFOAM  with no luck so far... Last edited by Heroic; November 7, 2012 at 05:09. 

November 1, 2012, 04:37 

#2 
Senior Member

Dear Heroic,
Probably this line in your checkMesh report is the most important: Overall domain bounding box (6535.35 380 2.98023e05) (16000 8077.06 7933) It says your domain is about 22.5 km in xdirection and similarly scaled (kilometers, not meters) in the other directions. this can be fixed using: Code:
transformPoints scale "(0.001 0.001 0.001)" Good luck, regards, Tom 

November 5, 2012, 06:46 

#3 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Hi Tom,
Thank you for your response. Since I am not too familiar with meshing and the mesh worked before, I assumed everything would be correct. I guess that is not the way to go First I coarsened a part of the mesh and I fixed a gap that should not be there. The aspect ratio "problem" seems to be gone, but the number of low determinant cells increased. Right now I am running a new simulation (with the mesh rescaled of course) to see if the boundary conditions are set correctly. Answers on my other questions are still welcome though! edit: I am getting Floating Point exception errors very quickly now. Looking at similar topics, I really suspect this has something to do with either the way my fvSchemes and fvSolution are set up or with my boundary conditions (especially p, k and epsilon, as also stated in the first post questions). It must be added that I decided to change internalField in 0/U to (0 0 0) instead of the earlier mentioned (v 0 0). Code:
[3] #0 Foam::error::printStack(Foam::Ostream&)[1] #0 Foam::error::printStack(Foam::Ostream&) in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [3] #1 Foam::sigFpe::sigHandler(int) in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [3] #2 in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [1] #1 Foam::sigFpe::sigHandler(int) in "/lib/x86_64linuxgnu/libc.so.6" [3] #3 Foam::DILUPreconditioner::calcReciprocalD(Foam::Field<double>&, Foam::lduMatrix const&) in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [3] #4 Foam::DILUPreconditioner::DILUPreconditioner(Foam::lduMatrix::solver const&, Foam::dictionary const&) in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [1] #2 in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [3] #5 Foam::lduMatrix::preconditioner::addasymMatrixConstructorToTable<Foam::DILUPreconditioner>::New(Foam::lduMatrix::solver const&, Foam::dictionary const&) in "/lib/x86_64linuxgnu/libc.so.6" [1] #3 Foam::DILUPreconditioner::calcReciprocalD(Foam::Field<double>&, Foam::lduMatrix const&) in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [3] #6 Foam::lduMatrix::preconditioner::New(Foam::lduMatrix::solver const&, Foam::dictionary const&) in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [1] #4 Foam::DILUPreconditioner::DILUPreconditioner(Foam::lduMatrix::solver const&, Foam::dictionary const&) in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [3] #7 Foam::PBiCG::solve(Foam::Field<double>&, Foam::Field<double> const&, unsigned char) const in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [1] #5 Foam::lduMatrix::preconditioner::addasymMatrixConstructorToTable<Foam::DILUPreconditioner>::New(Foam::lduMatrix::solver const&, Foam::dictionary const&) in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [3] #8 in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [1] #6 Foam::lduMatrix::preconditioner::New(Foam::lduMatrix::solver const&, Foam::dictionary const&) in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [1] #7 Foam::PBiCG::solve(Foam::Field<double>&, Foam::Field<double> const&, unsigned char) const[3] in "/opt/openfoam211/platforms/linux64GccDPOpt/bin/pisoFoam" [3] #9 in "/opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so" [1] #8 [3] in "/opt/openfoam211/platforms/linux64GccDPOpt/bin/pisoFoam" [3] #10 [1] in "/opt/openfoam211/platforms/linux64GccDPOpt/bin/pisoFoam" [1] #9 [3] in "/opt/openfoam211/platforms/linux64GccDPOpt/bin/pisoFoam" [3] #11 __libc_start_main in "/lib/x86_64linuxgnu/libc.so.6" [3] #12 [1] in "/opt/openfoam211/platforms/linux64GccDPOpt/bin/pisoFoam" [1] #10 [3] in "/opt/openfoam211/platforms/linux64GccDPOpt/bin/pisoFoam" [HPxw8600Workstation:29373] *** Process received signal *** [HPxw8600Workstation:29373] Signal: Floating point exception (8) [HPxw8600Workstation:29373] Signal code: (6) [HPxw8600Workstation:29373] Failing at address: 0x3ec000072bd [HPxw8600Workstation:29373] [ 0] /lib/x86_64linuxgnu/libc.so.6(+0x364a0) [0x7f7709ed54a0] [HPxw8600Workstation:29373] [ 1] /lib/x86_64linuxgnu/libc.so.6(gsignal+0x35) [0x7f7709ed5425] [HPxw8600Workstation:29373] [ 2] /lib/x86_64linuxgnu/libc.so.6(+0x364a0) [0x7f7709ed54a0] [HPxw8600Workstation:29373] [ 3] /opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so(_ZN4Foam18DILUPreconditioner15calcReciprocalDERNS_5FieldIdEERKNS_9lduMatrixE+0xac) [0x7f770af1131c] [HPxw8600Workstation:29373] [ 4] /opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so(_ZN4Foam18DILUPreconditionerC2ERKNS_9lduMatrix6solverERKNS_10dictionaryE+0x8e) [0x7f770af1148e] [HPxw8600Workstation:29373] [ 5] /opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so(_ZN4Foam9lduMatrix14preconditioner31addasymMatrixConstructorToTableINS_18DILUPreconditionerEE3NewERKNS0_6solverERKNS_10dictionaryE+0x3c) [0x7f770af116cc] [HPxw8600Workstation:29373] [ 6] /opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so(_ZN4Foam9lduMatrix14preconditioner3NewERKNS0_6solverERKNS_10dictionaryE+0x2b2) [0x7f770af05752] [HPxw8600Workstation:29373] [ 7] /opt/openfoam211/platforms/linux64GccDPOpt/lib/libOpenFOAM.so(_ZNK4Foam5PBiCG5solveERNS_5FieldIdEERKS2_h+0x6ba) [0x7f770af09b6a] [HPxw8600Workstation:29373] [ 8] pisoFoam() [0x433e03] [HPxw8600Workstation:29373] [ 9] pisoFoam() [0x4346eb] [HPxw8600Workstation:29373] [10] pisoFoam() [0x4184fd] [HPxw8600Workstation:29373] [11] /lib/x86_64linuxgnu/libc.so.6(__libc_start_main+0xed) [0x7f7709ec076d] [HPxw8600Workstation:29373] [12] pisoFoam() [0x418f5d] [HPxw8600Workstation:29373] *** End of error message ***  mpirun noticed that process rank 3 with PID 29373 on node HPxw8600Workstation exited on signal 8 (Floating point exception).  Code:
Starting time loop Time = 0.001 Courant Number mean: 2.52915e06 max: 0.387879 DILUPBiCG: Solving for Ux, Initial residual = 1, Final residual = 4.52238e05, No Iterations 1 DILUPBiCG: Solving for Uy, Initial residual = 1, Final residual = 9.27004e05, No Iterations 1 DILUPBiCG: Solving for Uz, Initial residual = 1, Final residual = 0.000369142, No Iterations 1 DICPCG: Solving for p, Initial residual = 1, Final residual = 0.00998977, No Iterations 602 DICPCG: Solving for p, Initial residual = 0.0565943, Final residual = 0.000530243, No Iterations 9 DICPCG: Solving for p, Initial residual = 0.0067374, Final residual = 6.72031e05, No Iterations 42 time step continuity errors : sum local = 1.23997e07, global = 1.65786e08, cumulative = 1.65786e08 DICPCG: Solving for p, Initial residual = 0.00123306, Final residual = 1.22963e05, No Iterations 418 DICPCG: Solving for p, Initial residual = 0.000477002, Final residual = 4.57706e06, No Iterations 56 DICPCG: Solving for p, Initial residual = 9.89045e05, Final residual = 9.67021e07, No Iterations 153 time step continuity errors : sum local = 1.76709e09, global = 4.42326e11, cumulative = 1.65344e08 DICPCG: Solving for p, Initial residual = 3.6719e05, Final residual = 3.63898e07, No Iterations 406 DICPCG: Solving for p, Initial residual = 1.81985e05, Final residual = 1.77353e07, No Iterations 36 DICPCG: Solving for p, Initial residual = 3.90771e06, Final residual = 9.89339e08, No Iterations 20 time step continuity errors : sum local = 1.80812e10, global = 2.50291e11, cumulative = 1.65094e08 DILUPBiCG: Solving for epsilon, Initial residual = 0.000234569, Final residual = 1.67314e06, No Iterations 1 bounding epsilon, min: 0.000497126 max: 0.0211597 average: 0.000419256 DILUPBiCG: Solving for k, Initial residual = 1, Final residual = 0.00336708, No Iterations 1 