Wednesday, 13 November 2013

Reply to 'FDS and the Challenge of Big Data'

While on the Tube's District Line from the office, I read the most recent blog article written by the developers of FDS. It is titled "FDS and the Challenge of Big Data".

For those of you who do not know it, FDS stands for Fire Dynamics Simulator, and it is the state of the art in fire modelling. It is a fine, advanced and excellent code of Computational Fluid Dynamics  (CFD), especially developed to simulate the behavior of flames and smoke in buildings and large open spaces. Its source code, in FORTRAN, is open and freely avaible to all. The work of development and maintenance is mostly carried by staff at NIST (VTT at Finland also plays a major role). NIST stand for National Institute of Standards and Technology, and it is a USA government agency which mission is to "promote U.S. innovation and industrial competitiveness by advancing science and technology".

Their blog article is mostly a complain. It is formulated around the apparent lack of good collaborations from academia around the world to support their difficult task of developing and maintaining FDS. They think the reason for this is down to the "publish or perish" stereotype, and use a blog article from a Cosmology researcher who expresses similar frustrations with academia.

I have four points to make regarding the FDS blog article:

0) Thank you. I felt bad that you think your contributions go thankless.This is unfair, because FDS is the state of the art, it is provided without cost and openly for all around the world to use. It has tremendously helped the Fire Safety Engineering community to develop further.

1) Test You Hypotheses. You should make sure you know and understand your potential collaborators, specially before you criticize them in the open. For example, the "primary currency of the academic reward structure" in engineering departments is funding and industrial relevance, not published papers. Be aware of using a Cosmology case to run your arguments against academic users of FDS who are mostly from engineering departments. Also, note that there are more jobs for research in Cosmology than for research in Fire Safety, so I find your final kick "there are only so many jobs available in cosmology" ill suited to the critique.

2) More Carrots. Find some of the true rewards that match the motivation of your potential collaborators. For example, I suggest you create the yearly NIST Award for Outstanding Contributions to FDS. This would create recognition and esteem which are highly valued in academic CVs, more than a bunch of papers, and in some cases it is the key for obtaining a position or promoting.

3) Elephant in the Room. Lets we forget that academia has played an essential role in the success of FDS, maybe also a thankless task. I have said in the past [*] that I believe the industrial success of FDS lies on three pillars; it is free, it is excellent for research, and there are hundreds of papers showing good modelling results (just google it). The third point is because no other fire code has ever received so much publication attention. The difference with other fire codes being of two orders of magnitude. This very high number of journal papers and the multiple open discussions taking place in any fire conference every single year have promoted somehow the image that FDS is 'the validated', 'the accepted' or 'the standard' model for a wide range of industrial designs. And industry uses FDS extensively, in thousands of fire protection projects worldwide in any one year, with the approval of the corresponding authorities. Hence, FDS contributes to fulfill the mission of NIST, and promotes innovation and industrial competitiveness by advancing science and technology, in the USA and also in the rest of the world. This industrial projection is thanks to all the research users of FDS who decided to publish and go public at some point. Thanks go to them too.

[*] Advantages and Disadvantages of Fire Modelling, Irish Chief Fire Officers Association Annual Conference, Dundalk, May 2012.

No comments:

Post a Comment