Problem in running a distributions_and_histograms.fds.

104 views
Skip to first unread message

H. M. Iqbal Mahmud

unread,
Jan 25, 2012, 1:48:36 AM1/25/12
to fds...@googlegroups.com


Hi,

i was trying to run distributions_and_histograms.fds file in FDS6, which i have downloaded from fds repository. I have got FDS6 by compiling the source codes at december 2011.

However, I could not run it. It showed a error message -

Forrtl: sever <408>: fort: <3>: subscript #1 of the array counts has value 0 which is less than the lower bound of 1


Would anyone of you know the reason of this error message?

Kind Reagrds

Kevin

unread,
Jan 25, 2012, 7:48:35 AM1/25/12
to FDS and Smokeview Discussions
This is a small diagnostic test case. It works for me with the latest
version of the FDS source code in the repository. What is the SVN
number and compile date of your version?

On Jan 25, 1:48 am, "H. M. Iqbal Mahmud" <hmiqbalmah...@gmail.com>
wrote:
> Hi,
>
> i was trying to run distributions_and_histograms.fds file in FDS6, which i
> have downloaded from fds repository. I have got FDS6 by compiling the
> source codes at december 2011.
>
> However, I could not run it. It showed a error message -
>
> *Forrtl: sever <408>: fort: <3>: subscript #1 of the array counts has value
> 0 which is less than the lower bound of 1*
> * *

H. M. Iqbal Mahmud

unread,
Jan 25, 2012, 8:27:31 PM1/25/12
to fds...@googlegroups.com
Hi Kevin,

Thanks for your reply. I downloaded the source codes at 23 Nov 2011
and compiled those at 24 Nov 2011. i could not find the SVN number in
the exe file.

However, this test file was uploaded at 23 Sep 2011. So, it should run
in my compiled version.

Do you think that i should compile the latest source codes again?

Thanks again.

Regards

H. M. Iqbal Mahmud

unread,
Jan 25, 2012, 9:03:31 PM1/25/12
to fds...@googlegroups.com
Hi Kevin,

I have downloaded the source codes today and compiled it again. The
SVN number is 9909. The test case is showing the same problem.

I am not sure that it the problem with the input file or the source codes.

Would you please help me to run this file?

Kind regards

Kevin

unread,
Jan 26, 2012, 8:53:47 AM1/26/12
to FDS and Smokeview Discussions
This is a bug. I'll fix it.

On Jan 25, 9:03 pm, "H. M. Iqbal Mahmud" <hmiqbalmah...@gmail.com>
wrote:
> Hi Kevin,
>
> I have downloaded the source codes today and compiled it again. The
> SVN number is 9909. The test case is showing the same problem.
>
> I am not sure that it the problem with the input file or the source codes.
>
> Would you please help me to run this file?
>
> Kind regards
>
> On 1/26/12, H. M. Iqbal Mahmud <hmiqbalmah...@gmail.com> wrote:
>
>
>
> > Hi Kevin,
>
> > Thanks for your reply. I downloaded the source codes at 23 Nov 2011
> > and compiled those at 24 Nov 2011. i could not find the SVN number in
> > the exe file.
>
> > However, this test file was uploaded at 23 Sep 2011. So, it should run
> > in my compiled version.
>
> > Do you think that i should compile the latest source codes again?
>
> > Thanks again.
>
> > Regards- Hide quoted text -
>
> - Show quoted text -

Kevin

unread,
Jan 26, 2012, 8:56:42 AM1/26/12
to FDS and Smokeview Discussions
I just posted new source code that should fix the problem.
> > - Show quoted text -- Hide quoted text -

H. M. Iqbal Mahmud

unread,
Jan 26, 2012, 7:38:45 PM1/26/12
to fds...@googlegroups.com
Hi Kevin,

Thanks a lot for your help.

However, still the test file is not running. It is showing the same
error message.

I have compiled the latest source codes today in visual studio without
showing any error or warning message.

If you please have a look to this matter, it would be so kind of you.

Kind regards

Iqbal

Topi Sikanen

unread,
Jan 27, 2012, 4:41:11 AM1/27/12
to FDS and Smokeview Discussions
I also was unable to reproduce this error. Try to compile a version
with debugging info and see if you can get it to report the offending
source code line.

H. M. Iqbal Mahmud

unread,
Jan 27, 2012, 7:27:46 AM1/27/12
to fds...@googlegroups.com
Hi Topi,
Thanks for your reply. However, would you please explain a little bit more?

Is it that when you had run that file, it run successful?