ExecutionTime = 128.41 s ClockTime = 135 s Time = 0.002 Courant Number mean: 0.000943259 max: 0.86726 DILUPBiCG: Solving for Ux, Initial residual = 0.446366, Final residual = 3.52989e05, No Iterations 1 DILUPBiCG: Solving for Uy, Initial residual = 0.134078, Final residual = 9.05337e05, No Iterations 1 DILUPBiCG: Solving for Uz, Initial residual = 0.228996, Final residual = 0.000167918, No Iterations 1 DICPCG: Solving for p, Initial residual = 0.01609, Final residual = 0.000158648, No Iterations 545 DICPCG: Solving for p, Initial residual = 0.52777, Final residual = 0.00498079, No Iterations 9 DICPCG: Solving for p, Initial residual = 0.120384, Final residual = 0.00115452, No Iterations 117 time step continuity errors : sum local = 1.22032e07, global = 9.93291e09, cumulative = 6.57644e09 DICPCG: Solving for p, Initial residual = 0.0264649, Final residual = 0.000261371, No Iterations 480 DICPCG: Solving for p, Initial residual = 0.0142866, Final residual = 0.000142056, No Iterations 64 DICPCG: Solving for p, Initial residual = 0.00290025, Final residual = 2.85768e05, No Iterations 151 time step continuity errors : sum local = 2.10434e09, global = 2.99905e11, cumulative = 6.60643e09 DICPCG: Solving for p, Initial residual = 0.00191043, Final residual = 1.89451e05, No Iterations 166 DICPCG: Solving for p, Initial residual = 0.00116943, Final residual = 1.15869e05, No Iterations 13 DICPCG: Solving for p, Initial residual = 0.000204794, Final residual = 9.79251e08, No Iterations 539 time step continuity errors : sum local = 7.1902e12, global = 6.33508e14, cumulative = 6.6065e09 DILUPBiCG: Solving for epsilon, Initial residual = 0.00101244, Final residual = 7.70349e05, No Iterations 2 bounding epsilon, min: 0.0247528 max: 139.839 average: 0.00316543 DILUPBiCG: Solving for k, Initial residual = 0.0296, Final residual = 0.00284824, No Iterations 2 bounding k, min: 0.149225 max: 1350.8 average: 0.0611313 ExecutionTime = 295.28 s ClockTime = 305 s Time = 0.003 Courant Number mean: 0.000945477 max: 1.4992 DILUPBiCG: Solving for Ux, Initial residual = 0.870503, Final residual = 0.0242976, No Iterations 2 DILUPBiCG: Solving for Uy, Initial residual = 0.774113, Final residual = 0.0326547, No Iterations 2 DILUPBiCG: Solving for Uz, Initial residual = 0.890679, Final residual = 0.0312396, No Iterations 2 DICPCG: Solving for p, Initial residual = 0.252153, Final residual = 0.00249363, No Iterations 171 DICPCG: Solving for p, Initial residual = 0.668941, Final residual = 0.00666033, No Iterations 49 DICPCG: Solving for p, Initial residual = 0.287678, Final residual = 0.00283792, No Iterations 23 time step continuity errors : sum local = 0.00120113, global = 1.11481e09, cumulative = 7.72131e09 DICPCG: Solving for p, Initial residual = 0.111563, Final residual = 0.00110069, No Iterations 42 DICPCG: Solving for p, Initial residual = 0.0527845, Final residual = 0.000505932, No Iterations 43 DICPCG: Solving for p, Initial residual = 0.0163874, Final residual = 0.000163869, No Iterations 422 time step continuity errors : sum local = 8.59736e05, global = 4.76738e06, cumulative = 4.7751e06 DICPCG: Solving for p, Initial residual = 0.039144, Final residual = 0.000377305, No Iterations 67 DICPCG: Solving for p, Initial residual = 0.0335431, Final residual = 0.000323097, No Iterations 24 DICPCG: Solving for p, Initial residual = 0.0125174, Final residual = 9.63604e08, No Iterations 708 time step continuity errors : sum local = 4.82243e08, global = 5.39547e10, cumulative = 4.77564e06 DILUPBiCG: Solving for epsilon, Initial residual = 0.999999, Final residual = 5.60009, No Iterations 1001 bounding epsilon, min: 4.18298e+13 max: 1.75542e+13 average: 6.67502e+07 DILUPBiCG: Solving for k, Initial residual = 1, Final residual = 0.699745, No Iterations 1001 bounding k, min: 1.09626e+13 max: 1.67478e+13 average: 2.35452e+07 ExecutionTime = 771.48 s ClockTime = 787 s Time = 0.004 Courant Number mean: 0.343598 max: 9.96555e+07 Last edited by Heroic; November 5, 2012 at 09:50. 

