[FDS-SMV Developer Blog] FDS-SMV 6 Beta Testing

[Kevin McGrattan]

It has been over two years since our last official release of version 5 of FDS and Smokeview, and we are ready to start beta testing FDS-SMV 6. Downloads are available at the home page fire.nist.gov/fds. Scroll down to the bottom of the page for the appropriate links.

Our primary interest in beta testing the new release is to determine if there is sufficient documentation to convert FDS 5 input files into FDS 6 input files. In general, the major parameters are the same, and the basic syntax of the input file is the same. However, some of the sub-models are significantly different, and there is no easy way to maintain perfect backward compatibility in FDS (Smokeview, in general, is backward compatible). By issuing error statements at start up, we want to alert you to important changes, which is why we prefer to stop the program with an error statement rather than accepting an outdated parameter or construct.

For those interested in helping with the beta testing, download the latest FDS 6 "release candidate" and try to run one of your old cases. Chances are that you will receive an error message. Chapter 1 of the new FDS User's Guide has a table that lists all of the parameters that have changed from FDS 5 to 6. We would like you to tell us how easy or difficult you find the conversion process, and whether or not the changes are well-documented. If you would like to be listed as a beta tester, send us via an email your name and affiliation, and we will compile a list in the new FDS User's Guide.

You can report your findings directly to us via email, or you can also start an Issue Tracker or Discussion Group thread. The latter two are preferable so that we can all learn about the changes in FDS and Smokeview. This will also provide us with the opportunity to describe in more detail the changes that have been made. Many of these changes are "under the hood," so to speak; that is, there have been changes to core algorithms that may not be apparent at first glance. But we can discuss these improvements as we look at your cases.

--
Posted By Kevin McGrattan to FDS-SMV Developer Blog at 11/04/2012 02:06:00 PM

[Barbro Maria Storm]

2012/11/4 Kevin McGrattan <mcgr...@gmail.com>:

> It has been over two years since our last official release of version 5 of
> FDS and Smokeview, and we are ready to start beta testing FDS-SMV 6.

Question about FDS6 and MPICH2 (as discussed in user guide of FDS6,
chapter 3.1.2) - I've had difficulties using FDS5 with right about any
version of MPICH2 from Argonne(1) exept old versions like 1.2.1 on
Windows Vista and 7.

This (v. 1.2.1 of MPICH2) seems to not be working with FDS6.

Has anyone found a version of MPICH2 that works with both? Does 1.4.1
work for anyone with FDS5?


1)http://www.mcs.anl.gov/research/projects/mpich2/index.php

--
Barbro Storm

[Kevin]

The MPI calls in FDS 5 only worked with the older version of MPICH2. We believe that this problem is fixed in FDS 6. We do not know if FDS 6 works with the old version of MPICH2. Here at NIST, we use a linux cluster using OpenMPI.

It would be interesting if someone currently using the old version of MPICH2 can run FDS 6.

[Simon Ham, FiSEC, UK]

Dear Kevin,

I have had a go at running one of my old models using FDS6 beta on my 64bit Windows 7 machine. It took a little time to get the gas burner component working as I had failed initially to incorporate the required &REAC line for the fuel.

Two problems have immediately manifested themselves. I apologise in advance if I have missed changes in the user's guide.

The first was QUANTITY FED is not appropriate for a SLCF. Has the syntax been changed for this?

I deleted all three of the FED slice lines in the Input file and it looked as if it was going to start running having  printed the head title. I then get forrt1:severe (157) Program Exception - access violation. I love this kind of message implying I have endeavoured to assail the virtue of my central processor unit, however this is outside my ability to resolve.

Any ideas?

Regards,

Simon J. Ham
Director, Fire Safety Engineering Consultants, UK

 

[Craig Weinschenk]

If you specify FED in an input file, then you are limited to a DEVC since FED is a time integrated quantity. This is the same as 5.5.3 syntax. New to FDS/SMV 6, if you want to display FED as a slice file, you need to specify three slice files one each for O2, CO2, and CO at each desired slice location. SmokeView will postprocess the species concentrations into an FED slice. Currently, the FED slice files are limited to the impact of these three species. You can also generate 3D slices files of these species and SmokeView will generate iso-surfaces, however as warning these files can become quite large.

 

--
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.

To view this discussion on the web visit https://groups.google.com/d/msg/fds-smv/-/KuVO4lvUfS4J.

For more options, visit https://groups.google.com/groups/opt_out.
 
 

--
Craig Weinschenk

NRC Postdoctoral Research Associate

Fire Engineering Division

National Institute of Standards and Technology
www.cweinschenk.com

[dr_jfloyd]

Post your input that got the error to the tracker.  If you can simplify it and still get the error that will help us.  That error is likely either a bug or some invalid input that we aren't trapping more gracefully.

[Daniel_eval_firegas]

Dear Kevin,

First of all, thank you and the FDS-Developers Community for your effort. I believe in the performance of your work as I'm testing different simulationtools.
Hope you are all motivated from the userfeedback. I guess you will get a lot of feedback after X-Mas.
I'll test FDS6 with measurement results and will post it in the next months. Especially Gas vs Smoke is an interessting topic.


Just a question about OpenMP: Is it supported with a command in FDS6 or will there be an other compiled version like in FDS5_openmp_winxx.
Maybe its a bit early for my question.

Thank you and kind regards
Daniel

[Kevin]

We hope to have an optional OpenMP build of FDS 6. We haven't yet tested it.

[Mohamed ASSAL]

Dear All,

Dear Kevin

Please add me to FDS6 testers list.

Mohamed ASSAL

CFD Engineer

CFD_Algeria.



2012/11/6, Kevin <mcgr...@gmail.com>:

> --
> 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.
> To view this discussion on the web visit

> https://groups.google.com/d/msg/fds-smv/-/DuAAQRr8u80J.

[Simon Ham, FiSEC, UK]

Jason has found and fixed the bug that was effecting my previous attempt at running one of my standard models.

I deleted the relevant affected slice and the model has started to run.

I now have a new problem however, paragraph 16.8 in the Users Guide suggests that there should be default outputs for Soot mass fraction and HRRPUV. When I run Smokeview these do not appear. Have I missed some inputs or does a Propane Burner burn too cleanly. The file I am running is almost identical to the one posted to the Tracker

I would offer some first impressions of FDS6 beta. It somehow feels very different. It appears to be slower probably because it is doing more. It offers a far greater level of control with more sophisticated inputs and therefore requires a more comprehensive  knowledge and technical understanding of fire and combustion. Trying to be clever, acting like a teenager with a new computer game, I thought I would skip printing the Technical Users Guide which was an elementary and stupid mistake.

If there are any UK or European users who would be interested in contributing and participating in a non-profit making weekend conference on using FDS6 I am prepared to set up such an event perhaps in January or February 2013. Please e-mail me directly to register your interest.

Simon J. Ham

[Kevin]

In FDS 6, we now require you to include a REAC line if you are
modeling a fire. You also must specify a SOOT_YIELD if you want to see
smoke. Previous versions of FDS used a default fuel and soot yield,
but we feel that this decision ought to be the user's, not ours.

As far as FDS 6 in general, while there are lots of new capabilities,
the basics are still the same. If all you are doing is simulating a
fire with a specified HRR, the input file should not be much different
at all.

