[Getdp] still facing discontinuities when using -restart option
Joseph Frippiat
joseph_frippiat at voo.be
Wed Apr 4 12:16:08 CEST 2018
For the restarts, I don't do a SaveSolution after the InitSolution.
My simulation is inspired by the HeatReservoir example on the old wiki.
I use the same principle: a bash script exchanging data with getdp
through a .pro file and restarting the simulation as needed.
By mimicking the example, I'd modified the file machine_magstadyn_a.pro
like this for the restart:
Operation {
If(par_batch==0) // start of the simulation
...
Else // restarts
InitSolution[A];
//ReadSolution[A]; // redundant => warnings "Prematured end of file"
InitSolution[M];
//ReadSolution[M]; // redundant => warnings "Prematured end of file"
InitSolution[T];
//ReadSolution[T]; // redundant => warnings "Prematured end of file"
...
EndIf // If(par_batch==0)
...
As in the example, for the restarts, I was also calling ReadSolution
(which caused problems) until Christophe has shown me that it was not
necessary.
I have only discontinuities for the system A (electromagnetical one) but
not for de Mechanical one and not for the Thermal one.
Joseph
Le 4/04/2018 à 10:03, Ruth Vazquez Sabariego a écrit :
> Having a look at “machine_magstadyn_a.pro”, I think that the problem
> may come from SaveSolution.
> You have:
> InitSolution[A];
> SaveSolution[A];
>
> InitSolution takes the data from your res file.
> SaveSolution saves the current solution, which is not yet computed,
> thus zero.
>
> Can you try by commenting it out:
> //SaveSolution[A];
>
> Let me know if that solves the issue.
>
> Best,
> Ruth
>
>
>
>
>> On 3 Apr 2018, at 20:57, Joseph Frippiat <joseph_frippiat at voo.be
>> <mailto:joseph_frippiat at voo.be>> wrote:
>>
>> Ruth,
>>
>> yes, the discontinuities coincide with the restart: the value of the
>> first point after a restart is not correct for the current.
>>
>> For the simulation, I use a modified version of
>> machine_magstadyn_a.pro. The current is the current in the circuit
>> connected to the stator coils (which is in fact another rotor...)
>>
>> For the restart, I do
>> getdp model1 -restart -solve CoSimulation -bin -v 3
>> (the post-processing is called from the Resolution as it is done in
>> the original machine_magstadyn_a.pro)
>>
>> No point seems to be missing as one can see in the log :
>>
>> end of restart 1:
>>
>> ...
>> Info : Theta Time = 0.12008873 s (TimeStep 1094, DTime 0.00010795)
>> Info : IterativeLoop converged (6 iterations, residual 6.57266e-07)
>> E n d P r o c e s s i n g
>> Info : Stopped (Tue Apr 3 13:17:21 2018, Wall = 77.2714s, CPU =
>> 261.174s, Mem = 38.8398Mb)
>>
>> beginning of restart 2:
>>
>> Info : Started (Tue Apr 3 13:17:22 2018, Wall = 0.0200088s, CPU =
>> 0.055353s, Mem = 4.40234Mb)
>> ...
>> P r e - P r o c e s s i n g . . .
>> Info : System 1/3: 745 Dofs
>> Info : System 2/3: 4 Dofs
>> Info : System 3/3: 575 Dofs
>> E n d P r e - P r o c e s s i n g
>> P r o c e s s i n g . . .
>> Info : Theta Time = 0.12019546 s (TimeStep 1095, DTime 0.00010673)
>> Info : IterativeLoop converged (5 iterations, residual 1.42442e-07)
>> ...
>>
>> Joseph
>>
>> Le 3/04/2018 à 11:33, Ruth Vazquez Sabariego a écrit :
>>> Hi Joseph,
>>>
>>> Do the discontinuities coincide with the restart?
>>> How are you doing the post-processing for getting the current?
>>> Does it behave in the same way if you do post-pro straightaway or
>>> just after?
>>>
>>> I mean
>>> getdp xxx.pro -restart -solve xxx -pos xxx
>>> OR
>>> getdp xxx.pro -restart -solve xxx
>>> getdp xxx.pro -pos xxx
>>>
>>> The behaviour should be the same but it could be that we are missing
>>> a time step…
>>> I do not use restart often but when I do I take the first option
>>> without observing any particular weird behaviour.
>>>
>>> Best regards,
>>> Ruth
>>>
>>>
>>>
>>>> On 3 Apr 2018, at 01:29, Joseph Frippiat <joseph_frippiat at voo.be
>>>> <mailto:joseph_frippiat at voo.be>> wrote:
>>>>
>>>> Hello,
>>>>
>>>> as one can see on the following picture, I get discontinuities each
>>>> time I restart my calculations with the -restart option.
>>>>
>>>> There is surely something that I miss but I can't find it.
>>>>
>>>> Did someone else encouter this problem that could point me to a
>>>> solution ?
>>>>
>>>> Thanks in advance,
>>>>
>>>> Joseph
>>>>
>>>> Here is the picture:
>>>>
>>>> <Figure 0.png>
>>>> _______________________________________________
>>>> getdp mailing list
>>>> getdp at onelab.info <mailto:getdp at onelab.info>
>>>> http://onelab.info/mailman/listinfo/getdp
>>>
>>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://onelab.info/pipermail/getdp/attachments/20180404/eb2cc0a4/attachment.html>
More information about the getdp
mailing list