November 6, 2012, 06:48 

#4 
Senior Member

Hello Heroic,
Been busy past few, days, but this line in your log may give you a first clue: Code:
Courant Number mean: 0.000945477 max: 1.4992 Your schemes and solution file seem ok, however it may be challenging with the mesh problems. maybe change the divScheme for the turbulent variables to upwind, especially near the start of the simulation. Regards, Tom 

November 7, 2012, 04:33 

#5 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Hi Tom,
I am aware of the suddenly increasing Courant number and that's pretty much the whole problem. I have been running some simulations using icoFoam (not realistic, but I thought it might give some insight in the p and U ICs) over the past few days with different timestep sizes, but every time it crashes really all of a sudden. The Courant number stays low (the log below is from a simulation where Co_max ranges between 0.1 and 0.2) and then it suddenly skyrockets) There are a few tiny areas where the cell sizes are very small and the velocities are very big. As we speak, I'm removing this danger zone in my mesh to see if that would solve the problem, but I doubt it. From the very small simulation time (see the partial log below) I could not find any weird behavior when it comes to the velocity of the flow. I also included a graphical view of the residuals. Please note that it is from the icoFoam simulation. PisoFoam messes up even faster! edit: I thought I mentioned it, but it seems like I did not. The mesh I am working with has been used for simulations in Ansys CFX with success. The mesh is not created by me, nor are the Ansys CFX simulations. I just need to get it up and running in OpenFOAM  with no luck so far... Code:
Time = 0.0259 Courant Number mean: 9.44589e05 max: 0.468547 DILUPBiCG: Solving for Ux, Initial residual = 3.58438e05, Final residual = 1.14984e08, No Iterations 1 DILUPBiCG: Solving for Uy, Initial residual = 5.40029e05, Final residual = 3.40818e07, No Iterations 1 DILUPBiCG: Solving for Uz, Initial residual = 9.90141e05, Final residual = 2.87787e07, No Iterations 1 DICPCG: Solving for p, Initial residual = 0.0368753, Final residual = 0.000303587, No Iterations 8 DICPCG: Solving for p, Initial residual = 0.00704458, Final residual = 6.96844e05, No Iterations 130 DICPCG: Solving for p, Initial residual = 0.0017646, Final residual = 1.72086e05, No Iterations 185 time step continuity errors : sum local = 1.61201e12, global = 2.19119e13, cumulative = 1.70655e10 DICPCG: Solving for p, Initial residual = 0.00160778, Final residual = 1.58644e05, No Iterations 187 DICPCG: Solving for p, Initial residual = 0.000779738, Final residual = 7.53291e06, No Iterations 78 DICPCG: Solving for p, Initial residual = 0.000158196, Final residual = 1.56612e06, No Iterations 345 time step continuity errors : sum local = 1.47087e13, global = 2.64781e14, cumulative = 1.70681e10 ExecutionTime = 18994.3 s ClockTime = 24337 s Time = 0.026 Courant Number mean: 9.44625e05 max: 0.567149 DILUPBiCG: Solving for Ux, Initial residual = 3.69846e05, Final residual = 1.64708e08, No Iterations 1 DILUPBiCG: Solving for Uy, Initial residual = 5.56901e05, Final residual = 3.84811e10, No Iterations 2 DILUPBiCG: Solving for Uz, Initial residual = 0.000102508, Final residual = 3.98556e07, No Iterations 1 DICPCG: Solving for p, Initial residual = 0.0467224, Final residual = 0.000439405, No Iterations 8 DICPCG: Solving for p, Initial residual = 0.00877503, Final residual = 8.00621e05, No Iterations 145 DICPCG: Solving for p, Initial residual = 0.00234424, Final residual = 2.21612e05, No Iterations 339 time step continuity errors : sum local = 2.27776e12, global = 2.71224e13, cumulative = 1.7041e10 DICPCG: Solving for p, Initial residual = 0.00243512, Final residual = 2.40603e05, No Iterations 158 DICPCG: Solving for p, Initial residual = 0.00114083, Final residual = 1.11688e05, No Iterations 10 DICPCG: Solving for p, Initial residual = 0.000197753, Final residual = 1.97666e06, No Iterations 470 time step continuity errors : sum local = 2.06328e13, global = 9.89029e16, cumulative = 1.70411e10 ExecutionTime = 19074.6 s ClockTime = 24500 s Time = 0.0261 Courant Number mean: 9.44675e05 max: 0.705859 DILUPBiCG: Solving for Ux, Initial residual = 3.88916e05, Final residual = 3.15394e08, No Iterations 1 DILUPBiCG: Solving for Uy, Initial residual = 5.8256e05, Final residual = 6.794e10, No Iterations 2 DILUPBiCG: Solving for Uz, Initial residual = 0.000107994, Final residual = 5.74618e07, No Iterations 1 DICPCG: Solving for p, Initial residual = 0.0623028, Final residual = 0.000542297, No Iterations 9 DICPCG: Solving for p, Initial residual = 0.0116734, Final residual = 0.00011547, No Iterations 141 DICPCG: Solving for p, Initial residual = 0.00307717, Final residual = 3.06705e05, No Iterations 338 time step continuity errors : sum local = 3.58674e12, global = 3.63346e13, cumulative = 1.70047e10 DICPCG: Solving for p, Initial residual = 0.00361425, Final residual = 3.59638e05, No Iterations 158 DICPCG: Solving for p, Initial residual = 0.00175589, Final residual = 1.73841e05, No Iterations 9 DICPCG: Solving for p, Initial residual = 0.000305585, Final residual = 2.95085e06, No Iterations 475 time step continuity errors : sum local = 3.51875e13, global = 1.93796e15, cumulative = 1.70049e10 ExecutionTime = 19148.8 s ClockTime = 24651 s Time = 0.0262 Courant Number