On Nov 7, 11:25 am, "Simon Ham, FiSEC, UK" <simon...@btconnect.com>
wrote:

[Simon Ham, FiSEC, UK]

Dear Kevin,

I have amended the input file to incorporate the following lines and re-run it:

&SURF ID='FIRE', HRRPUA=503.,PART_ID='smoke', COLOR= 'FIREBRICK' /
&REAC     FUEL             =     'PROPANE'
    SOOT_YIELD        =     0.01
    CO_YIELD            =    0.02
    HEAT_OF_COMBUSTION    =    464124. /
&VENT XB=0.4,2.5,0.4,2.5,2.5,2.5, SURF_ID='FIRE' /
&PART ID='smoke',  MASSLESS=.TRUE., SAMPLING_FACTOR=1, COLOR='BLACK', AGE=10. /

I am still not getting either the soot mass fraction or HRRPUV when I run Smokeview

There is obviously a conflict or syntax error here, undoubtebly simple but fundamental, but I cannot spot it.

Regards,

Simon J. Ham, FiSEC, UK

[Kevin]

Could you start an Issue Tracker for this, and attach the entire file.
I don't see anything obviously wrong here, and I need to run the case.

On Nov 8, 3:27 am, "Simon Ham, FiSEC, UK" <simon...@btconnect.com>
wrote:

[Paul Williams]

Hi Simon

 

I don't know whether it is the entire problem but your HEAT_OF_COMBUSTION = 464124. is an order of magnitude of 10 greater than propane and most other fuels.

 

Try running it with a reduced HEAT_OF_COMBUSTION and see if that fixes the issue.

 

Kind regards,

 

Paul

[danie...@gmail.com]

Hello Kevin,

Great work! I would like to be added as a Beta tester.

Cheer, Daniel Bak

Sent via BlackBerry from SingTel!

---

From: Kevin McGrattan <mcgr...@gmail.com>

Sender: fds...@googlegroups.com

Date: Sun, 04 Nov 2012 19:06:21 +0000

To: <fds...@googlegroups.com>

ReplyTo: fds...@googlegroups.com

Subject: [fds-smv] [FDS-SMV Developer Blog] FDS-SMV 6 Beta Testing

--

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.

[Kevin]

Thanks to all of you who have volunteered to be beta testers. Let me
explain a bit more about what we mean by this. As Kris Overholt
pointed out in our latest blog post:

http://blog.fds-smv.net/2012/11/resources-for-fds-smv-development-and.html

we have developed an automated process for testing FDS and Smokeview
which mainly involves running verification and validation test cases.
The former are the sample cases that you get when you download FDS/
SMV; the latter are simulations of a wide variety of experiments. The
verification cases are run and checked each night using an automated
program script we call "firebot." The "bot" is short for robot, which
is a slang term for what is formally known as "continuous
integration." A popular program for doing CI is Buildbot, and you
often see others play on the suffix "bot." Anyway, what this really
means is that we have a pretty good system for testing algorithms
within FDS and Smokeview, but we do not have a formalized system for
assessing usability. That is where you come in. We try to design the
FDS input file and the Smokeview menus to be as intuitive as possible,
but what makes sense to us does not necessarily make sense to you. So
we rely on feedback from users to determine if new features are easy
or difficult for people to use.

The beta testing has two aims: (1) catch bugs that our test cases
might have missed, and (2) assess usability. For both of these, we
would like beta testers to open up a new Issue Tracker to report their
findings. Label the Issue something like "FDS/SMV Beta Testing --
Whatever" to let us know what the post is about. At the top, report
which release candidate you are using, your operating system (like
Windows 7, Mac animal, flavor of linux), and a general description of
the calculation you ran. Give a brief summary of how easy or difficult
it was to convert your old input file to FDS 6. Add a comment about
Smokeview if you notice something different, good or bad. And if you
find a bug, report it here. Then we can assign the Issue to the
appropriate person who will work with you until the Issue is resolved.

Finally, remember to give your full name, affiliation, and location so
that we can add you to the list of beta testers that we will maintain
in the FDS User's Guide.

[mop-mop]

MPICH" version "mpich2-1.3.2p1-win-x86-64" works for FDS6.

[Nat Ong]

Hi Kevin,

I would like to be added to beta testing group. Natalie Ong, Arup Fire Singapore, Analyst.

I have playing around with a model by using FDS6. Surprisingly,I can get the model started without changing any command line (Not even include the magic word FUEL in the reaction line! Instead, I put in the formula of C,H,O,N respectively.) 

I am interested to see the soot deposition to the surface and have started a test run. Just wonder if there is any direct co-relation/impact between this and visibility of smoke near wall/floor/ceiling in FDS6? Does visibility slices loaded take into account the new feature of soot deposition in FDS6?  

Next, I have a read at the User Guide and get confused with the term "CO_PRODUCTION" and "CO_YIELD". is there any difference between these two terms? I am refer to CO_PRODUCTION from User Guide page7, Table 1.1: changes to input parameters FDS Version5-6 and CO_YIELD in page120&page234. 

Thanks,

Natalie

[dr_jfloyd]

Yes, any soot removed by deposition is no longer in the gas and hence will not contribute to a reduction in visibility.  Note: the deposition model should still be considered as under development.

CO_PRODUCTION in FDS 5 referred to a two-step combustion model for CO formation and destruction. That model has been removed and replaced with a recommended set of reaction definitions.  CO_YIELD refers to the amount of post-flame CO that is produced for the single step reaction.

[Julio Cesar Silva]

Hi to All,

Kevin and the others FDS Developers, Thank you so much for your work!

Until now, I am beta testing the User’s Guide and I would like to
present my initial considerations.
1- In 1.2 “What’s New in FDS6” – This item will gain a much more value
if you could add a link or a page number where we can find a deeper
description about each item. Or a procedure used later in Table 1.1
where the section which have the information is listed sided to the
information.
2- Congratulations in adding the feature of choose the correspondence
between mesh and mpi process.
3- I have difficulties in found the information about the Immersed
Boundary Method -
https://groups.google.com/group/fds-smv/msg/3a43c731b8d6ae6b - Is it
working on FDS6?
4- I think the User’s Guide should have more information and guidance
about (maybe highlight) the parameters “measure of turbulence
resolution” and “normalized wavelet error measure” as exposed by
Randy McDermott in the paper - Quality Assessment in FDS: A Bridge to
Reliable Simulations -
http://www.thunderheadeng.com/2011/08/d1-4-mcdermott/

About the FDS6 executables:
We use a linux (fedora) cluster and I would like to compile the
executable from the source code as I always did with FDS5.
I have downloaded the version r13632 via SVN. However the source code
is not working with the makefile.
When I tried to compile the code, I get error #6731 in the func.f90.
../FDS_Source/func.f90(591): error #6731: Object is not a pointer object [WC]
WC%NOM_IB(1:) => OS%INTEGERS(I1:I2,STORAGE_INDEX) ; IF (NEW) WC%NOM_IB(1:6) = 0
Has someone the same message or found how to proceed through this error?

I have also tried the precompiled version of the release candidate,
however to run the simulations with MPI, there are a lot of changes to
make in all machines.

Is the precompiled version the only way to get FDS6?

In time: My data to be included as a beta tester:
Julio Cesar Silva, Civil Engineering Program, COPPE, Federal
University of Rio de Janeiro, Brazil.

Again, thank you all for this amazing work in fire simulation community,
Sincerely yours,

Julio Cesar

[dr_jfloyd]

Re compilation: You need to update your compiler; it is not compliant with the current Fortran standard.

