Issue 17: Version 5 of Fire Dynamics Simulator and Smokeview Released
By Kevin McGrattan, Glenn Forney, Bryan Klein, Jason Floyd, Simo Hostikka and Timo Korhonen
In October 2007, the latest version of Fire Dynamics Simulator (FDS)
and the visualization program, Smokeview, were released. FDS is a
computational fluid dynamics (CFD) model of fire, and Smokeview is a
graphics program that conveys the results of FDS in various forms of 3-D
The first official version of FDS/Smokeview, released in 2000, was
aimed at large scale simulations of smoke movement from prescribed,
well-ventilated fires. This first version was ideal for design work
where the fire's heat release rate is not predicted by the model, but
rather specified by the model user. However, during the NIST
Investigations of the World Trade Center collapse and the Station
Nightclub fire, it became evident that FDS needed improved sub-models to
make it an effective forensic tool. That is, there was now a need to
simulate actual fires, rather than just hypothetical ones.
What's new in FDS/Smokeview?
Combustion: Previous versions of FDS generally assumed that fuel and
oxygen readily burned on contact, a very good assumption for
well-ventilated fires. However, most fire scenarios involve some degree
of under-ventilation, in which case the production rate of carbon
monoxide and soot, along with other toxic by-products, can dramatically
increase. The combustion model in FDS has been improved so that it can
now model extinction and the user now has the option to model rather
than specify CO production.
Pyrolysis: The FDS solid phase algorithms have been overhauled to
provide a better description of heat transfer and pyrolysis. First,
there is an increased functionality that allows multiple layers of
materials. This was an obvious improvement to make, given the experience
with insulated steel in the WTC, and polyurethane foam covering plywood
walls in the Station Nightclub. Next, the pyrolysis algorithm has been
generalized to allow for multiple materials undergoing multiple
reactions. The intent here is to retain the simplicity of past versions,
but open up the possibility of increased complexity in describing how
real materials burn.
Material Property Input: Versions 1 through 4 of FDS included a
"database" of properties for common building materials, furnishings and
other potential "fuels." These properties were a collection of
bench-scale measurements, handbook values and "best estimates" that were
often misunderstood and misapplied by the model users. Originally, the
database was to serve as a set of examples, but given the scarcity of
data for real materials, it became a de facto standard. As a
result, starting with FDS version 5, the database has gone away. It is
hoped that this will spur a renewed effort by the fire research
community to develop new, or re-interpret existing, robust, practical
methods of obtaining material properties, with an emphasis on
fire-specific phenomena like pyrolysis and char formation.
Evacuation: FDS 5 contains an evacuation model developed at VTT,
Finland, that allows the simulation of the fire and evacuation processes
simultaneously. The obvious benefits of implementing an evacuation
model directly into FDS are the straightforward computation of how the
fire affects people, possibility for two-way coupling (people open doors
and change the ventilation) and the time savings when the simulation
geometry is defined for only one piece of software. In addition to the
actual movement of people, the model tries to account for some of the
evacuee decision making processes as well. The evacuation model is still
in its early stages but has already found applications in practical
fire safety engineering. More information can be found at www.vtt.fi/fdsevac.
Visualization: Smokeview remains the key to understanding the fairly
complex FDS calculations that can consume days or weeks of computing
time on an entire network of computers. Just processing the tremendous
amount of data is a challenge in itself, and the rendering of clear,
easy to understand animations means keeping up with the monthly advances
in graphics hardware and software. Also, Smokeview has been designed to
render the output of the NIST zone model CFAST.
Verification and Validation: In the past several years, organizations such as SFPE1 and the US Nuclear Regulatory Commission2
have conducted Verification and Validation (V&V) studies of
different fire models. In short, V&V is a process to assess the
accuracy of the mathematical solvers and the underlying physical
assumptions of the models. For FDS, this was done informally, via
journal articles, conference proceedings, and student theses, both at
NIST and elsewhere. However, there was no systematic approach to the
process. Now, there are "suites" of case studies and experimental
datasets against which FDS will be checked periodically to ensure that
the model retains the accuracy of previously reported studies. While
this can be a somewhat tedious task, it is essential for maintaining
Support: FDS and Smokeview originated as "research" codes, and to
this day can be a bit daunting to anyone new to fire modeling. However,
there is a new emphasis on user support. By exploiting several free
on-line services provided by Google, FDS/Smokeview now has an easy to
use Discussion Group where users can post questions and comments, a
simple bug-reporting service where cases can be assigned to one of the
developers for rapid resolution, and a simple point and click download
facility. In addition to improved support for users, FDS is now being
developed and maintained using fairly common software development tools.
The source code and manuals are stored in an external repository and
can be accessed by all members of the team and anyone else who is
interested in working with the actual source code. This idea of "version
control" is not unusual to a commercial software development team, but
it is relatively new for researchers whose primary education is not in
computer science or information technology.
Finally, the most notable change in FDS/Smokeview 5 is the
realization that the development and maintenance of sophisticated fire
modeling software will require a collective effort by the fire research
community. Although NIST remains the primary caretaker, professional
societies, testing laboratories, universities, and consulting companies
are playing an increasing role. Many of the changes made to the software
and to its development infrastructure have been designed to facilitate
its long-term maintenance by the various stakeholders. No single person
or organization can do it alone.
Kevin McGrattan, Glenn Forney and Bryan Klein are with the National
Institute of Standards and Technology. Jason Floyd is with Hughes
Associates, and Simo Hostikka and Timo Korhonen are with VTT, Finland.
Evaluation of the Computer Fire Model DETACT-QS, Society of Fire Protection Engineers, Bethesda, MD, 2002.
NUREG-1824 and EPRI 1011999, "Verification and Validation of Selected
Fire Models for Nuclear Power Plant Applications," Vols. 1-7, U.S.
Nuclear Regulatory Commission, Washington, DC and Electric Power
Research Institute, Palo Alto, CA, 2007.
For questions concerning delivery of this e-Newsletter, please contact our Customer Service Department at (216) 931-9934 or magazine.sfpe.org.