mean: 9.44746e05 max: 0.914305 DILUPBiCG: Solving for Ux, Initial residual = 4.22995e05, Final residual = 8.03984e08, No Iterations 1 DILUPBiCG: Solving for Uy, Initial residual = 6.2494e05, Final residual = 3.19853e07, No Iterations 1 DILUPBiCG: Solving for Uz, Initial residual = 0.000116662, Final residual = 8.73911e07, No Iterations 1 DICPCG: Solving for p, Initial residual = 0.0884359, Final residual = 0.000741313, No Iterations 9 DICPCG: Solving for p, Initial residual = 0.0163983, Final residual = 0.000163449, No Iterations 89 DICPCG: Solving for p, Initial residual = 0.00398394, Final residual = 3.80192e05, No Iterations 160 time step continuity errors : sum local = 5.36471e12, global = 5.80222e13, cumulative = 1.69469e10 DICPCG: Solving for p, Initial residual = 0.00559329, Final residual = 5.41431e05, No Iterations 158 DICPCG: Solving for p, Initial residual = 0.00283293, Final residual = 2.63392e05, No Iterations 13 DICPCG: Solving for p, Initial residual = 0.000508933, Final residual = 5.0394e06, No Iterations 183 time step continuity errors : sum local = 7.11153e13, global = 1.09061e14, cumulative = 1.69458e10 ExecutionTime = 19191 s ClockTime = 24736 s Time = 0.0263 Courant Number mean: 9.44858e05 max: 1.25322 DILUPBiCG: Solving for Ux, Initial residual = 4.92494e05, Final residual = 2.87712e07, No Iterations 1 DILUPBiCG: Solving for Uy, Initial residual = 7.03779e05, Final residual = 1.42621e07, No Iterations 1 DILUPBiCG: Solving for Uz, Initial residual = 0.00013109, Final residual = 1.45978e09, No Iterations 2 DICPCG: Solving for p, Initial residual = 0.13273, Final residual = 0.00105482, No Iterations 8 DICPCG: Solving for p, Initial residual = 0.0252117, Final residual = 0.000248345, No Iterations 9 DICPCG: Solving for p, Initial residual = 0.00455824, Final residual = 4.50067e05, No Iterations 132 time step continuity errors : sum local = 8.47003e12, global = 5.34656e13, cumulative = 1.69993e10 DICPCG: Solving for p, Initial residual = 0.00987514, Final residual = 9.77687e05, No Iterations 154 DICPCG: Solving for p, Initial residual = 0.00460621, Final residual = 3.75006e05, No Iterations 13 DICPCG: Solving for p, Initial residual = 0.000840936, Final residual = 8.21884e06, No Iterations 472 time step continuity errors : sum local = 1.52967e12, global = 5.13435e14, cumulative = 1.70044e10 ExecutionTime = 19247.8 s ClockTime = 24852 s Time = 0.0264 Courant Number mean: 9.45054e05 max: 1.8557 DILUPBiCG: Solving for Ux, Initial residual = 6.73309e05, Final residual = 5.8597e08, No Iterations 2 DILUPBiCG: Solving for Uy, Initial residual = 9.20102e05, Final residual = 8.2857e09, No Iterations 2 DILUPBiCG: Solving for Uz, Initial residual = 0.000164266, Final residual = 9.55881e09, No Iterations 2 DICPCG: Solving for p, Initial residual = 0.210654, Final residual = 0.00200213, No Iterations 53 DICPCG: Solving for p, Initial residual = 0.0594574, Final residual = 0.000571596, No Iterations 139 DICPCG: Solving for p, Initial residual = 0.0143112, Final residual = 0.000141844, No Iterations 355 time step continuity errors : sum local = 4.94039e11, global = 8.93856e12, cumulative = 1.78983e10 DICPCG: Solving for p, Initial residual = 0.0266814, Final residual = 0.000222965, No Iterations 12 DICPCG: Solving for p, Initial residual = 0.00560012, Final residual = 5.58345e05, No Iterations 82 DICPCG: Solving for p, Initial residual = 0.0014299, Final residual = 1.42898e05, No Iterations 176 time step continuity errors : sum local = 6.15084e12, global = 4.23144e13, cumulative = 1.79406e10 ExecutionTime = 19305.4 s ClockTime = 24969 s Time = 0.0265 Courant Number mean: 9.45492e05 max: 2.87359 DILUPBiCG: Solving for Ux, Initial residual = 0.000162167, Final residual = 1.10896e06, No Iterations 2 DILUPBiCG: Solving for Uy, Initial residual = 0.000297795, Final residual = 8.46494e07, No Iterations 2 DILUPBiCG: Solving for Uz, Initial residual = 0.000353394, Final residual = 1.62101e06, No Iterations 2 DICPCG: Solving for p, Initial residual = 0.840711, Final residual = 0.00681527, No Iterations 9 DICPCG: Solving for p, Initial residual = 0.342861, Final residual = 0.00331792, No Iterations 96 DICPCG: Solving for p, Initial residual = 0.204708, Final residual = 0.00200217, No Iterations 148 time step continuity errors : sum local = 1.17995e08, global = 1.23396e09, cumulative = 1.05455e09 DICPCG: Solving for p, Initial residual = 0.445196, Final residual = 0.00411219, No Iterations 153 DICPCG: Solving for p, Initial residual = 0.218903, Final residual = 0.00213873, No Iterations 80 DICPCG: Solving for p, Initial residual = 0.106291, Final residual = 0.00104989, No Iterations 143 time step continuity errors : sum local = 2.99229e08, global = 4.02546e10, cumulative = 1.4571e09 ExecutionTime = 19351.3 s ClockTime = 25061 s Time = 0.0266 Courant Number mean: 0.000119108 max: 1353.46 DILUPBiCG: Solving for Ux, Initial residual = 0.664363, Final residual = 0.00373814, No Iterations 11 DILUPBiCG: Solving for Uy, Initial residual = 0.789425, Final residual = 0.00126608, No Iterations 14 DILUPBiCG: Solving for Uz, Initial residual = 0.76283, Final residual = 0.00623744, No Iterations 11 DICPCG: Solving for p, Initial residual = 0.772438, Final residual = 0.0076202, No Iterations 213 DICPCG: Solving for p, Initial residual = 0.308032, Final residual = 0.00275654, No Iterations 16 DICPCG: Solving for p, Initial residual = 