[Andrew Louie]

If you are using Intel Fortran Compiler version 11, it is a bug with the compiler that Intel fixed in their latest release. (I think Intel should provide the fix for fortran 11 instead of forcing customers to have to pay for a new version).

http://software.intel.com/en-us/forums/topic/295038

--

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.

To view this discussion on the web visit https://groups.google.com/d/msg/fds-smv/-/VCC9EexB1s8J.

For more options, visit https://groups.google.com/groups/opt_out.
 
 

--
-Andrew Louie :wq

[Julio Cesar Silva]

Thank you Andrew and Jason,

I updated my intel compiler and MPICH2.
FDS (serial and mpi) r- 13921 sucessfuly compiled using MPICH2 1.4.1
and IFORT 12.

Thank you,
Julio

2012/11/29 Andrew Louie <lou...@gmail.com>:

Julio Cesar Gonçalves da Silva, M.Sc.
Doutorando - Programa de Engenharia Civil - COPPE/UFRJ
Laboratório de Estruturas e Materiais - LABEST
(21) 2562-8493 Ramal 31
(21) 8886-1235

[Nat Ong]

Hi All,

I have recently carried a few modelling to find out more about the new function of soot deposition in FDS6.

It's a simple box with low level makeup and high level opening for ventilation purposes.

Test 1 (soot1.fds)

AEROSOL=.TRUE.

BNDF SOOT=on

In this model,i turned on both the command line as to model soot deposition to walls of a box and expected to see similar result as in Figure 12.3,page 133 of User Guide.

Observation: 

• No result on smoke visibility is shown. 
• No result of 3D smoke is shown
• No result of soot boundary files.
• Smoke temperature looks OK.
• HRR of the prescribed fire looks normal. 

I have attached the input file for your reference.

Test 2 (soot2.fds)

AEROSOL=.FALSE.

BNDF SOOT=off

Observation:

• No result on smoke visibility is shown.
  
• No result of 3D smoke is shown
  
• No soot boundary files.   --->   This is expected
  
• Smoke temperature looks OK.
  
• HRR of the prescribed fire looks normal.

Test 3 (soot3.fds)

AEROSOL=.TRUE.

BNDF SOOT=off

Observation:

• No result on smoke visibility is shown.
  
• No result of 3D smoke is shown
  
• No soot boundary files.   --->   This is expected
  
• Smoke temperature looks OK.
  
• HRR of the prescribed fire looks normal

From test 2 and test3, it seem like there is no difference in term of output generated no matter aerosol is turned on/off.

I am guessing this is due to the fact that we are just prescribing aerosol and didn't have output to measure it, and hence we can't observe its impact.

However, i couldn't understand why does the prescription of AEROSOL species to our model will halt the 3D smoke activities and even smoke visibility while the HRR and smoke temperature look fine.

Besides, could you please have a look at my model and advise if any line of codes in Test1 leads to wall soot deposition function fail?  

Understood soot deposition function is still under development but i am still welcoming any thoughts/ideas. 

Thanks,

Natalie   

On Monday, November 5, 2012 3:06:23 AM UTC+8, Kevin wrote:

[Nat Ong]

Due to some network problem, i couldn't upload any files for you to review however the input file for soot1.fds is as below:

 

&HEAD CHID='Soot1', TITLE='1MW steady-state fire in a box'/

&DUMP RENDER_FILE='Soot1.ge1', DT_DEVC=120, DT_PL3D=120, DT_RESTART=1200, NFRAMES=1200, PLOT3D_QUANTITY='TEMPERATURE','U-VELOCITY','V-VELOCITY','W-VELOCITY','VISIBILITY'/

&MISC TMPA=32.00/

&TIME T_END=600/

&MESH ID='MESH01', IJK=160,150,20, XB=-2.00,62.00,-2.00,58.00,0.00,8.00/

&REAC ID='POLYURETHANE',

      C=6.30,

      H=7.10,

      O=2.10,

      N=1.00,

      IDEAL=.TRUE.,

      HEAT_OF_COMBUSTION=2.3000000E004,

      SOOT_YIELD=0.10/

&SPEC ID='SOOT',AEROSOL=.TRUE./  

&SURF ID='FIRE',

      HRRPUA=500/

&VENT SURF_ID='FIRE', XB=22.00,24,2.50,3.50,0.4,0.4, COLOR='RED'/ fire Vent 1MW

&OBST SURF_ID='INERT', XB=22.00,24,2.50,3.50,0,0.4, COLOR='GRAY 60'/ 

&VENT SURF_ID='OPEN', MB='XMIN'/ Mesh Vent: MESH [XMIN]

&VENT SURF_ID='OPEN', MB='XMAX'/ Mesh Vent: MESH [XMIN]

&VENT SURF_ID='OPEN', MB='YMAX'/ Mesh Vent: MESH [YMAX]

&VENT SURF_ID='OPEN', MB='YMIN'/ Mesh Vent: MESH [YMIN]

&VENT SURF_ID='OPEN', MB='ZMAX'/ Mesh Vent: MESH [ZMAX]

&OBST XB=0.00,60.00,0.00,0.50,0.00,5.0, COLOR='GRAY 60', SURF_ID='INERT'/ Wall

&OBST XB=0.00,60.00,49,50,0.00,5.0, COLOR='GRAY 60', SURF_ID='INERT'/ Wall[1]

&OBST XB=59,60.00,0.00,50,0,5.00, COLOR='GRAY 60', SURF_ID='INERT'/ Wall[1]

&OBST XB=0,1.00,0.00,50,0,5.00, COLOR='GRAY 60', SURF_ID='INERT'/ Wall[1]

&OBST XB=0,60,0,50,4.50,5.00, COLOR='GRAY 60', SURF_ID='INERT'/ ceiling

&HOLE XB= 30,34,30,34,4,5/

&HOLE XB= -1,1,0,50,0,0.5/

&DEVC ID='THCP 1', QUANTITY='TEMPERATURE', XYZ=23,3,2.00/

&DEVC ID='THCP 2', QUANTITY='TEMPERATURE', XYZ=23,3,3.00/

&DEVC ID='THCP 3', QUANTITY='TEMPERATURE', XYZ=23,3,4.00/

&DEVC ID='THCP 4', QUANTITY='TEMPERATURE', XYZ=23,3,5.00/

&BNDF QUANTITY='SURFACE DEPOSITION', SPEC_ID='SOOT'/

&SLCF QUANTITY='VELOCITY', VECTOR=.TRUE., PBY=2/

&SLCF QUANTITY='VELOCITY', VECTOR=.TRUE., PBY=7/

&SLCF QUANTITY='VELOCITY', VECTOR=.TRUE., PBZ=4.5/

&SLCF QUANTITY='VELOCITY', VECTOR=.TRUE., PBZ=2/

&SLCF QUANTITY='VISIBILITY', PBZ=2.5/

&SLCF QUANTITY='TEMPERATURE', PBZ=2.5/

&SLCF QUANTITY='VISIBILITY', PBX=23/

&SLCF QUANTITY='TEMPERATURE', PBX=23/

&SLCF QUANTITY='VISIBILITY', PBX=35/

&SLCF QUANTITY='TEMPERATURE', PBX=35/

&SLCF QUANTITY='VISIBILITY', PBY=3/

&SLCF QUANTITY='TEMPERATURE', PBY=3/

&SLCF QUANTITY='VISIBILITY', PBY=5/

&SLCF QUANTITY='TEMPERATURE', PBY=5/

&SLCF QUANTITY='VISIBILITY', PBY=7/

&SLCF QUANTITY='TEMPERATURE', PBY=7/

&SLCF QUANTITY='VISIBILITY', PBY=3.78/

&SLCF QUANTITY='TEMPERATURE', PBY=3.78/

&SLCF QUANTITY='VISIBILITY', PBZ=2.1/2.1m ab FFL

&SLCF QUANTITY='VISIBILITY', PBZ=2.5/2.5m ab FFL

&SLCF QUANTITY='TEMPERATURE', PBZ=2.5/2.5m ab FFL

&TAIL /

[Kristopher Overholt]

Hi Natalie:

Thanks for testing out the new features in FDS v6. As mentioned in the group before, the soot deposition routines are in a developmental/experimental/testing phase.

