
[Sponsors] 
December 26, 2011, 18:59 
Highly unexpected occurence of Enforce_bounds error

#1 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 9 
Hi,
I am facing a very strange and unexpected situation. I have converged results for a conjugate heat transfer problem with the option ' Essential' selected in Output control. I have used komega turbulence model in my simulation. I need yplus values n my results and I realized that yplus values are not available when ' Essential' option is selected in output control. So, to get yplus values also in results, I just reran my simulation by changing the Output control option to 'Standard' and using the already available converged results as initial condition with Continue history option enabled in solver. The simulation proceeds and reaches convergence without any issue. However, just before forming the result file, solver reports Enforce_bounds error and stops without forming the result file !!! The location of Enforce_bounds error is reported to be one of the wall boundary conditions and for fluid density. However, the maximum and minimum fluid densities in the fluid domain are within physical limits. I have already checked my simulation setup, also tried with 'Selected variables' option in Output control and just reran the old simulation for confirmation. I have concluded that the error occurs only when I change the Output control option from ' Essential' to either ' Standard' or ' Selected variables' . I have no clue about the source of this problem now and any inputs will be of immense help. I am attaching the .out file for your reference. Best regards.. 

December 26, 2011, 19:03 

#2 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 9 
The .out file is attached herewith..


December 27, 2011, 06:19 

#3 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 14,077
Rep Power: 109 
The problem is probably due to you now evaluating a variable when you generate the larger results file. The essential results file is not including the variable but now it is in there and is causing the problem.
You have a lot of CEL in there and I suspect that either something is wrong with your CEL, or an expression which should not be evaluated somewhere (maybe at a wall). Alternately you have added only two variables to your variables list for the results file, they are also two key ones to look at. 

March 5, 2012, 06:25 

#4 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 9 
Hi everyone,
I am still stuck with this strange problem. I have contacted Ansys support and implemented the suggestion given by them. However, the issue is still unsolved. I have further analyzed the problem and I am describing my findings here below. Simulation setup: A. I am simulating conjugate heat transfer in a heat sink with turbulent flow. I am using komega model for turbulence. B. The fluid properties are temperature dependent. I have used CEL expressions for fluid properties which give fluid properties as function of local temperature. C. The simulation proceeds without any problem and converges. The issue: I need solver yplus values which are available in CFX post only if 'Standard' option is used in output control. However, by using this option, the solver reports unphysical fluid temperature values (of the order of 1e4 K) at a few mesh nodes on fluidsolid wall while forming the result file (and AFTER the simulation has converged). The error message is shown below : ****** Notice ******  While evaluating Static Entropy,  Temperature on boundary IPO Inlet manifold top adiabatic wall  went outside of its upper limit. Its maximum value was  3.4538E+04. The bounds error was handled by clipping.  If this situation persists, consider increasing the table range. ****** Notice ******  While evaluating Static Entropy,  Temperature on boundary Outlet side fluid adiabatic walls  went outside of its upper limit. Its maximum value was  6.8304E+05. The bounds error was handled by clipping.  If this situation persists, consider increasing the table range. The boundaries reported above belong to two different domains in the simulation. The error happens inspite of the fact that .out file reports temperature values within physical limits for both the domains after the simulation is finished and just before reporting this error!!. The relevant parts of .out file is shown below ================================================== ==================== Termination and Interrupt Condition Summary ================================================== ==================== CFD Solver: All target criteria reached (Equation residuals AND global imbalances) ++  Variable Range Information  ++ Domain Name : Inlet side fluid ++  Variable Name  min  max  ++  Density  9.95E+02  9.96E+02   Specific Heat Capacity at Constant Pressure 4.18E+03  4.18E+03   Dynamic Viscosity  7.54E04  7.97E04   Thermal Conductivity_xx  6.16E01  6.20E01   Thermal Conductivity_yy  6.16E01  6.20E01   Thermal Conductivity_zz  6.16E01  6.20E01   Thermal Conductivity_xy  0.00E+00  0.00E+00   Thermal Conductivity_xz  0.00E+00  0.00E+00   Thermal Conductivity_yz  0.00E+00  0.00E+00   Static Entropy  4.23E+02  4.59E+02   Velocity u  4.20E01  1.43E+00   Velocity v  8.90E01  3.01E01   Velocity w  8.47E01  9.45E01   Pressure  9.81E+02  2.53E+03   Turbulence Kinetic Energy  1.73E11  3.72E01   Turbulence Eddy Frequency  8.21E+01  3.11E+05   Eddy Viscosity  4.87E13  1.04E01   Temperature  3.03E+02  3.06E+02   Static Enthalpy  1.26E+05  1.37E+05  ++ Domain Name : Outlet side fluid ++  Variable Name  min  max  ++  Density  9.94E+02  9.95E+02   Specific Heat Capacity at Constant Pressure 4.18E+03  4.18E+03   Dynamic Viscosity  7.25E04  7.57E04   Thermal Conductivity_xx  6.19E01  6.23E01   Thermal Conductivity_yy  6.19E01  6.23E01   Thermal Conductivity_zz  6.19E01  6.23E01   Thermal Conductivity_xy  0.00E+00  0.00E+00   Thermal Conductivity_xz  0.00E+00  0.00E+00   Thermal Conductivity_yz  0.00E+00  0.00E+00   Static Entropy  4.56E+02  4.85E+02   Velocity u  4.57E01  1.74E+00   Velocity v  1.40E+00  4.71E01   Velocity w  1.30E+00  1.13E+00   Pressure  9.03E+02  1.13E+03   Turbulence Kinetic Energy  2.91E13  2.35E01   Turbulence Eddy Frequency  1.64E+02  1.02E+06   Eddy Viscosity  5.11E15  4.62E02   Temperature  3.06E+02  3.08E+02   Static Enthalpy  1.36E+05  1.45E+05  ++ The fact that temperature remains within physical range till the simulation is converged is also shown by monitor plots of maximum temperature in various domains in the simulation.See attached figure monitorplots.png. There are other domains also in simulation setup but there is no problem in those domains. Following interaction with Ansys support, I tried the following two things which have not made any difference in the issue. 1) Constraining the temperature to within some limits while evaluating the fluid properties through CEL expressions. I did the following to constrain the temperature within 2070 deg C as the temperature cannot go outof this range in my simulation TC = min(max(T,293.15),343.15) and then use this to evaluate fluid properties 2) I even tried with constant fluid properties by setting TC to a constant value. 3) I removed extra output variables from output control. There was one minor change after doing steps 1 and 2 above. Before trying these changes, the solver was stopping after convergence, giving a Enforce_bounds error, and not forming any result file. However with both the changes 1) and 2) above, the solver did form a result file and I was able to see where the unphysical temperature was . Also, there was one minor difference between cases 1 and 2 above. With the case 1 above, the solver reported the bounds error as shown in this post above in red and formed the result file. However with change 2, the solver did not report any bounds error and formed the result file. But the result file in both the cases 1) and 2) above, showed same unphysical temperature at a few points on the two fluidwall boundaries. I am showing the unphysical temperature on of the fluid wall boundaries. See attached figure temperature.png All these problems do not occur , if I just use the option 'Essential' in output control. I am showing the same temperature plot with the option 'Essential' in the attached figure temperature2.png Any suggestions to get around this situation will be extremely helpful. Sorry for the long post. Regards, Chander 