Other thing is that I compiled the source codes in debugging mode and it was compiled successfully without showing any error message.

If those information complied with your message then I have understood your message, otherwise I have misunderstood.

Kind regards

Iqbal

Kevin

unread,
Jan 27, 2012, 8:07:41 AM1/27/12
to FDS and Smokeview Discussions
I compiled FDS serial version, SVN 9910, on a Windows XP 32 bit laptop
using the Intel compiler. I used the makefile that is in
FDS_Compilation of the FDS repository. Both release and debug
compilations run the case successfully.

Which compiler are you using, and are you using the makefiles in the
repository?

On Jan 27, 7:27 am, "H. M. Iqbal Mahmud" <hmiqbalmah...@gmail.com>
wrote:

dr_jfloyd

unread,
Jan 27, 2012, 8:21:45 AM1/27/12
to fds...@googlegroups.com
Have you done a clean build? (that is deleting all module and object files so that all source files are compiled). 

H. M. Iqbal Mahmud

unread,
Jan 27, 2012, 8:22:09 AM1/27/12
to fds...@googlegroups.com
Hi Kevin,

I have compiled the source codes in Visual Studio 2008. No, I have not used the makefile as I know that I don't need makefile to compile in visual studio.

Is it the problem with not to use makefile...???

Kind regards.



Kevin

unread,
Jan 27, 2012, 8:32:00 AM1/27/12
to FDS and Smokeview Discussions
I stopped using Visual Studio several years ago because I find it very
difficult to know exactly what options are being used. The makefile
provides us all with a consistent set of options. Otherwise we have
this current situation where Topi and I are able to run this cases
successfully and you are not.

If you have compiled in debug mode, what is the line number and the
error that occurs?

On Jan 27, 8:22 am, "H. M. Iqbal Mahmud" <hmiqbalmah...@gmail.com>
wrote:

H. M. Iqbal Mahmud

unread,
Jan 27, 2012, 8:39:46 AM1/27/12
to fds...@googlegroups.com

Hi Floyd,

I removed all previous files and added new source files and then compiled.
However, I will clean and then build again.



Hi Kevin,

I will run the file and post the error message window as a picture.

Thanks to all of you for your kind support.

H. M. Iqbal Mahmud

unread,
Jan 29, 2012, 6:24:03 PM1/29/12
to fds...@googlegroups.com
Hi kevin,
I have attached the error message as attachment. May be, you can
determine the problem regarding this and give me advice to solve this
problem.


Hi Floyd,
I cleaned the solution and then built again. It did not change
anything. However, i have attached the error message.


Kind regards

Error Message.JPG

Topi Sikanen

unread,
Jan 30, 2012, 3:54:30 AM1/30/12
to FDS and Smokeview Discussions
There was a bug in UPDATE_HISTOGRAM that would result in array out of
bounds if the value went out of the LIMITS defined on PDPA namelist
group. I just committed a version that should fix this.

On 30 tammi, 01:24, "H. M. Iqbal Mahmud" <hmiqbalmah...@gmail.com>
wrote:
>  Error Message.JPG
> 949 ktNäytäLataa

H. M. Iqbal Mahmud

unread,
Jan 30, 2012, 9:23:59 PM1/30/12
to fds...@googlegroups.com
Hi Topi,

That's great. The compiled version of new source codes is working now.

I was looking to the csv output files. However, I am confused about
the output parameters, as i have not found anything about the
distribution and histogram in the downloaded FDS6_user_guide.txt
(converted to pdf file).

I was expecting droplet size distribution data to draw a distribution
diagram or histogram from the csv output file. But, it seems something
different.

However, would you please explain me about those outputs or can i get
droplet size distribution data to draw a distribution diagram or
histogram?

If you have any document or reference regarding those parameters, i
would be happy to great from you.

Thanks again for your kind help.

Regards

Iqbal

> --
> You received this message because you are subscribed to the Google Groups
> "FDS and Smokeview Discussions" group.
> To post to this group, send email to fds...@googlegroups.com.
> To unsubscribe from this group, send email to
> fds-smv+u...@googlegroups.com.
> For more options, visit this group at
> http://groups.google.com/group/fds-smv?hl=en.
>
>

Topi Sikanen

unread,
Jan 31, 2012, 2:18:37 AM1/31/12
to FDS and Smokeview Discussions

The PDPA_HISTOGRAM=.TRUE. parameter on the PDPA line creates a
histogram with PDPA_HISTOGRAM_NBINS equal width bins between
PDPA_HISTOGRAM_MAX and PDPA_HISTOGRAM_MIN. Note that for diameters the
min and max must be given in meters (not micrometers like the DIAMETER
parameter). The output .csv file contains the bin counts for each
bin. Values that fall outside the MIN and MAX are counted to the
first and last bins respectively.