The issue with your input file is that, you are mixing reaction notation in the input file. You are specifying a reaction with primitive species while also defining a SOOT species that is not being produced by your primary reaction. The specified soot yield of 0.10 in your reaction is not producing the species SOOT with AEROSOL=.TRUE. that you specified. Soot is being produced, but not the SOOT species that you specified.

To explicitly specify soot and have that species generated as part of your products, you need to specify the chemical reaction as in the example file Species/propane_flame_deposition.

I recently created a calculator to help you specify a soot yield in a reaction with a defined SOOT species:

http://www.koverholt.com/chemical-equation-balancer/

So, for example, if you wish to specify polyurethane as you have defined, you would enter "C6.3H7.10O2.10N1.00" as the fuel, with a soot yield of 0.10. Then you would modify the REAC and SPEC lines to produce the amount of soot and other products accordingly (see the Species/propane_flame_deposition example). I will be adding outputs from that calculator soon to output FDS input lines, but for now you can enter the stoichiometric coefficients manually. That way your SOOT species will actually be produced, and the deposition routines will act on them accordingly.

Kris

--

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.

To view this discussion on the web visit https://groups.google.com/d/msg/fds-smv/-/cdNR7Ww-ocQJ.

For more options, visit https://groups.google.com/groups/opt_out.
 
 

--
Kristopher J. Overholt
Ph.D. Candidate | Fire Research
The University of Texas at Austin
http://www.koverholt.com
Mobile: (832) 736-3473

[Franck Didieux @ LNE]

Hi Kevin,

I'd like to join the team of beta-testers for the v6. I have a couple of test cases that I think about as good candidates to test v6 on (and a specific one that gave me a hard time when going from v4 to v5 : I'll definitely test it in v6 !).

Best regards,

[Konrad Wilkens]

Hi Kevin,

I would like to join the team of beta testers for FDS6, I have just recently joined Lund University as a PhD candidate in Fire Safety Engineering and i think we have a few good cases that also have experimental data which can be compared against the FDS6 outputs. 

Especially, we have data from a full-scale test that is performed here every year as part of the bachelor degree, students completing this degree as part of one of their subjects (simulation of fires) then compare an FDS simulation to the experimental data they got. This means I have lots of previous data to compare against a simulation using FDS6 (and also to test the ease of converting files from FDS5 to 6).

Also as part of the research i will be undertaking, there will be lots of times when i get experimental data that may be used as validation studies, so I would like to help out in anyway that i can.

Regards,

Konrad Wilkens - PhD Candidate, Lund University, Sweden 

[Kevin]

We would like to release FDS 6 officially soon. Unfortunately, many
have offered to be beta testers, but to date only about 10 people have
actually reported any results. I tried to make it clear to people that
all we want is for them to re-run an existing case that works with FDS
5 to ensure that we have properly documented the changes in the input
parameters.

The validation study is something that can be done at any time. We
have enough existing validation data to check FDS 6. In the longer
term, it would be nice to add more data sets to our collection. So the
beta testing and validation study are two separate things. We would
like to see results of the beta testing within the month, and we can
take a longer time for validation work.

Concerning validation studies -- again, many users have offered to do
validation work for FDS. However, 9 out of 10 times we do not ever see
the results of these studies. We assume that there is a masters or PhD
thesis written, and sometimes there is a paper, but in most cases we
do not get copies and we cannot possibly read all those journals. We
would much prefer, and I cannot emphasize this enough, that you work
closely with us (on-line or off-line, your choice) so that your
validation study can be added to our Validation Guide. Sometimes
students send us a gigantic Excel file after they graduate and say
"here you go, here is some data!" If the experiments are of value
(properly documented, well-controlled, publicly available), we have to
redo everything from scratch. If you work with us, we can tell you how
to set up the input files and experimental data files so that the
processing of the results can be done automatically. The entire FDS
Validation Guide is processed with a script. We cannot possibly re-run
all these cases and make all these plots one by one. It is done with a
script, and the way this all works will be of great interest to you
because it involves some pretty slick numerical techniques that will
serve you well in life. So consider working with us throughout your
project. Please do not give us a paper or your thesis at the very end
-- at that time it will be too late to implement the "hooks" necessary
to automate the processing of the simulations and experimental data.

[Andrew Louie]

Regarding validation and automation:

Everything about FDS is open source except for matlab. Is there any movement to adopt an open source tool in place of matlab?

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.

To post to this group, send email to fds...@googlegroups.com.

For more options, visit https://groups.google.com/groups/opt_out.

--
-Andrew Louie :wq

[Kevin]

You are confusing the terms "open source" and "free".

We adopt the best tools for the job at hand. We have not found a
better mathematical analysis/plotting package than Matlab for our
purposes. We use LaTeX for word processing not because it is free, but
because it is far superior to Word in almost all respects. We use a
commercial Fortran compiler from Intel because it is the fastest, best
supported Fortran compiler that we have tested. We know that there are
free Fortran compilers, but they are not as good.

All this being said, if you want to suggest a free alternative to
Matlab, we (or you) could try it on our V&V scripts.

> > For more options, visithttps://groups.google.com/groups/opt_out.
>
> --
> -Andrew Louie :wq

[Andrew Louie]

Well I mean open source in that the FDS/SMV code itself is open source, I can read and compile the source.

SmokeView renders using OpenGL which is also open source

the makefiles  are open source, and the make utility itself is open source, The LaTeX code for the manuals is open source and LaTeX itself is open source.

The entire project is stored/managed with subversion which is also open source. 

that leaves matlab as the oddball :)

Anyway I am interested in finding an open source alternative to the matlab scripts, it's on my interests list. I'll keep you updated if I find anything worthwhile. 

For more options, visit https://groups.google.com/groups/opt_out.

--
-Andrew Louie :wq

[Kevin]

By your definition, Matlab is open source too, I think. I believe that
all the Matlab intrinsic scripts are readable. Matlab is the graphical
user interface and the compiler. FDS has a non-free GUI called
PyroSim. FDS uses non-free compilers from Intel. So, again, open
source does not necessarily imply that something is free of cost. We
all need to get paid in one way or another.

Good luck with your pursuit of a free Matlab. You might find that you
get what you pay for.

[dr_jfloyd]

All the matlab scripts are text files.  With some effort you should be able to convert them over to some other tool provided that the other tool has the same set of capabilities. 