March 5, 2012, 06:46 

#5 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 14,077
Rep Power: 109 
So your simulation runs and converges, but when you restart it to get the additional parameter then you get this bounds error?
So that sounds like a problem on restarts and/or the calculation of the additional output file variables, doesn't it? 

March 5, 2012, 06:49 

#6 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 9 
Well it is not a restart error as I ran the simulation again from beginning to rule out this possibility.
Yes the issue occurs when I ask CFX to give me some more output variables by using the 'Standard' option instead of 'Essential'. 

March 5, 2012, 15:55 

#7 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 9 
So..any ideas?


March 5, 2012, 18:19 

#8 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 14,077
Rep Power: 109 
This sounds like a bug in CFX. I know you have already talked to CFX support but I suspect you need to make it clear that the issue appears to be a bug.
But I would only bother reporting it if you are using the latest version of the solver. They do not case about bugs in old versions. If you are not using the latest version then you should upgrade  you are never going to get anywhere on this issue if you don't. 

March 8, 2012, 11:21 

#9 
Senior Member
Join Date: Oct 2010
Location: Zurich
Posts: 176
Rep Power: 9 
Hello Glen,
Well I cannot upgarde to CFX 14 at the moment. The basic reason why I need the result file with Standard option is that I need y+ values. Actually for my calculations, I need y+ value for every mesh cell/node. What I am trying to do is to plot local profiles of entropy generation for my problem and it is for this purpose that the above info is required. I now realize that Standard option will only give me y+ values for first mesh cells from wall while, as I said above, I need these values for all mesh cells. Is there a way by which I can calculate y+ value for any mesh cell/node? I think, to calculate this for any mesh node, I will need to find actual distance of that mesh node from nearest wall and velocity gradients at that corresponding location (projected location of the node) on that wall (to calculate wall shear stress). Is there a way by which I can do this myself in CFXPost or outside? If there is, then I can simply use the result files with Essential option that I already have and not bother about the current problem. 

March 8, 2012, 18:16 

#10 
Super Moderator
Glenn Horrocks
Join Date: Mar 2009
Location: Sydney, Australia
Posts: 14,077
Rep Power: 109 
Personally I think you are wasting your time working around bugs on old versions of the software, but it is up to you.
Obviously you can define your own y+ as a user variable set with the definition of y+. But y+ is defined from wall shear stress and that only exists at the walls so will not help you much. You might be able to use a if/step function to put zeros in for nodes away from the walls. 

Tags 
cfx 12.1 
Thread Tools  
Display Modes  


Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
CGNS Compiling  Diego  Main CFD Forum  17  December 21, 2014 02:40 
Accessing phi from a fvPatchField at same patch  johndeas  OpenFOAM  1  September 13, 2010 20:23 
UDF: DEFINE_CG_MOTION for vertical jump motion of an electrode!  alban  Fluent UDF and Scheme Programming  2  June 8, 2010 18:54 
compile errors of boundary condition "expDirectionMixed"  liying02ts  OpenFOAM Bugs  2  February 1, 2010 21:11 
Installation of Netgen in SuSE Linux 92  edvardsenpriv  Open Source Meshers: Gmsh, Netgen, CGNS, ...  23  January 16, 2009 07:12 