0.0732167, Final residual = 0.000669005, No Iterations 498 time step continuity errors : sum local = 5.42255e08, global = 1.44381e08, cumulative = 1.58952e08 DICPCG: Solving for p, Initial residual = 0.227006, Final residual = 0.00146796, No Iterations 53 DICPCG: Solving for p, Initial residual = 0.0640002, Final residual = 0.000543174, No Iterations 162 DICPCG: Solving for p, Initial residual = 0.0275658, Final residual = 0.000241115, No Iterations 218 time step continuity errors : sum local = 7.3091e08, global = 1.83035e08, cumulative = 3.41987e08 ExecutionTime = 19435.3 s ClockTime = 25231 s Time = 0.0267 Courant Number mean: 0.000215413 max: 1387.06 DILUPBiCG: Solving for Ux, Initial residual = 0.425437, Final residual = 0.00202656, No Iterations 6 DILUPBiCG: Solving for Uy, Initial residual = 0.631993, Final residual = 0.00102398, No Iterations 13 DILUPBiCG: Solving for Uz, Initial residual = 0.808149, Final residual = 0.00784158, No Iterations 4 DICPCG: Solving for p, Initial residual = 0.39313, Final residual = 0.00348448, No Iterations 324 DICPCG: Solving for p, Initial residual = 0.418668, Final residual = 0.00412241, No Iterations 5 DICPCG: Solving for p, Initial residual = 0.0721179, Final residual = 0.000681379, No Iterations 93 time step continuity errors : sum local = 1.41891e06, global = 4.70344e07, cumulative = 5.04543e07 DICPCG: Solving for p, Initial residual = 0.768661, Final residual = 0.00697122, No Iterations 14 DICPCG: Solving for p, Initial residual = 0.146318, Final residual = 0.00115033, No Iterations 184 DICPCG: Solving for p, Initial residual = 0.0575492, Final residual = 0.000504504, No Iterations 18 time step continuity errors : sum local = 5.74277e06, global = 2.65267e06, cumulative = 3.15721e06 ExecutionTime = 19484.5 s ClockTime = 25330 s Time = 0.0268 Courant Number mean: 0.00359019 max: 15825.2 DILUPBiCG: Solving for Ux, Initial residual = 0.998627, Final residual = 0.0043253, No Iterations 39 DILUPBiCG: Solving for Uy, Initial residual = 0.996133, Final residual = 0.00333772, No Iterations 31 DILUPBiCG: Solving for Uz, Initial residual = 0.994139, Final residual = 0.00726301, No Iterations 17 DICPCG: Solving for p, Initial residual = 0.887243, Final residual = 0.545828, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.208399, Final residual = 0.00196286, No Iterations 746 DICPCG: Solving for p, Initial residual = 0.132203, Final residual = 0.00402091, No Iterations 1001 time step continuity errors : sum local = 0.000880529, global = 0.000151345, cumulative = 0.000154502 DICPCG: Solving for p, Initial residual = 0.781695, Final residual = 0.00780975, No Iterations 139 DICPCG: Solving for p, Initial residual = 0.101759, Final residual = 0.00928835, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.0607304, Final residual = 0.00312274, No Iterations 1001 time step continuity errors : sum local = 0.0100497, global = 0.0016949, cumulative = 0.00154039 ExecutionTime = 19845.5 s ClockTime = 26057 s Time = 0.0269 Courant Number mean: 1.39003 max: 6.30005e+06 DILUPBiCG: Solving for Ux, Initial residual = 0.999623, Final residual = 0.431531, No Iterations 1001 DILUPBiCG: Solving for Uy, Initial residual = 0.99988, Final residual = 0.0924515, No Iterations 1001 DILUPBiCG: Solving for Uz, Initial residual = 0.999948, Final residual = 0.125014, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.999374, Final residual = 3.66073, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.840675, Final residual = 1.59847, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.82918, Final residual = 16.5741, No Iterations 1001 time step continuity errors : sum local = 1.09465e+06, global = 26.0149, cumulative = 26.0164 DICPCG: Solving for p, Initial residual = 0.971531, Final residual = 2.14693, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.867684, Final residual = 2.55421, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.800374, Final residual = 22.508, No Iterations 1001 time step continuity errors : sum local = 9.44426e+09, global = 17257.8, cumulative = 17283.8 ExecutionTime = 20737.5 s ClockTime = 27856 s Time = 0.027 Courant Number mean: 1.6153e+10 max: 1.3835e+19 DILUPBiCG: Solving for Ux, Initial residual = 0.999982, Final residual = 113.182, No Iterations 1001 DILUPBiCG: Solving for Uy, Initial residual = 0.999999, Final residual = 16.3833, No Iterations 1001 DILUPBiCG: Solving for Uz, Initial residual = 0.999998, Final residual = 17.1423, No Iterations 1001 DICPCG: Solving for p, Initial residual = 1, Final residual = 358965, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.153481, Final residual = 0.00100004, No Iterations 119 DICPCG: Solving for p, Initial residual = 0.00642703, Final residual = 0.00222331, No Iterations 1001 time step continuity errors : sum local = 3.17175e+34, global = 3.81634e+21, cumulative = 3.81634e+21 DICPCG: Solving for p, Initial residual = 0.000773474, Final residual = 0.00553084, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.014366, Final residual = 0.058737, No Iterations 1001 DICPCG: Solving for p, Initial residual = 0.000218239, Final residual = 0.000554938, No Iterations 1001 time step continuity errors : sum local = 9.60439e+40, global = 2.95678e+28, cumulative = 2.95678e+28 ExecutionTime = 21583.8 s ClockTime = 29557 s Time = 0.0271 Courant Number mean: 1.34852e+41 max: 5.43802e+49 Last edited by Heroic; November 7, 2012 at 05:12. 