Early on we had tried the route of using open source plotting packages.  None of them quite gave us the control over plots that Matlab gave us and none of them were as easy to use as Matlab was.  Additionally, most of the developers already had some level of proficiency with Matlab which means that it is easy for us to create new scripts vs. some other tool where we would have to spend a lot of time learning the tool first.

open source means just that.  That the source is made available.  All of our V&V inputs, V&V data, manuals, source codes, and scripts are all made available. 

[Lukas A.]

Might Python be an alternative here? I have been using Matlab a lot, however since a couple of years I switched to Python and would (in general) recommend python for tasks like those addressed here.

Btw, currently I use Python for FDS input file generation and for FDS output analysis.

Best,
Lukas

[Randy McDermott]

I am sure Bryan Klein is smirking as he reads this thread.  :-)

There is a history here.  In the very earliest V&V scripts, Bryan set these up in Python.  When he joined Pyrosim, we had to scramble to recreate the process in a hurry because neither Kevin nor I were Python experts.  But I am pretty good with Matlab.  Thus, it is my fault that everything got moved to Matlab.  From there things have grown to the point where a conversion back to Python will be somewhat painful... not impossible, but nontrivial.  The good news is that I made it a point this year to learn Python and at this point I am getting pretty comfortable with it.  And actually I now believe it is much more powerful than Matlab.  Matlab has problems we have not been able to overcome (which I won't go into here).

So, Lukas, I completely agree with you.  My preference (and I know Kris Overholt concurs) is to transition back to Python.  But this will not happen overnight.  We have a couple of things to do: first we have to build a complete clone of the current V&V process in Python.  Then it is also necessary that the other developers become as handy with Python as they are with Matlab.  Again, this is all possible, but won't happen overnight.

[Randy McDermott]

One more thing Kevin just reminded me of: the pdf plots produced by Matplotlib are about 100KB whereas the Matlab pdf plots are about 6KB.  This is a show-stopper because were are literally making thousands of plots for the validation guide.  I am sure there is a solution, but definitely something we need to sort out.

[Glenn Forney]

did you mean the plots produced by python are 100KB (what is Matplotlib?)

To view this discussion on the web visit https://groups.google.com/d/msg/fds-smv/-/yIaWoUcTTTYJ.

For more options, visit https://groups.google.com/groups/opt_out.
 
 

--
Glenn Forney

[Mohamed ASSAL]

http://matplotlib.org/

2013/2/28, Glenn Forney <gfo...@gmail.com>:

>>>> >>>>> visithttps://groups.google.**com/groups/opt_out<http://groups.google.com/groups/opt_out>.

>>>>
>>>> >>
>>>> >>>> --
>>>> >>>> -Andrew Louie :wq
>>>> >>
>>>> >>> --
>>>> >>> You received this message because you are subscribed to the Google
>>>> Groups
>>>> >>> "FDS and Smokeview Discussions" group.
>>>> >>> To unsubscribe from this group and stop receiving emails from it,
>>>> send an
>>>> >>> email to fds-smv+u...@googlegroups.com.
>>>> >>> To post to this group, send email to fds...@googlegroups.com.
>>>> >>> For more options,

>>>> >>> visithttps://groups.google.**com/groups/opt_out<http://groups.google.com/groups/opt_out>.

>>>>
>>>> >>
>>>> >> --
>>>> >> -Andrew Louie :wq
>>>> >
>>>> > --
>>>> > You received this message because you are subscribed to the Google
>>>> Groups "FDS and Smokeview Discussions" group.
>>>> > To unsubscribe from this group and stop receiving emails from it,
>>>> > send
>>>> an email to fds-smv+u...@googlegroups.com.
>>>> > To post to this group, send email to fds...@googlegroups.com.
>>>> > For more options, visit

>>>> > https://groups.google.com/**groups/opt_out<https://groups.google.com/groups/opt_out>.

[Lukas A.]

Python itself is a "very sparse tool". However, it comes with A LOT of additional modules, like matplotlib, which allows to generate very nice scientific figures.

Lukas

[Lukas A.]

Randy,

> There is a history here.

as always ;-)

> The good news is that I made it a point this year to learn Python and at this point I am getting pretty comfortable with it.

Sounds good. Python is a very nice language, and very easy to learn -- however very powerful due to the available modules (you can even do parallel processing, e.g. via MPI, with Python).

> So, Lukas, I completely agree with you. My preference (and I know Kris Overholt concurs) is to transition back to Python. But this will not happen overnight. We have a couple of things to do: first we have to build a complete clone of the current V&V process in Python. Then it is also necessary that the other developers become as handy with Python as they are with Matlab. Again, this is all possible, but won't happen overnight.

As I am doing a lot of programming (including Python), I'm happy to give you a hand here -- only if needed of course. However, I am quite occupied till end of April. But I could put it on my agenda and we could organize the first steps -- just get in touch with me (hpc.o...@gmail.com).

Lukas

> To view this discussion on the web visit https://groups.google.com/d/msg/fds-smv/-/d2TV5Y4YxTIJ.

[Bryan Klein]

There were methods to my madness. :)

On Thursday, February 28, 2013 6:13:11 AM UTC-8, Randy McDermott wrote:

I am sure Bryan Klein is smirking as he reads this thread.  :-)

Also, there is a difference between 'Open File Formats' and 'Open Source' tools.  You are all correct when you say that the .m scripts are in an open format, but this is not the same as open source.  Matlab is a closed source commercial program, just like other tools Excel and PyroSim come to mind.  Python is open source, open format and non-commercial.

I am not sure when the PDF file sizes from Matplotlib were tested last, but it has been a few years since the switch to Matlab from Python and things might have been updated in the module since then.  I seem to remember it had something to do with embedding fonts into the resultant PDF files.  I bet that it is much nicer to use now, than it was back then.

Anyway, my old Python scripts should be in the repository.  I am sure that you can make them much better now, but they might be a good place to start as a reference.

-Bryan

[Randy McDermott]

Bryan,

I tested the Matplotlib pdf size yesterday... same result.

R

[Bryan Klein]

I wonder if this new PGF feature would help with file size issues.

http://matplotlib.org/1.2.0/users/pgf.html#pgf-tutorial

-Bryan

[Daniel Gutmacher]

Hi all,

Python is a very cool open source tool which I recommend to everyone. I use it every day for creating measurement tools (with gui, wxglade) and also for plotting and calculations. But most often I use matlab for nicer plots and publication data. Till now I didn't get the python matplotlib to the same level as in matlab. But I do have to switch completely one day to python, it's worth!

but you have to be aware if you want to switch that the following things will take time:

contra or awareness:

- installation process of python is horrible: you need many packages ( depending on what you want to do), you need to set up the python path manually and correct ( it's not for everyone and not so easy to use like matlab)

- python updates and different versions are not not always compatible. there are different versions at that point, I used python 2.6. Independent there is now python 3 officially released as the new python . the most important for analyzing and calculating FDS results is to take the same import math libraries. you have to be sure that the FDS community uses the same functions to get the same results! now can you guarantee that? you could if you upload the mani script and that has always to be used and updated correctly to actual python versions.

- as a matlab user it take some time to get into the object code of python

- students and universities are important for FDS evolution. They usually use matlab and don't know python ( that's a sad marketing strategy of matlab, by the way: same thing for labview marketing). when I was a student I only knew matlab, labview and some programming of java and c. 