To draw a traditional histogram you simply need to normalize the data
in the csv-file.

H. M. Iqbal Mahmud

unread,
Jan 31, 2012, 6:37:45 PM1/31/12
to fds...@googlegroups.com
Hi Topi,

Thanks a lot for those useful information. Now, i am clear about those
parameters.

However, i would be glad to be clear about few issues in this regard.

I have three csv output files from the simulation. Two of them are -

1. distributions_and_histograms_devc.csv
2. distributions_and_histograms_hist.csv

The first one is containing data with respect to time. The data
correspond to time is "LOGNORMAL", "ROSIN-RAMMLER",
"ROSIN-RAMMLER-LOGNORMAL" and "USER".

However, i have not found any data with respect to droplet diameter,
i.e. droplet size distribution data to draw histogram. So, would you
please explain me that what those data are and how i can get the
droplet size distribution data?

And another thing is that how i have to normalize the data to draw a
traditional histogram from the csv file.

Thanks again for your kind help and I will be glad to get further help from you.

Kind regards

Iqbal

Topi Sikanen

unread,
Feb 1, 2012, 1:55:34 AM2/1/12
to FDS and Smokeview Discussions
The data in the historgram .csv files is just the counts from each
bin. It doesn't contain the binning info. So for example from the
definitio of "LOG-NORMAL" PDPA device


PDPA_HISTOGRAM=.TRUE.
PDPA_HISTOGRAM_NBINS=20
PDPA_HISTOGRAM_LIMITS=0,2000E-6
/

There will be 20 bins (20 values in the csv file). The bins are equal
width, so each will be (2000 - 0)/20=100 microns in width. This means
that the first number in the csv file telss you the number of droplets
with diameter <100 microns. The second number in the file gives you
the number of droplets between 100-200 microns and so forth.

Normalization of the data depends on what you want to do with the
data. For traditional histogram you divide the numbers on the row with
the row sum.

H. M. Iqbal Mahmud

unread,
Feb 1, 2012, 8:10:28 PM2/1/12
to fds...@googlegroups.com
Hi Topi,

That's great. That information is very useful and i have understood
the use of those parameters. Thanks a lot for your kind help. However,
i am still not clear about some values of output file. It would be so
kind of you, if you please could make me clear about those.

Just to be confirmed that the data for histogram is in
distributions_and_histograms_hist.csv" file, isn't it? If i normalize
the data of this file, it will give me histogram. However, i am not
sure about the "distributions_and_histograms_devc.csv". What are those
data?

Another thing, why the data in "distributions_and_histograms_hist.csv"
are negative? And all the data of different distribution same, i.e.
value of "LOGNORMAL", "ROSIN-RAMMLER", "ROSIN-RAMMLER-LOGNORMAL" and
"USER" are same. The distribution of different types should be
different, isn't it?

Moreover, the initial droplet number, declared in the input file, was
10. But, the droplet number in the output file is much greater then
this number.

I would be glad, if you patiently make me understand those issues.

Thanks again for your kind help.

Regards

Topi Sikanen

unread,
Feb 2, 2012, 2:28:53 AM2/2/12
to FDS and Smokeview Discussions
The .devc file gives the PDPA sensor output. In this case its the
arithmetic mean diameter.

The values in distributions and histograms shouldn't be negative. I
don't really know how you could end up with negative values there. If
the values are negative there is something wrong. And yes the data
for different distributions should be different. If they aren't then
this is a bug.

The initial number of droplets declared in the input file is 10000 for
each particle class not 10.

H. M. Iqbal Mahmud

unread,
Feb 2, 2012, 4:16:54 AM2/2/12
to fds...@googlegroups.com
Hi Topi,

Yes, you are right. The initial number of droplets declared in the input file is 10000 for each particle class.

However, as I wrote to you, the output values in _hist.csv file are negative and those values are so large that it seems meaningless to me. The output values are constant in each bin and it is -6.28E+66. I could not understand that why this value is so large.

The output of devc.csv is also something else.

Would you please have a look in those output files sothat I can know whether I am right or wrong?

I am grateful to you for your continuous support. Thanks a lot.

Kind regards.

Topi Sikanen

unread,
Feb 3, 2012, 3:55:22 AM2/3/12
to FDS and Smokeview Discussions

This case is working for me and giving the expected results. By this I
mean that the _devc file contains the arithmetic means of the
distributions and the _hist.csv file contains correct looking
distributions.

