[Nek5000-users] Guidelines for Tolerances

nek5000-users at lists.mcs.anl.gov nek5000-users at lists.mcs.anl.gov
Sat Mar 31 13:44:49 CDT 2018


In the PnPn formulation you don’t project into a divergence free space so you don’t prescribe the divergence. It’s a simulation result.

> On 31 Mar 2018, at 19:09, "nek5000-users at lists.mcs.anl.gov" <nek5000-users at lists.mcs.anl.gov> wrote:
> 
> I guess in the end the chosen/requested tolerances depend on what one is 
> interested in, and how sensitive a specific flow case is. Also, the 
> spatial distribution of the divergence error might be an interesting 
> quantity to look at.
> 
> However, I have a question to Stefan: I am not sure I fully understand 
> your comment regarding spatial resolution. Do you mean that the spatial 
> errors are so large that a specific divergence tolerance cannot be 
> fulfilled?
> 
> Philipp
> 
>> On 2018-03-31 11:20, nek5000-users at lists.mcs.anl.gov wrote:
>> Your tolerances look good to me. Typically 1e-5 for pressure is accurate enough. Note, in the Pn/Pn-2 formulation the tolerance is equivalent to the desired divergence error. This is in contract to Pn/Pn where such a relation does not exists. In most cases the divergence error is dominated by the spatial resolution. That's something you can try - just lower the tol to say 1e-5. If you'll get a similiar divergence error (compared to 1e-8) try to lower it even further.
>> 
>> --
>> Stefan
>> 
>> 
>> -----Original message-----
>>> From:nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
>>> Sent: Saturday 31st March 2018 11:07
>>> To: nek5000-users at lists.mcs.anl.gov
>>> Subject: [Nek5000-users] Guidelines for Tolerances
>>> 
>>> Dear Nek experts,
>>> 
>>> sorry, my first mail was incomplete. I will start again.
>>> 
>>> I am simulating turbulent pipe flow with 6 thermal fields treated as passive scalars.
>>> The first three thermal fields use Pr=0.71 and the remaining three use Pr=0.025.
>>> 
>>> In the par file, I set residualTol = 1e-08 for pressure velocity and all thermal fields. Having read the recent discussion about another logfile, I was wondering whether these settings might be too tight.
>>> 
>>> The log for one timestep is given below (sorry about the formatting).
>>> As you can see, I need 16 - 21 iterations for the scalar fields of Pr=0.025.
>>> Do you think I could reduce the residualTol for the low Pr fields to speedup the simulation?
>>> 
>>> Best regards,
>>> Steffen
>>> 
>>> 
>>> Step  99999, t= 8.4999900E+02, DT= 1.0000000E-03, C=  0.291 3.3389E+04 3.2523E-01
>>>              Solving for Hmholtz scalars
>>>       99999  Project TEMP           9.7160E-08   5.5444E+00   5.7064E+07   8   8
>>>       99999  Hmholtz TEMP       6   8.0673E-09   3.1410E-05   1.0000E-08
>>>       99999  Project PS 1           1.9038E-07   6.6127E+00   3.4734E+07   8   8
>>>       99999  Hmholtz PS 1       7   6.2818E-09   6.1081E-05   1.0000E-08
>>>       99999  Project PS 2           5.7194E-06   1.1100E+01   1.9408E+06   7   8
>>>       99999  Hmholtz PS 2       8   4.7345E-09   1.5839E-03   1.0000E-08
>>>       99999  Project PS 3           1.6110E-09   2.0874E-01   1.2958E+08   8   8
>>>       99999  Hmholtz PS 3      16   8.1941E-09   6.8549E-07   1.0000E-08
>>>       99999  Project PS 4           1.8743E-09   2.5528E-01   1.3620E+08   8   8
>>>       99999  Hmholtz PS 4      16   9.3461E-09   7.7774E-07   1.0000E-08
>>>       99999  Project PS 5           4.9138E-08   4.4469E-01   9.0499E+06   7   8
>>>       99999  Hmholtz PS 5      21   8.2110E-09   1.3142E-05   1.0000E-08
>>>       99999  Scalars done  8.5000E+02  1.7601E-01
>>>              Solving for fluid
>>>       99999  Project VELX           5.5774E-09   6.9874E-03   1.2528E+06   8   8
>>>       99999  Hmholtz VELX       4   2.7784E-09   1.7303E-06   1.0000E-08
>>>       99999  Project VELY           5.4643E-09   6.9118E-03   1.2649E+06   8   8
>>>       99999  Hmholtz VELY       4   2.7477E-09   1.6969E-06   1.0000E-08
>>>       99999  Project VELZ           5.0326E-09   6.9213E-03   1.3753E+06   8   8
>>>       99999  Hmholtz VELZ       4   6.0439E-09   2.0347E-06   1.0000E-08
>>>       99999  Project PRES           1.5397E-04   9.8607E-02   6.4043E+02  14  20
>>>       99999  U-PRES gmres      13   9.9860E-09   8.3983E-08   1.0000E-08   4.3257E-02   7.0065E-02
>>>       99999  volflow Z              1.8796E-02   8.0416E-06   7.8539E-01   7.8540E-01
>>>       99999  Fluid done  8.5000E+02  1.2875E-01
>>> ________________________________________
>>> Von: Straub, Steffen (ISTM)
>>> Gesendet: Samstag, 31. März 2018 10:45
>>> An: nek5000-users at lists.mcs.anl.gov
>>> Betreff: AW: Nek5000-users Digest, Vol 109, Issue 32
>>> 
>>> Dear Nek experts,
>>> 
>>> I am simulating turbulent pipe flow with 6 thermal fields treated as passive scalars.
