Fire Banner

Friday, July 30, 2010

Research and the release rate of FDS versions

I have recently posted in the Discussion Group site for FDS and Smokeview some questions related to the release rate of FDS versions that I would like to post here as well.

Note: Fire Dynamics Simulator (FDS) is a computational fluid dynamics model of fire-driven fluid flow developed by NIST.

First, note that I am not a code developer but an user. I am a particular type of user: a researcher. An academic who studies methodologies to use some of the tools available to fire safety engineers. Hopefully I have helped or will help to develop the state of the art in fire modelling. I certainly support improvements to FDS and greatly admire the work of the FDS developers team.

For the last two years I am facing two problems stemming from the different versions of FDS that are released with some frequency. Research (my research at least) has a characteristic time that seems to be significantly longer that the time between version releases. This means that when we start a research project on modelling that includes FDS we use the newest version available but by the time the work is finishing and concluding, there are one or two new versions of FDS available. Note: PhD thesis in the UK last for three or four years.

This has led to these issues in my research group:

I - Different results from different versions
We have accidentally observed two times already (in two different PhD thesis work) that the same input file produces significantly different results in different but consecutive versions of FDS (eg v5.2.0, v5.2.5 and v5.3.0). In one case it took us months to figure out the problem. Both of these problems were discussed in due time in the FDS forum. My personal recommendation to my students is that they fix the FDS version of their interest at the beginning of the project and stay with it until the very end of the thesis.

I wonder if other researchers would like to express their views on this issue?

II - Peer reviewed of FDS results
We have had reviewers complaining and requesting rejection of our modelling papers because results are not in the latest version of FDS. NOTE: just the review process in fire and combustion typically takes anything from 3 to 16 months.

About modelling work on the Dalmarnock Fire Tests (Chp 2 of this thesis), one anonymous reviewer recently said "Given that [version] 5 has significant changes in the combustion model and other submodels, there is no value to the community in publishing a paper on FDS 4 unless the paper can demonstrate that the conclusions on FDS performance will remain valid for the current version and versions under active development".

To demonstrate that the conclusions of the work apply to the newest version of FDS effectively implies repeating the thesis work. This is not affordable in most cases, at least not for my group. Should academics fear every new version of FDS because it means one extra year of work for PhD students?

I wonder if other researchers would like to express their views on this issue?

III - Corollary
Following from issues I and II, I am confronted with the question, what happens with buildings which design was aided and approved using previous versions of FDS. What are the implications to forensic investigations that reached conclusions in court using a previous version of FDS?

1 comment:

Guillermo Rein said...

The Chp 2 in the PhD Thesis of my student Wolfram Jahn that I refer to in the blog post was published in 2011in Building and Environment.
Full reference is:

- W Jahn, G Rein, JL Torero, A Posteriori Modelling of the Growth Phase of Dalmarnock Fire Test One, Building and Environment 46 (5), pp. 1065-1073, 2011.