Do other sprinkler / particle cases work for you? Do they give
sensible results?
Try adding
QUANTITIES="PARTICLE DIAMETER"
on the PART lines. Look at the case in Smokeview. Do the particle
diameters seem reasonable ?

What compiler are you using?

I suspect that there is something wrong with your build system. You
originally discovered a bug when FDS tried to write to index 0 in the
histogram bin counts array. However this can only happen if the had a
diameter equal to or under zero. So this alone seems to indicate that
the executable you are using is somehow broken.

H. M. Iqbal Mahmud

unread,
Feb 5, 2012, 6:28:57 PM2/5/12
to fds...@googlegroups.com
Hi Topi,

I could not understand the problem in my compiled version. I am using
this for other input files and it is working properly. However, it is
not working in case of distributions_and_histograms.fds.

I am using Visual studio 2008 in the project category of Intel Visual
Fortran for the compilation of FDS6 source codes. The source files are
compiled successfully without any error message.

I have changed the particle diameter and it has not any effect to the
output file. Moreover, the smokeview shows particle symbolically, it
does not show the actual diameter of the particles, isn't it?

However, do you think that there is any problem in my compilation? I
have used the guidelines found in a discussion post among Kevin,
Zatarra, Simo and others. The link of the post is -
http://markmail.org/message/pbxholypcabgsv3a#query:+page:1+mid:pu3i7bceflbscrmn+state:results.

Do you think that I should compile it using the make_fds .bat file? I
have tried to compile using the information in FDS wiki page
guidelines. However, I have not understood those instructions.

Here, i am uploading the _devc.csv and _hist.csv, which i got from my
simulation. It may help you to understand the problem and to give me
suggestion in solving this problem.

Thanks a lot for your so cordial help in solving my problem.

Kind regards

Iqbal

distributions_and_histograms_hist.csv
distributions_and_histograms_devc.csv

Topi Sikanen

unread,
Feb 6, 2012, 7:33:44 AM2/6/12
to FDS and Smokeview Discussions
I made one more fix regarding this issue (Uninitialized variable). Way
too many of these little errors in such a short piece of code.

Anyway. Your _devc file shows that the USER and ROSIN-RAMMLER
distributions have wrong arithmetic diameters. Some fixes were
recently committed to the svn. So I guess what we should try is

1) Update your source to the latest svn version
2) Clean all object flies and excecutable files left over from earlier
builds
3) Build a new executable and try it out.

H. M. Iqbal Mahmud

unread,
Feb 6, 2012, 11:46:27 PM2/6/12
to fds...@googlegroups.com
Hi Topi,

I am very happy to say that i am getting continuous support from you.

However, i have compiled the latest source file and run the fds input
file. And i am worried about the output file. The values have not
seemed reasonable to me. I have attached the csv files sothat you can
identify the bug in the source codes.

Thanks again for your so kind patience from the very beginning.

Kind Regards

Iqbal

distributions_and_histograms_hist.csv
distributions_and_histograms_devc.csv

Topi Sikanen

unread,
Feb 7, 2012, 7:09:45 AM2/7/12
to FDS and Smokeview Discussions
Everything except the 'USER' distribution seems to be working for you
now. Good. We are making progress.

If you could add CHECK_DISTRIBUTION = .TRUE. on the PART line that
defines the 'USER' part class and send me the USER_cdf.csv file that
results.

Do you have some particular application in mind for these functions?
The distribution outputs and possibly the distribution routines
themeselves are probably going to undergo some changes in recent
future. So we might have to do this all over again in the near future.

We probably should have handled this in the issue tracker instead of
filling the group with these postings but I guess it's too late for
that now.

H. M. Iqbal Mahmud

unread,
Feb 7, 2012, 6:38:53 PM2/7/12
to fds...@googlegroups.com
HI Topi,

Thanks a lot. I am happy to hear that things are progressing.

I have run the simulation and attached the USER_cdf.csv file. Are the
values of this file Cumulative Volume Fraction, F(d)? If it is then it
is also a useful parameter.

However, I am not clear about the values of _hist.csv file. Some
values are very large, like 1.43E+8. This is the droplet number in a
bin of a particular drop size, isn't it? If it is then how it could be
possible? The number of droplets in a bin (1.43E+8) is greater than
the total number of droplets (10000). Is it reasonable?

About the application of those functions, at this moment i am trying
to simulate the distribution pattern of droplets injected from a
sprinkler, i.e. distribution pattern, size distribution etc. I have
also a plan to do some experiments. Therefore, the measurement of
those functions numerically and comparison might be a very good
exercise for me.