- matlab is better and faster for calculations ( it was made for that)

- matlab plots look nicer

PRO for python:

- it's open source and free

- programming style is more object and syntax is very easy

- you will love it 

- you can create exe files with all included packages !!! (for free)

- it has many packages and can interact with all kind or other things ( platform plattform independen, internet, mysql, drivers, parser, etc.....)

so et the end i think you should switch to Python for yourself. and ich a Student wants to do Some validation with officially FDS Python scritps then he will install Python. but Some will not.

Thats my opinion, what do you think about?

best wishes 

Daniel

PS www.python.org

--

You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msg/fds-smv/-/qsrLJuzhigoJ.

[Kristopher Overholt]

Hi Daniel:

Thanks for your comments. I agree with most of your points, and I also use Python in my day-to-day research workflow including data acquisition, data analysis/reduction, generating publication-quality plots, automating a large number of FDS and CFAST simulations, Bayesian inference and other statistical applications, and more. Also, it is worth mentioning that some of the third-party FDS tools, such as BlenderFDS and my website calculators, are coded in Python.

Taking from your list of cons, I'd like to add some notes about my experiences with Python (and for others that may be following along). One thing that addresses about half of your cons is to use one of the more popular bundled Python distributions. My favorite is EPD Free from Enthought, which is a free and cross-platform distribution that includes Python, numpy, scipy, matplotlib, ipython, and a few other common packages.

All of our computing clusters here at UT Austin have the full version of EPD installed, and Enthought also offers a free, full-featured academic version of EPD for university students and faculty. I have been using EPD Free for a few years now, and it eliminates the problems with installations and versions that you mentioned. And you can make use of the PyPi package repository (using a tool like pip) to easily install extra modules and packages.

Also, the performance of Python is quite fast when using packages such as NumPy and SciPy, which include matrix math and numerical optimization tools that run on optimized C and Fortran code. There are also a number of guides for Matlab vs. Python usage, such as "Numpy for Matlab Users", which show equivalent commands in Python (which are often the same as their Matlab counterpart). Finally, I have found that the Python and scientific Python community are quite vast and helpful resources if one runs into problems along the way.

All that being said, for your last point about students doing validation work with FDS, it is recommended that a student or researcher planning to do V&V work contact the FDS development team early on (as Kevin mentioned in a previous blog post). This way, whatever the methods we are using (which are all well documented on the FDS-SMV Wiki Pages), the efforts and FDS cases can continue working as an integral and ongoing part of the V&V suite as opposed to only existing in a publication or other static writeup.

Kris

[dr_jfloyd]

Let's not use track that whatever we use needs to meet the needs developers.  If python is not sufficiently stable that scripts for one version are not likely to work with a later version, that is a major issue.  We don't have the time or resources to be constantly rewriting scripts.  If the install process is cumbersome and time consuming, that is also an issue.  Also, at this point the developers know Matlab well enough to make use of it and it has well written documentation.  There are only a few people who day-to-day do FDS development and time we spend installing and learning python or some other software is time not spent developing FDS.

[Kevin]

Let me echo that sentiment. We have appealed for several years for
students and other researchers to help us with our V&V efforts. The
response has been very disappointing. While there are many reasons for
this, one of them is that Word, Excel and Powerpoint are the
overwhelming tools of choice for word processing, data analysis, and
technical drawing. My own children, ages 17 and 19, have grown up
using these tools, and when I tell them that there is another way to
write a paper or make a plot, they roll their eyes. They can't
understand why anyone would need another way to do these things. It's
reassuring to see that not everyone is trapped in this morass of
publishing mediocrity, but we can't completely abandon stable
commercial products that produce high quality documents and figures. I
agree that Matlab and Python have their pros and cons, but I think the
real problem is that 95% of our user base rely on MS products and that
has become a much greater impediment to progress compared to Matlab vs
Python.

> > in a previous blog post<http://blog.fds-smv.net/2012/12/fds-6-verification-and-validation.html>).

> > This way, whatever the methods we are using (which are all well documented

> > on the FDS-SMV Wiki Pages<https://code.google.com/p/fds-smv/wiki/Verification_Case_Setup_Example>),

> > the efforts and FDS cases can continue working as an integral and ongoing
> > part of the V&V suite as opposed to only existing in a publication or other
> > static writeup.
>
> > Kris
>

> > On Tue, Mar 5, 2013 at 2:51 AM, Daniel Gutmacher <D.Gut...@gmx.net<javascript:>

> >> Am 04.03.2013 um 18:36 schrieb Bryan Klein <kl...@thunderheadeng.com<javascript:>
> >> >:

> >> email to fds-smv+u...@googlegroups.com <javascript:>.
> >> To post to this group, send email to fds...@googlegroups.com<javascript:>
> >> .

> >> To view this discussion on the web visit
> >>https://groups.google.com/d/msg/fds-smv/-/qsrLJuzhigoJ.

> >> For more options, visithttps://groups.google.com/groups/opt_out.

>
> >>  --
> >> You received this message because you are subscribed to the Google Groups
> >> "FDS and Smokeview Discussions" group.
> >> To unsubscribe from this group and stop receiving emails from it, send an

> >> email to fds-smv+u...@googlegroups.com <javascript:>.
> >> To post to this group, send email to fds...@googlegroups.com<javascript:>
> >> .
> >> For more options, visithttps://groups.google.com/groups/opt_out.

[Tony Bova]

A possible alternative for those who don't wish to purchase Matlab is the GNU Octave project. Octave is free and open source, and its commands and syntax are, with relatively few exceptions, identical to those in Matlab. You don't get the nice development environment that comes with Matlab (I just use a text editor), but much of the functionality is the same. As an example, I use Octave to run Simo Hostikka's slice file reader function (slread.m in the Matlab scripts folder) without any alteration to the script. I haven't used it for the V&V scripts, but my guess is that they would not need much tweaking, if any, to work in Octave.

[Daniel_eval_firegas]

Hi All

I agree with Kevin: Matlab vs Python is an other discussion point. Even if python or an other tool can do the same job as matlab, the MAIN PROBLEM I here from you is to reach the attraction of students and researchers to validate. I did ones read the V&V guide but to be honest I can’t remember because as a beginner of FDS it was to complicated for me how to get all this done and the main focus was to get my first simulation done. I do wish to contribute to the community after I have finished my measurements and simulation results. And what happens at the end? Guess what, the time is missing and I do only plot FDS results versus measurement insteat of using the good quality analysis of the V&V matlab scripts I have seen in your documentation. So with all respect to your great work, you should maybe get the people by making V&V more attractive for themselves.

What you can do to get the attention of the “MS Office” users:

1)      Create an easy to use V&V tool  (fds_vv.exe  è EXE is a must)

2)      Implement your analysis tools and generate plots (automatically as option)

3)      Write a short user guide version

4)      INPUT: fds_vv.exe simulation_devc.csv

5)      The OUPUT of the V&V tool is the MAIN MOTIVATION: Plots of the just finished simulation.

6)      And the final step: Implement an option to INPUT experimental data
(example: fds_vv.exe simulation_devc.csv measurement.csv)

7)      A dream would be to have a GUI where the user can select the right experimental data.

But I have to say that I can’t really discuss with you about that because I forgot what is written in the V&V user guide. The only thing a know as a “MS user ;-)” is that my measurements have to be compared with experimental data, that’s the second V.

Yes, I know maybe to much fantasy and why can’t I do some help: It’s because as a “MS user ;-)” I do not want to wait until the tool is available and analysis and ploting of FDS results is done by myself and at the end in a lower quality than the existing V&V tools in Matlab which has good statistical analysis according to the plots I have seen.