>>> The first three thermal fields use Pr=0.71 and the remaining three use Pr=0.025.
>>> 
>>> 
>>> ________________________________________
>>> Von: Nek5000-users <nek5000-users-bounces at lists.mcs.anl.gov> im Auftrag von nek5000-users-request at lists.mcs.anl.gov <nek5000-users-request at lists.mcs.anl.gov>
>>> Gesendet: Freitag, 30. März 2018 17:02
>>> An: nek5000-users at lists.mcs.anl.gov
>>> Betreff: Nek5000-users Digest, Vol 109, Issue 32
>>> 
>>> Send Nek5000-users mailing list submissions to
>>>         nek5000-users at lists.mcs.anl.gov
>>> 
>>> To subscribe or unsubscribe via the World Wide Web, visit
>>>         https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>> or, via email, send a message with subject or body 'help' to
>>>         nek5000-users-request at lists.mcs.anl.gov
>>> 
>>> You can reach the person managing the list at
>>>         nek5000-users-owner at lists.mcs.anl.gov
>>> 
>>> When replying, please edit your Subject line so it is more specific
>>> than "Re: Contents of Nek5000-users digest..."
>>> 
>>> 
>>> Today's Topics:
>>> 
>>>    1. Re: logfile (nek5000-users at lists.mcs.anl.gov)
>>>    2. Re: logfile (nek5000-users at lists.mcs.anl.gov)
>>>    3. Re: logfile (nek5000-users at lists.mcs.anl.gov)
>>>    4. Re: logfile (nek5000-users at lists.mcs.anl.gov)
>>> 
>>> 
>>> ----------------------------------------------------------------------
>>> 
>>> Message: 1
>>> Date: Fri, 30 Mar 2018 15:34:49 +0200
>>> From: nek5000-users at lists.mcs.anl.gov
>>> To: nek5000-users at lists.mcs.anl.gov     <nek5000-users at lists.mcs.anl.gov>
>>> Subject: Re: [Nek5000-users] logfile
>>> Message-ID:
>>>         <mailman.12194.1522416912.19102.nek5000-users at lists.mcs.anl.gov>
>>> Content-Type: text/plain; charset=utf-8
>>> 
>>> That par file doesn't currently support a relative and absolute tolerance for the Helmholtz solver. I think there are two options to tackle your tolerance problem:
>>> 
>>> * Specify the parameters below in usrdat3 of your usr file.
>>> * Solve a non-dimensional problem and use a residual tolerance of e.g. 1e-8 (for velocity and temp) and 1e-5 for pressure
>>> 
>>> --
>>> Stefan
>>> 
>>> -----Original message-----
>>>> From:nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
>>>> Sent: Friday 30th March 2018 14:19
>>>> To: nek5000-users at lists.mcs.anl.gov
>>>> Subject: Re: [Nek5000-users] logfile
>>>> 
>>>> Thanks, Paul,
>>>>     Could you please tell me the parameters in terms of ###.par file.
>>>> I am using the latest version.
>>>> 
>>>> Thank you
>>>> 
>>>>> On Fri, Mar 30, 2018 at 5:35 PM,  <nek5000-users at lists.mcs.anl.gov> wrote:
>>>>> 
>>>>> 
>>>>>        1520  Hmholtz TEMP      12   3.8789E-11   4.5120E+07   1.0000E-10
>>>>> 
>>>>> 
>>>>> On Step 1520, it took 12 iterations to reach a residual of 3.8e-11, starting
>>>>> with a residual of 4.5e7, with target residual 1.e-10.
>>>>> 
>>>>> 
>>>>> [ This is actually too tight of a tolerance, since you are asking for 17
>>>>> orders of magnitude reduction in residual and are working only with 16
>>>>> significant digits. It seems like you might need to change your .rea file in
>>>>> the following way:
>>>>> 
>>>>> 
>>>>>        set p21 and p22 to 0
>>>>> 
>>>>> 
>>>>>        set p24, tolrel , to 1.e-3
>>>>> 
>>>>>        set p25, tolabs, to 1.e-3
>>>>> 
>>>>> 
>>>>> The only time you'll need tolrel/abs tighter than this is if you're doing an
>>>>> analytic convergence study where you know the answer to an infinite number
>>>>> of digits.   For engineering calculations, 1.e-2 to 1.e-3 is fine.   The
>>>>> residual tolerances are set conservatively based on these input values.