November 8, 2012, 05:04 

#6 
Senior Member

Hi Heroic,
If the sudden increase in the courant number is not caused by the flow reaching some point where this could be explained I think that your settings should be okay from the time discretisation. As it is the pressure that seems to go first in your residuals, you may want to change to use the AMG solver. It could help. In fvSchemes use something like this: Code:
p { solver GAMG; tolerance 1e12; relTol 0.001; smoother GaussSeidel; cacheAgglomeration true; nCellsInCoarsestLevel 20; agglomerator faceAreaPair; mergeLevels 1; } pFinal { solver GAMG; tolerance 1e12; relTol 0; smoother GaussSeidel; cacheAgglomeration true; nCellsInCoarsestLevel 20; agglomerator faceAreaPair; mergeLevels 1; } Code:
laplacianSchemes { default Gauss linear limited 0.333;//corrected; // laplacian(nuEff,U) Gauss linear corrected; // laplacian((1A(U)),p) Gauss linear corrected; // laplacian(DkEff,k) Gauss linear corrected; // laplacian(DepsilonEff,epsilon) Gauss linear corrected; // laplacian(DREff,R) Gauss linear corrected; // laplacian(DnuTildaEff,nuTilda) Gauss linear corrected; } Code:
snGradSchemes { default limited 0.333;//corrected; } Regards, Tom 

November 9, 2012, 06:36 

#7 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Hi Tom,
Thanks again for your suggestions. I have now updated my fvSchemes and fvSolution files and a simulation is running as we speak. Running on the original mesh now, as it seems my coarsened mesh did not really do any good (it basically crashed within 5 timesteps with a Co going from 0.07 to 0.3 to 1.5 to 1000 towards infinity )... Could be my settings as well (I am running the simulation with my original solvers for the turbulent parameters). Simplifying and updating the mesh is harder than I expected, so I am now getting familiar with Ansys ICEM . I still assume the original mesh should be fine, as it has been able to run in Ansys CFX (or is that no guarantee?), but a working, more coarse mesh would definately speed things up for me, so that is what I am aiming for now. 

November 9, 2012, 07:50 

#8 
Senior Member

Hello Heroic,
Well Ansys CFX uses different numerics, so it is no guarantee that if it works in Ansys it will work in OpenFOAM, generally. If your Co number ramps up that fast, than I think your time step is too large, just set it 10 times as low and see if it still ramps up that fast, I am not sure it should. On the other hand, if it is speeding up so fast, just write data every time step and see if you can identify a reason why it blows up and determine whether it is physical or not. Regards, Tom 

November 9, 2012, 08:04 

#9 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Hi Tom,
That was exactly my plan. I have done it with the icoFoam case already, but nothing weird showed up in paraView. It simply went wrong from 1 timestep to another without any visibly weird behavior. I had to terminate my simulation for now, but I will let it run over the weekend using a very small timestep and a high amount of data to be saved just to be sure. Next Monday I will give an update on this. For now I will focus on my mesh and see if I can eliminate all low determinant areas. Thanks a lot for your input so far! 

November 12, 2012, 04:49 

#10 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
The results are in... I tried to simulate both with and without a turbulence model enabled.
With the kepsilon model included the simulation crashed within 10 timesteps (probably because of the solver not being set to an upwind scheme?). The simulation without any turbulence effects went on for a few hours before it crashed. I am still not entirely sure whether my BC/ICs are correct, but there seems to be one "line" (intersection between symmetry plane and outflow plane, see attachments for a small snapshot of the area in question) where the velocity gets really high (it should be O(1) or O(10), but it is O(100) over there) and the pressure goes negative. I do not know whether this happens because of my increasing Co number or that my Co number increases because there is something wrong near this spot. The mesh seems to be fine at that particular area, but I am going to have a good look and see if something could be changed over there and then I will try again. 

November 12, 2012, 09:07 

#11 
Member
Aathavan
Join Date: Nov 2012
Posts: 70
Rep Power: 4 
Dear Heroic,
I am also new to openFoam, however I want to ask you something. most probably simulation diverge because of improper mesh. second is I have seen your output it is showing epsilon bounding, so what value you are using for your kappa and epsilon. third is what is your y+ value for pisoFoam it should be 30 to 300. check the answer for the above question, post your answers here, then we can go further. thanks, Aadhavan 

November 13, 2012, 04:31 

#12 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Hi Aadhavan,
As I said in my previous post, I am slowly starting to think there may be some problems with my mesh as well. The area where it goes wrong is properly meshed though, so I have included all files from my 0 folder, including the k and epsilon ones. These are simply calculated in a similar fashion as has been done in the OpenFOAM manual (assuming U' to be 5% of U in all directions, using the 7% of the hydraulic diameters of my inlets and outlets as characteristic length). I am unsure where to find my y+ value, so I cannot really say anything about that. If 30 or 300 is a default value, it is probably what I am using as well, as I copied and modified the files from one of the tutorial folders. Kind regards. edit: I was also wondering whether it is possible to (visually) locate extreme values for p/U using paraView, even if it is one out of a million cells. The program says there are some values that are extremely off, but I cannot seem to find them. 