What do you think about fds_vv.exe or much more attractive: fds_val.exe ?

Best wishes and keep the gread work on! I love it! Hopefully some fruits are coming out of my results to share with you! And I want to run your validation scripts!

Daniel

[Daniel_eval_firegas]

Hi again,

I just read the Validation guide again. Its a good track of existing validation. Also there is an example on how to quantify the model uncertainty (chapter 4, p45) with mathemical explanations.
This is the way students could calculate the predictive capaiblitly. But now there is no template or script avalaible. A scatter plot is a very simple think, but there a thousands of different versions made by students or researchers. So the best way to have is ot share the matlplot scripts (maybe they are already published, but not in the validation guide?).
So I think a short version of Validations guide with chapter 4 is necessary with code example (which is not much code for standard deviation, and scatterplotings).
The validation tracking and history shouldn't be plottet in a Guide?

Last thing, comming back to the "MS" users, or a student should exactly take more time on calculating the model uncertainty.
So you sould highlight chapter 4 (10 page) and a quick introduction with code-example, so that the sudent gets motivated to do the same thing. The results will be in a form that you could better use in the community than big excel files.

Greetings
Daniel

[dr_jfloyd]

All the scripts needed to run the cases, process the outputs, and make the plots and pictures are in the repository. There is a wiki page on how to use them.

[Daniel Gutmacher]

yes i know that. its ok this explained: http://code.google.com/p/fds-smv/wiki/FDS_Validation_Process

but to be honest 95% of the Users can follow the instructions. even for me with programming skills

it is hard to first get all the needed software and then debugging I until I can run my first validation

for the community. 95% don't use Linux. I do not have any Linux/unix installed.

a version for windows users will give you more validation feedback!!

somebody has to do the .bat files, or here an exe file out of python could help to do all that stuff.

with all respect to your work: the todo list on the wikipage is so long and expects the user know all tools.

for example: running latex requires miktex on windows: and even to install this until it get to work with all packages requires at least a day.

other example: SVN is great and I use it daily, but 95% do not know what SVN means: so you will get lease than 5% validation back from the users.

Am 06.03.2013 um 02:59 schrieb dr_jfloyd <drjf...@gmail.com>:

All the scripts needed to run the cases, process the outputs, and make the plots and pictures are in the repository.  There is a wiki page on how to use them.

--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.

To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.
To post to this group, send email to fds...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msg/fds-smv/-/_rb1P-H1MdcJ.

[Kevin]

Daniel

Most of the work in doing validation is (1) reading and understanding the experimental test report, (2) creating the FDS input files, and (3) organizing the experimental data in a usable form. When I process a new validation case, this is where I spend 90% of my time. This has nothing to do with Matlab, linux, LaTeX, or anything else other than FDS and whatever extra tools you prefer to use, like Excel. What we really need from students are a set of input files and a set of experimental data files, preferably in the form of simple comma-delimited spreadsheets (.csv files). FDS output in the _devc.csv file can then be compared column by column with data in the experimental file. Setting up the Matlab scripts and adding the case to our automated system is something we can do quickly. The student need not do it, even though we highly recommend to them that they learn these things because it will make them a much more knowledgeable researcher. But that is up to the students. If they want to limit their IT skills to Word, Excel, and PPT, that is OK with us.

This is what usually happens. A student does some FDS  validation work and we hear about it at a conference or maybe we get his/her paper to review. It is very rare that the student or professor contacts us directly. My guess is that they feel that a published paper is the most effective form of communication. This is not the case. The reason is that once the paper is published, the student is long gone, working at a new job. He/she does not have the time, nor the desire, to revisit what they just spent two long years working on. The professor has no interest in it either. The paper is published, after all. It's now on to the next project. At best, I get sent a single gigantic Excel file. In it is all the data, analysis and plots that went into the thesis. Now I either have to figure out what is what, or just give up and get the original experimental report and start over. I literally have to redo everything from scratch. All too often, I discover that there is a very important input parameter that has not been measured or reported. I discover that the student just made up a value for the missing input parameter. The lack of this data invalidates the whole exercise -- I learn that this really wasn't a validation exercise after all. At best, the student calibrated the model. The sad part of all of this is that had the student and professor contacted us early on in the project, we could have easily told them that this missing data needs to be found or measured, or otherwise the experiments are of no value for model validation. We could also have advised the student as to how to set up the input files, answered questions, gave advice, described new algorithms that might help in the effort. In short, we could have greatly added to value of this student's education.

[Daniel_eval_firegas]

Hi Kevin,

thank you for describing the case. I totaly agree with you.
I just wonder how you could get more feedback?

Keep on the great work and I will try to contribute also experimental data when I have finished all measurements.
In my special case its about fire gases and its propagation. As a gas sensing researcher I published only experimental data. The simulation with FDS is a goal i would like to do with your review in my thesis to publish quality data for the community (The thesis is in German :-)..., but I would like to write a simulation paper in english for "Fire Safety Journal" with the validation of the simulation under your review.
I measured the propagation of CO in an EU standard test room.
I hope you have free access to: http://www.sciencedirect.com/science/article/pii/S0925400511010586

What I try is to simulate the CO (and H2) propgations out of a smoldering beech wood fire. Using measurement data as boundarie condition ( CO, flux out of weightloss measurement, Temperature of the hotplate etc...)
But for details I will contact you via email this month or April.

wish you lots of feedback from other students!

bye bye
Daniel


Some other publication:

[szilagyics]

Dear All,

I have run a model with FDS6 in an underground garage. The results are interesting the CFL numbers are similars, but the time steps are larger. I had got problem with Jet fans, but it was my mistake. I didn't well understand perfectly according to User's guide description (Thanks for Dr Floyd) .  There is an other interesting  action with CTRL s. When I would like to guide a supply or an exhaust surface and jet fans with same control, the jet fans works immediately.  Is there anything change in the code?
It is working well in FDS5.

Thank you for your attention,
Csaba

[Szilagyi Csaba]

Dear All,

I have attached also a Pressure slide from the FDS6 simulation. It was
taken from a multiple mesh simulation. It's true there is a suddenly
supply and exhaust in these time, but we can see the pressure
differences at the boundary. On the velocity slides you can see the
problem within the velocity vectors. The movement of the air is
following the waving of the pressure. I also have attached a part from
out file from the same time step.
In this file in the FDS5 this problem generated a numerical error. Now
it seems to handled.

Thank you for your attention,

Csaba

2013/3/11 szilagyics <csaba.s...@gmail.com>:

> --
> You received this message because you are subscribed to the Google Groups
> "FDS and Smokeview Discussions" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to fds-smv+u...@googlegroups.com.
> To post to this group, send email to fds...@googlegroups.com.
> To view this discussion on the web visit

> https://groups.google.com/d/msg/fds-smv/-/mQXCR2wFM28J.

[Kevin]

Post a very very simple test case to the Issue Tracker. Do not use the
case shown below. Create a new, simple test case that demonstrates the
problem.

[Szilagyi Csaba]

Dear Kevin,

I have posted a simple test case in issue tracker from the first
problem from the CTRL.
At the second problem with the pressure and velocity:
I would liked to compare the result between the FDS5 and FDS6 in a real project.
I'm going to try to make more simple test case.

