[Nek5000-users] Helmholtz error
nek5000-users at lists.mcs.anl.gov
nek5000-users at lists.mcs.anl.gov
Tue Jun 29 14:27:22 CDT 2010
David,
It looks as though you're hitting steady state.
If you set param 94 to zero in your .rea file, this will
correct the problem. It has to do with generation of an
initial guess and having linearly-dependent vectors in the
approximation space --- this can happen when there is no
dynamics in the system.
Note that your CFL is very small. You likely could increase
your DT -- perhaps set param 12 to +2.e-4 (i.e., > 0).
You could also set COURANT to be, say, 0.1, to be safe -- which
is sometimes necessary for Boussinesq problems at early times (only).
I would in this case also set IFCHAR to F, since there is no point
in using the characteristics timestepper if CFL ~ 0.1. (See
www.mcs.anl.gov/~fischer/oifs.pdf for an explanation.)
Finally,
> 321 DNORM, DIVEX 7.71505530689053011E-6 2.65666629955944499E-9
Indicates that you have a tighter tolerance for the pressure solver
( ~2.6e-9 ) than is realizable ( 7.7e-6 ). I'm guessing by the
peculiar tolerance that you have set DIVERGENCE=0 in the .rea file
(which is fine), and that the tolerance is thus set by TOLREL.
You might be able to dial that back a bit --- but given that you
have no flow and very little pressure iteration count, I'd leave
it as is.
Paul
On Tue, 29 Jun 2010, nek5000-users at lists.mcs.anl.gov wrote:
> Hi nek users,
>
> I'm getting an error I don't understand, perhaps someone can help. I'm
> simulating the Boussinesq equations in a 2D box of aspect ratio 2 with
> periodic side BC, stress-free velocity and fixed-flux temperature on the top
> and bottom, and at a Rayleigh number of 10^8. My time step is 2^-6; and
> longer and there are CFL problems. The part of the logfile where the error
> first appears is below. Any idea?
>
> Thanks a lot,
>
> David
>
>
> 321 Hmholtz VELX: 1 1.8722E-04 9.4841E-04 3.0177E-02
> 321 5.5836E+03 3.0971E-03 1.8028E+06 16 alph1x
> 321 halpha 16 9.1147E+00 1.6433E-01 1.6202E-03 1.5295E-03
> 3.2457E-04 9.6963E-05 5.3573E-05 2.3766E-05 1.5049E-05 1.1283E-05
>
> 321 Hmholtz VELY: 1 4.5111E-04 3.0971E-03 3.0177E-02
> 321 16 alpha: 4.3959E-02 1.3360E-03 1.2680E-04 6.4401E-06
> -3.8972E-06 3.2052E-06 7.4030E-06 3.4569E-06 2.8993E-06 1.4221E-06
> 321 16 3.5585E-04 1.2617E-07 2.8203E+03 alph12
> 321 U-Pres gmres: 3 2.6567E-09 4.1491E-09 1.2617E-07
> 1.5726E-03 2.5913E-03
> 321 DNORM, DIVEX 7.71505530689053011E-6 2.65666629955944499E-9
> 321 6.4200E-04 6.4809E-03 Fluid done
> 321 6.420000E-04 7.911223E-01 1.711772E+00 umax
> 321 0.6420E-03 t_av= 0.5000E+00 Ek= 0.1953E+00 Nu= 0.1000E+01
> Step 322, t= 6.4400000E-04, DT= 2.0000000E-06, C= 0.000 5.4661E+00
> 9.7695E-03
> Solving for heat
> Solving for fluid
> 322 Hmholtz TEMP: 1 9.9844E-03 6.0396E-01 2.4836E-02
> 322 6.4400E-04 1.0539E-03 Heat done
> 322 2.0339E+03 5.8756E-03 3.4616E+05 17 alph1x
> 322 halpha 17 3.3193E+00 1.0787E-01 5.3974E-04 9.8243E-04
> 3.7291E-04 2.3705E-04 5.1605E-05 3.4660E-05 1.1899E-05 1.1993E-05
>
> 322 Hmholtz VELX: 1 1.2039E-03 5.8756E-03 3.0705E-02
> 322 5.6870E+03 NaN NaN 17 alph1x
> 322 halpha 17 9.2834E+00 1.7756E-01 1.8590E-03 1.7684E-03
> 3.7603E-04 1.1302E-04 6.3138E-05 2.9137E-05 1.9115E-05 1.5046E-05
>
> 322 5000 **ERROR**: Failed in HMHOLTZ: VELY NaN NaN
> 3.0705E-02
> 322 17 alpha: NaN NaN NaN NaN
> NaN NaN NaN NaN NaN NaN
> 322 17 NaN NaN NaN alph12
> 322 U-Pres gmres: 100 NaN 4.1491E-09 NaN
> 5.2696E-02 8.8512E-02
> 322 DNORM, DIVEX NaN NaN
> 322 6.4400E-04 6.0996E-01 Fluid done
> 322 6.440000E-04 -9.900000E+21 -9.900000E+21 umax
> 322 0.6440E-03 t_av= 0.5000E+00 Ek= NaN Nu= NaN
> Step 323, t= 6.4600000E-04, DT= 2.0000000E-06, C= 0.000 6.0793E+00
> 6.1319E-01
> Solving for heat
> Solving for fluid
> 323 5000 **ERROR**: Failed in HMHOLTZ: TEMP NaN NaN
> 2.4836E-02
> 323 6.4600E-04 5.2047E-01 Heat done
> 323 NaN NaN NaN 18 alph1x
> 323 halpha 18 NaN NaN NaN NaN
> NaN
>
More information about the Nek5000-users
mailing list