>>>>> 
>>>>> 
>>>>>   1520  Helmh3 fluid      17   3.6594E-11   4.7524E-01   1.0000E-10
>>>>>        1520  U-PRES gmres     120   1.3494E-08   4.5675E-06
>>>>> 1.0000E-10   7.8799E-01   3.0171E+00
>>>>> 
>>>>> 
>>>>> You have 17 iterations for the moment equations.
>>>>> 
>>>>> 
>>>>> You have 120 iterations for pressure.   Again, your tolerances are too tight
>>>>> for engineering calculations -- you do not need 1.e-10 for divergence (i.e.,
>>>>> for the pressure residual tolerance).
>>>>> 
>>>>> 
>>>>> hth,
>>>>> 
>>>>> Paul
>>>>> 
>>>>> 
>>>>> ________________________________
>>>>> From: Nek5000-users <nek5000-users-bounces at lists.mcs.anl.gov> on behalf of
>>>>> nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
>>>>> Sent: Friday, March 30, 2018 4:52:26 AM
>>>>> To: nek5000-users at lists.mcs.anl.gov
>>>>> Subject: Re: [Nek5000-users] logfile
>>>>> 
>>>>> The following is my logfile entry (sample lines). The first line is
>>>>> understandable. What do the values in the third and following lines
>>>>> denote?
>>>>> 
>>>>> 
>>>>> ------------------------------------------------------
>>>>> Step   1520, t= 7.6000000E-02, DT= 5.0000000E-05, C=  0.072 5.3831E+03
>>>>> 3.5403E+00
>>>>>              Solving for Hmholtz scalars
>>>>>        1520  Hmholtz TEMP      12   3.8789E-11   4.5120E+07   1.0000E-10
>>>>>        1520  Scalars done  7.6000E-02  8.3823E-02
>>>>>              Solving for fluid
>>>>>        1520  Helmh3 fluid      17   3.6594E-11   4.7524E-01   1.0000E-10
>>>>>        1520  U-PRES gmres     120   1.3494E-08   4.5675E-06
>>>>> 1.0000E-10   7.8799E-01   3.0171E+00
>>>>>        1520  Fluid done  7.6000E-02  3.3650E+00
>>>>> --------------------------------------------------------
>>>>> 
>>>>>> On Fri, Mar 30, 2018 at 3:05 PM,  <nek5000-users at lists.mcs.anl.gov> wrote:
>>>>>> I am not sure what you are talking about?
>>>>>> 
>>>>>> -----Original message-----
>>>>>>> From:nek5000-users at lists.mcs.anl.gov <nek5000-users at lists.mcs.anl.gov>
>>>>>>> Sent: Friday 30th March 2018 10:30
>>>>>>> To: nek5000-users at lists.mcs.anl.gov
>>>>>>> Subject: [Nek5000-users] logfile
>>>>>>> 
>>>>>>> Hi Neks,
>>>>>>> 
>>>>>>>      What are the entries in the logfile that is output from running
>>>>>>> the code.  Please let me know where the entries are determined in the
>>>>>>> code.
>>>>>>> 
>>>>>>> Thanks
>>>>>>> Swarandeep
>>>>>>> _______________________________________________
>>>>>>> Nek5000-users mailing list
>>>>>>> Nek5000-users at lists.mcs.anl.gov
>>>>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>>>>>> 
>>>>>> _______________________________________________
>>>>>> Nek5000-users mailing list
>>>>>> Nek5000-users at lists.mcs.anl.gov
>>>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>>>> _______________________________________________
>>>>> Nek5000-users mailing list
>>>>> Nek5000-users at lists.mcs.anl.gov
>>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>>>> 
>>>>> _______________________________________________
>>>>> Nek5000-users mailing list
>>>>> Nek5000-users at lists.mcs.anl.gov
>>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>>>> 
>>>> _______________________________________________
>>>> Nek5000-users mailing list
>>>> Nek5000-users at lists.mcs.anl.gov
>>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>>> 
>>> _______________________________________________
>>> Nek5000-users mailing list
>>> Nek5000-users at lists.mcs.anl.gov
>>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>>> 
>> _______________________________________________
>> Nek5000-users mailing list
>> Nek5000-users at lists.mcs.anl.gov
>> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
>> 
> _______________________________________________
> Nek5000-users mailing list
> Nek5000-users at lists.mcs.anl.gov
> https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users



More information about the Nek5000-users mailing list