November 13, 2012, 06:56 

#13 
Member
Aathavan
Join Date: Nov 2012
Posts: 70
Rep Power: 4 
Dear Heroic,
If you are sure about your k and epsilon value, then the problem can be in your mesh. first calculate the y value to make sure the y+ value to be 30 300 you can have a look wiki for the calculation. In wiki you can get the detailed information about the "estimation of y+ value" Regarding your present y+ value just type the command yPlusRAS in terminal. if your y+ value also fine. you can send your whole case, I can have a look in my machine. Thanks and regards, Aadhavan 

November 13, 2012, 10:53 

#14 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Hi Aadhavan,
I have been reading a little about Y+ and I realized that I am in the low Re region for which I appear to need values of Y+ < 1 in case of a kepsilon model. For the laminar simulation I have been running (which also crashed), Y+ is not really relevant if I understood everything correctly. I ran the command you mentioned on my (crashed) turbulent pisoFoam and for the 0/yPlus it gave constant values of about 0.11. That should be correct for this case then, right? I do not think I am allowed to share/spread the mesh I am working with, so that is not going to work. Other than that, most of the relevant files are here already. Once things go as they should, my velocities should not be over ~10 m/s anywhere at all, but right now it mainly seems to go wrong where the pipe outflow intersects with the symmetry plane as you can see in the snapshot I have provided in a previous post. This is what mostly troubles me right now. The mesh at that particular area seems to be fine, so I am still in doubt whether the problem really is in the mesh or maybe in my 0/p or 0/U. Thank you for the input! Kind regards. edit: never mind about the low Re. That is not really the case for me. I will have a better look at the Y+ and how it is determined in the mesh I am working with. Maybe something needs to be altered, but if I "assume" laminar flow (e.g. I disable the turbulence model in pisoFoam) this value should not matter, right? Last edited by Heroic; November 14, 2012 at 04:11. 

November 14, 2012, 05:54 

#15 
Member
Aathavan
Join Date: Nov 2012
Posts: 70
Rep Power: 4 
Dear Heroic,
Fine, I had an experience problem in y+ value, that is why I was suggesting you to check that. Actually your y+ value will increase or decrease according to your velocity. Please check your k and epsilon 0/k and 0/epsilon I have just rearranged the epsilon value for your OUTFLOW_PIPE. I dont know how useful it is, but still you can try once. Have you tried komega SST (for low Re), try this as well. if you are not getting any solution. you can generate new mesh. with out seeing your mesh, I can not give you suggestion. Thanks and Regards, Aadhaven 

November 15, 2012, 12:05 

#16 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Hi Aadhavan,
The turbulence models are of no priority at the moment, since simulations also crash when using laminar flow. It seems you have swapped two values, but I do not expect the problems to be in those values anyway. A colleague gave me the suggestion to increase the accuracy of the outflow velocity in order to make it really perpendicular to the outflow plane. Unfortunately this did not solve anything either. I have run a quick simulation with a coarser and simplified mesh, and the problem is there as well. I really suspect it to be there due to having both a fixedValue for 0/p as well as a fixedValue for 0/U. As a test, I have now "swapped" the inflows and outflows to see if the problem moves to different places. At this moment I am pretty much clueless what can be the cause. The mesh may not be accurate everywhere, but also with a simplified mesh the problem persists. It occurs at a place where there are no problems in the mesh, and it is exactly on one intersection line... 

November 15, 2012, 12:15 

#17 
Member
Aathavan
Join Date: Nov 2012
Posts: 70
Rep Power: 4 
Dear Heroic,
I feel you need to generate new mesh. That may help you. other than that i have know information to share with you. Try to generate the new mesh, then try. thanks, Aadhavan 

November 19, 2012, 03:56 

#18 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Hello,
I would like to give an update on the case. Just for testing purposes I decided to remove the outlet velocity and set it to a zeroGradient instead. It is not exactly what I wanted to simulate, but it has been generating data until the hard drive did not have any space remaining. In other words, it did not crash, so I am quite sure the initial problem is in the boundary conditions as was stated before. My idea right now is to set a pressure difference in such a way that I can get approximately the desired outlet velocity, unless someone has any other suggestions for my BCs/ICs on the symmetry plane and pipe outlet. Kind regards. 

November 26, 2012, 03:44 

#19 
New Member
B
Join Date: Oct 2012
Posts: 17
Rep Power: 4 
Another update. I have been able to run long simulations on my coarsened mesh with great convergence and no crashes (not sure if I am overthinking things, but could it be that a noslip wall and a fixed velocity outlet directly touching eachother cause a conflict?).
I must add that I managed to do this with different OpenFOAM solvers, but I cannot get the turbulence model working in any of them without the simulation crashing again. Right now the plan is to play around (and get more familiar) with the fvSchemes and fvSolution and see if I can get it running, but it would be appreciated if some pointers could be given on what might be the cause of the turbulence model still not working. Thanks a lot. Kind regards. 

December 2, 2012, 06:48 

#20 
Member
Zifei Yin
Join Date: Sep 2012
Location: Shanghai & Ames
Posts: 33
Rep Power: 4 
You can add time adjustment in solver.
I believe it is the mesh problem. There are too many nonortho meshs 

Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
SimpleFoam k and epsilon bounded  nedved  OpenFOAM Running, Solving & CFD  13  November 4, 2013 15:13 
epsilon and K blowing up.  sivakumar  OpenFOAM Running, Solving & CFD  1  October 25, 2012 04:50 
k epsilon boundary conditions for pisofoam ras solver  chamoun  OpenFOAM  4  May 10, 2011 05:30 
Compressible epsilon blows up  swahono  OpenFOAM  10  November 26, 2010 06:38 
SimpleFoam k and epsilon bounded  nedved  OpenFOAM Running, Solving & CFD  1  November 25, 2008 21:21 