Csaba

2013/3/11 Kevin <mcgr...@gmail.com>:

> --
> You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to fds-smv+u...@googlegroups.com.
> To post to this group, send email to fds...@googlegroups.com.

[Chris Salter]

There appears to be an error in the documentation on page 198 where thermal radiation is being discussed. The document refers to Radiative Heat Flux Gas (from what I gather in my testing, is no longer in the software).

I'm trying to add points in an area to see the radiation a person would be subjected to in that location. In FDS 5, this was done with the RADIATIVE HEAT FLUX GAS device but this isn't valid anymore. From reading the User Guide, this appears to be now RADIATIVE HEAT FLUX (and now requires an orientation)?

The command previously was

&DEVC ID='RHF-2.2m', QUANTITY='RADIATIVE HEAT FLUX GAS', XYZ=4.0,4.1,2.2/

I've tried changing it to 

&DEVC ID='RHF-2.2m', QUANTITY='RADIATIVE HEAT FLUX', XYZ=4.0,4.1,2.2/

where I'm then told to set orientation using IOR. The issue then is that the RHF is being measured on a surface? I want this to be measured in the gas phase (for example in the middle of a corridor) but I can't see the changes to do this in FDS 6?

On Sunday, 4 November 2012 19:06:23 UTC, Kevin wrote:

It has been over two years since our last official release of version 5 of FDS and Smokeview, and we are ready to start beta testing FDS-SMV 6. Downloads are available at the home page fire.nist.gov/fds. Scroll down to the bottom of the page for the appropriate links.

Our primary interest in beta testing the new release is to determine if there is sufficient documentation to convert FDS 5 input files into FDS 6 input files. In general, the major parameters are the same, and the basic syntax of the input file is the same. However, some of the sub-models are significantly different, and there is no easy way to maintain perfect backward compatibility in FDS (Smokeview, in general, is backward compatible). By issuing error statements at start up, we want to alert you to important changes, which is why we prefer to stop the program with an error statement rather than accepting an outdated parameter or construct.

For those interested in helping with the beta testing, download the latest FDS 6 "release candidate" and try to run one of your old cases. Chances are that you will receive an error message. Chapter 1 of the new FDS User's Guide has a table that lists all of the parameters that have changed from FDS 5 to 6. We would like you to tell us how easy or difficult you find the conversion process, and whether or not the changes are well-documented. If you would like to be listed as a beta tester, send us via an email your name and affiliation, and we will compile a list in the new FDS User's Guide.

You can report your findings directly to us via email, or you can also start an Issue Tracker or Discussion Group thread. The latter two are preferable so that we can all learn about the changes in FDS and Smokeview. This will also provide us with the opportunity to describe in more detail the changes that have been made. Many of these changes are "under the hood," so to speak; that is, there have been changes to core algorithms that may not be apparent at first glance. But we can discuss these improvements as we look at your cases.

--
Posted By Kevin McGrattan to FDS-SMV Developer Blog at 11/04/2012 02:06:00 PM

[dr_jfloyd]

Based on user feedback during beta testing of fds6 that the inputs for doing RADIATIVE HEAT FLUX GAS using particles were difficult to deal with, we added back in the DEVC type. The release version of FDS 6 will have RADIATIVE HEAT FLUX GAS as a DEVC type. Details are in the latest builds of the documentation which can be downloaded from links on the google code site for fds.

[Chris Salter]

Following the guidance in the documentation, I added the line (as in the user guide)

&DEVC ID='flux', QUANTITY='RADIATIVE HEAT FLUX GAS', XYZ=0.45,0.0,0.3, ORIENTATION=-1,0,0 /

Doing so, and then trying to run the FDS file gives me an error on run, stating - 

ERROR: DEVC QUANTITY RADIATIVE HEAT FLUX GAS not found

However, noting that you say it's a shortcut to the code: 

&DEVC ID='flux', INIT_ID='f1', QUANTITY='RADIATIVE HEAT FLUX' /
&INIT ID='f1', XYZ=4.0,4.1,2.2, N_PARTICLES=1, PART_ID='rad gauge' /
&PART ID='rad gauge', STATIC=.TRUE., ORIENTATION(1:3,1)=-1,0,0, SURF_ID='target' /
&SURF ID='target', RADIUS=0.001, GEOMETRY='SPHERICAL' /

I entered that and FDS starts off and proceeds as normal. I note that you say that the RADIATIVE HEAT FLUX GAS is a shortcut to the extended code, meaning FDS 5 compatibility it maintained. It appears that this part is buggy at the minute and FDS 6 isn't seeing the shortcut.

I'll continue testing to see if I can find anything else that isn't performing as it should be.

If possible, could I be added to the beta test list? Chris Salter, Hoare Lea and Partners, UK.

Many thanks
Chris

[dr_jfloyd]

The versions available for download are beta test versions not the release version (FDS 6 has not yet been formally released). The feature was added back after beta version RC3 was released for testing. If you want to use it now you will need to download and compile the source.

[dr_jfloyd]

[Chris Salter]

Apologies, I glossed over the fact you said release version.

[Nikolai Ortiz]

Hi..

Since from yesterday was released the FDS 6.0 Official.

I wonder if at last there will be a OpenMp version release?

Thanx. 

Nikolai

El martes, 6 de noviembre de 2012 08:50:31 UTC-5, Kevin escribió:

We hope to have an optional OpenMP build of FDS 6. We haven't yet tested it.

On Tuesday, November 6, 2012 4:23:05 AM UTC-5, Daniel_eval_firegas wrote:

Dear Kevin,

First of all, thank you and the FDS-Developers Community for your effort. I believe in the performance of your work as I'm testing different simulationtools.
Hope you are all motivated from the userfeedback. I guess you will get a lot of feedback after X-Mas.
I'll test FDS6 with measurement results and will post it in the next months. Especially Gas vs Smoke is an interessting topic.


Just a question about OpenMP: Is it supported with a command in FDS6 or will there be an other compiled version like in FDS5_openmp_winxx.
Maybe its a bit early for my question.

Thank you and kind regards
Daniel

[Daniel_eval_firegas]

Hi Nikolai,

yes the OpenMP version was for FDS5 was userfriendly and hat a great performance comparing to MPICH, by using an uniform mesh size.
Try to install MPICH2 1.4.1, which worked successfully with FDS6.0.0 on my windows 7 machine with 6 cores.
You can assign multiple meshes to MPI_processes, see user guide. Try to spread the nodes equally over different mpi_processes (on a single PC with multiple cores), so you can save time like the OpenMPI version did.

I wonder how big is the impact/overhead by sharing the MPI_processes over multiple PC in the network. I believe buliding your one cluster seems to be quite complicated and i believe that synchronizing will slow down the simulation to the slowest PC in the cluster. If the boundary-data is big, the networkspeed will also slowdown the simulation? I really wonder how much time the FDS cluster-experts save ;-)

Kind regards
Daniel

PS. Just playing with the radiactive_fraction value for an ethanol fire, where i expect a lower fraction of 16-20 % inseat of default 35%, but still the experiments which were made in the past decade are lacking about this topic. I will post as soon as possible some comparable exp_data about temperature in an EN54 room + Ethanol fire (43x43cm pool).

[Nikolai Ortiz]

Hi Daniel

Ok.. 

I will put myself in the homework and see.. 

I hope share some results and guides 

and I agree about that OpenMp was more user friendly, as you said before

Regards

Nikolai