However, you are talking that those distribution outputs and
distribution routines are probably going to undergo some change in
recent future. If it is going to be changed then still will those
functions be available? If yes, it might be very helpful for me.

Thanks again for your kind support.

Regards

Iqbal

distributions_and_histograms_USER_cdf.csv

Topi Sikanen

unread,
Feb 8, 2012, 3:13:08 AM2/8/12
to FDS and Smokeview Discussions
The values in the _cdf file are the Cumulative Number Fraction of the
particle size distribution.

It is impossible to simulate all the droplets in a spray, there are
simply too many. What FDS doe sis that each particle in the simulation
represents several particles. So even though there are only 10000
particles in the simulation, these particles represent a much larger
number of particles.

The histogram capabilities will stay. The output format will probably
change a bit.

Anyway the _cdf.csv contents seems to be correct.

H. M. Iqbal Mahmud

unread,
Feb 8, 2012, 7:31:25 PM2/8/12
to fds...@googlegroups.com
Hi Topi,

That's great. Now, most of the points are clear to me. However, just i
would be glad to get a little bit more explanation about the number of
droplets. Yes, you are right. FDS does not simulate every particle in
the domain. It simulates a number of representative droplets.

In case of a sprinkler, the flow rate and other few parameters are
declared and FDS produce a distribution of droplets. However, in the
input file of distributions_and_histograms.fds, just the number of
initial droplets is declared, not the total number. I wonder to know
that how FDS determine the number of droplets in each bin. Does it
multiply the number of representative droplets with any multiplying
factor? Where is it decleared the total number of droplets in the
input file or how is it count the total number of droplets in the
domain?

Another thing is that the _cdf.csv contents seem to be correct. But,
values in hist.csv does not seem to be reasonable.

In this mail, i have also attached some graphs of distribution of
droplet number with respect to droplet diameter. Would you please have
a look to it that it is reasonable or not?

I would like to appreciate you to solve this problem and make me to
understand the new features in FDS6. I have enjoyed your support.

Kind regards

Iqbal

distributions_and_histograms_hist.xlsx

dr_jfloyd

unread,
Feb 8, 2012, 8:39:31 PM2/8/12
to fds...@googlegroups.com
The first source for answering questions should always be to check the documentation.  How the droplet weighting factor is computed is given in the Technical Reference Guide.

Topi Sikanen

unread,
Feb 9, 2012, 7:02:04 AM2/9/12
to FDS and Smokeview Discussions
The contents of the _hist.csv file seem reasonable in the sense that
the histogram output device seems to be working correctly.

The USER distribution clearly doesn't work for you and I'm not sure
why. The Rosin-Rammler distribution seems dubious. I'll try to get
back to this in a few days. I'll post updates on the progress on these
distribution issues on issue 1571

http://code.google.com/p/fds-smv/issues/detail?id=1571

H. M. Iqbal Mahmud

unread,
Feb 9, 2012, 8:22:00 PM2/9/12
to fds...@googlegroups.com
Hi Floyd,
That's a good suggestion. I will look at the technical guide.

Hi Topi,
Thanks a lot. I wish that those issues will be solved soon and I can
use those parameter for my study. However, just to let you know that I
have run some simulation using those parameters for a sprinkler. The
diagram of size distribution pattern is seemed quite absurd to me.

Thanks again.

Kind regards

H. M. Iqbal Mahmud

unread,
Feb 26, 2012, 7:40:33 PM2/26/12
to fds...@googlegroups.com
Hi Topi,

I have come to know from your previous message that the values in the


_cdf file are the Cumulative Number Fraction of the particle size
distribution.

However, would you please let me know that how I can get the
"Cumulative Volume or Mass fraction" as csv output? I would be helpful
if I can get this value as output.

Kind regards

Topi Sikanen

unread,
Feb 27, 2012, 3:07:41 AM2/27/12
to FDS and Smokeview Discussions
There is currently no way to output the Cumulative Volume Fraction.
However it's not difficult to calculate the CVF from the CNF.

H. M. Iqbal Mahmud

unread,
Feb 27, 2012, 5:03:36 PM2/27/12
to fds...@googlegroups.com
Hi Topi,

Thanks a lot. Yes, you are right. I have already done it.

Regards

On 2/27/12, Topi Sikanen <topi.s...@gmail.com> wrote:
> There is currently no way to output the Cumulative Volume Fraction.
> However it's not difficult to calculate the CVF from the CNF.
>

Reply all
Reply to author
Forward
0 new messages