[heavy_user]

Hi All,

i am experiencing some spooky things here...

I am running a case with reactingFoam.
It is running fine, results seem to look nice...so far..

BUT: i can check flowtime in the log, he is writing save files as he is supposed to...Adictionally he is writing files from a flowtime which is not yet reached and which dont match the specified write intervall at all!
(I attached a picture; the picture was taken when log said he is at flowtime 0.0028 sec)

does anyone know Why?
Dont want to have to delete and sort out all the rubish being written ...

regards


here the control dict:

{
version 2.0;
format ascii;
class dictionary;
location "system";
object controlDict;
}
// * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * //


application reactingFoam;

startFrom startTime;

startTime 0;

stopAt endTime;

endTime 10;

deltaT 0.000001;

writeControl timeStep;

writeInterval 100;

purgeWrite 0;

writeFormat ascii;

writePrecision 6;

writeCompression uncompressed;

timeFormat general;

timePrecision 6;

runTimeModifiable no;

adjustTimeStep no;

maxCo 0.5;

[gschaider]

Originally Posted by heavy_user

Hi All,

i am experiencing some spooky things here...

I am running a case with reactingFoam.
It is running fine, results seem to look nice...so far..

BUT: i can check flowtime in the log, he is writing save files as he is supposed to...Adictionally he is writing files from a flowtime which is not yet reached and which dont match the specified write intervall at all!
(I attached a picture; the picture was taken when log said he is at flowtime 0.0028 sec)

VERY WILD GUESS: did you start a "parallel" run without the -parallel-option? That would start N _serial_ processes (all of them calculating the same thing). If one is faster than the one that writes the log it might look like there are time-directories from the future. But this still wouldn't explain the odd intervalls

[heavy_user]

HI Bernhard,

thx for the answer!

I did start the run with "reactingFoam > log ".
I did NOT split the domain for several processors.
I have no file (as far as i know) which tells reactingFoam to go parallel...
CPU usage is 48% = 1 Processor being bussy(dualcore)..

...perhaps from a parallel universe ?? to bring a VeRyWiLdGuEsS on :-P ...

regards!

P.S.: would you mind taking a look at my post from the 20.01.2010 at 13.24 in "specifying nonuniform boundary conditions" ??
i am absolutely stuck here... and thx again, i think i might owe you a beer(or two) by now...

[gschaider]

Originally Posted by heavy_user

I did start the run with "reactingFoam > log ".
I did NOT split the domain for several processors.
I have no file (as far as i know) which tells reactingFoam to go parallel...
CPU usage is 48% = 1 Processor being bussy(dualcore)..

Only a guess, as I said. No idea where that kind of behaviour can come from

Originally Posted by heavy_user

...perhaps from a parallel universe ?? to bring a VeRyWiLdGuEsS on :-P ...

Your guess is better

Originally Posted by heavy_user

P.S.: would you mind taking a look at my post from the 20.01.2010 at 13.24 in "specifying nonuniform boundary conditions" ??

I knew it was a mistake to answer in this thread

I think I skipped that one, as I find it very hard to read unformatted C++-sources (see Mark's remark on the importance of the CODE-tag elsewhere on the forum - today or yesterday)

No idea. Where do the graphs come from? sample?

Originally Posted by heavy_user

i am absolutely stuck here... and thx again, i think i might owe you a beer(or two) by now...

The rumors that I drink that stuff are without any foundation

[heavy_user]

Hi Bernhard,

the dimension-leak seems to have shut...I did not change anything (just restartet it, since it crashe due to a bug making T run out of boundaries) ...and it stopped writing that stuff ...

Originally Posted by gschaider

I knew it was a mistake to answer in this thread

...too late ...

Originally Posted by gschaider

I think I skipped that one, as I find it very hard to read unformatted C++-sources (see Mark's remark on the importance of the CODE-tag elsewhere on the forum - today or yesterday)

very sorry for that one, I sticked to code-convention, writing it in texteditor...but when i copy pasted, it messed up...
Whats mark's username, can neither find "mark" nor a thread called "code-tag" and browsing the posts of today and yesterday did not work out either (my eyes are tooo tired from reading error messages ALL day, i guess...)

the file basically says (or should say) something like:

internalFiled.y < value -> CH4 =1;
elsewhere CH4 = 0 ;
but I end up with a z-dependece of the distibution ...

I changed the post and added the original file - if you would mind to take a look...

Originally Posted by gschaider

No idea. Where do the graphs come from? sample?

I did the graphes with "sample" since i discovered the mess in paraFiew and i wanted to check on it...

Originally Posted by gschaider

The rumors that I drink that stuff are without any foundation

i dont drink it either, but some kind of goblin drops his empty bottles in my room at night .. or the dimension-crack is hounting me...

Give me your adress and I will send you a bottle of milk, if you prefere that ..

regards!

[gschaider]

Originally Posted by heavy_user

very sorry for that one, I sticked to code-convention, writing it in texteditor...but when i copy pasted, it messed up...
Whats mark's username, can neither find "mark" nor a thread called "code-tag" and browsing the posts of today and yesterday did not work out either (my eyes are tooo tired from reading error messages ALL day, i guess...)

Copy pasting is OK. What he said in the post was that people should first click on the #-Icon and past beween the CODE-tags. That way the Board won't try to reformat (BTW: olesen is the username)

Originally Posted by heavy_user

the file basically says (or should say) something like:

internalFiled.y < value -> CH4 =1;
elsewhere CH4 = 0 ;
but I end up with a z-dependece of the distibution ...

How is the patch aligned?

Originally Posted by heavy_user

I changed the post and added the original file - if you would mind to take a look...

I think this should be discussed in that thread

Originally Posted by heavy_user

I did the graphes with "sample" since i discovered the mess in paraFiew and i wanted to check on it...

Make sure that what you get is only the fdace-values and not something interpolated

Bernhard

[heavy_user]

Originally Posted by gschaider

Copy pasting is OK. What he said in the post was that people should first click on the #-Icon and past beween the CODE-tags. That way the Board won't try to reformat (BTW: olesen is the username)

How is the patch aligned?

hi Bernhard,
I updated the post, I hope you like it now...and you have all the information you need!

Originally Posted by gschaider

I think this should be discussed in that thread
Bernhard

We will

Originally Posted by gschaider

Make sure that what you get is only the fdace-values and not something interpolated

Bernhard

Even if it interpolates, the distribution should be independent of z.
The values between 0 and 1, which should not be there acording to file 0/CH4, could be due to interpolation if they were only in areas somewhere between areas with value 1 and areas with value 0...but they are not...there are no areas with 1 if i check with sample or paraFoam, eventhough the file 0/CH4 has ONLY entrys with 0 and 1..
Adictionaly I checked with paraFoam, and when I do "wireframe" he should not be interpolating(?) ...and there is no value 1 at all, he seems to cut down my peak value !
...so i am totaly confused...

have a nice weekend!

regards!